Professional Documents
Culture Documents
Version history
Issue
Date
Author
16/03/12
Finalised report
11/05/12
Acknowledgements
Section 20 Technical Advisory Group
Andrea Johnstonova (Chair)
SEPA
Drew Aitken
SEPA
Heather Forbes
SEPA
Julia Garritt
Lorna Harris
SEPA
Kirsty Jack
SEPA
Richard Jefferies
SEPA
Roy Richardson
SEPA
David Scott
SEPA
Nadeem Shah
Forest Research
Mark Williams
SEPA
Dominic Habron
Mark McLaughlin
Executive summary
The Flood Risk Management (Scotland) Act 20091 (the FRM Act) introduces a new
sustainable approach to managing flood risk and places new duties on SEPA and
Responsible Authorities. One responsibility placed on SEPA is to assess and
consider the role which Natural Flood Management can play in reducing flood risk.
Natural Flood Management was defined by SAIFF (2011)2 as:
Natural Flood Management can be defined as those techniques that aim to work
with natural hydrological and morphological processes, features and characteristics
to manage the sources and pathways of flood waters. These techniques include the
restoration, enhancement and alteration of natural features and characteristics, but
exclude traditional flood defence engineering that works against or disrupts these
natural processes.
The aim of this study is to advance development of the methodology for assessing
the contribution that Natural Flood Management can make to managing flood risk.
This report builds on previous work undertaken by SEPA, Jacobs and others in the
field of Natural Flood Management. The main tasks have included:
Review the previous work by Jacobs and the proposed GIS method for
assessing Natural Flood Management;
As part of the delivery of Section 20 of the FRM Act SEPA appointed Jacobs to
develop a methodology to assess Natural Flood Management measures (Jacobs,
2011)3. The Jacobs method was founded on a comprehensive and up-to-date
literature review (Werritty et al., 2012)4. Unfortunately this methodology is associated
with licensing issues, technical difficulties and a catchment based approach which
did not allow national identification of opportunities led to this tool not being adopted
by SEPA.
In this study SEPA has sought to build on the Jacobs literature review by extending it
to include all flood mechanisms (pluvial, fluvial, groundwater, coastal and urban) and
by expanding it to include a wider literature base. The extended literature review has
formed the foundation for the development of more robust methods for screening and
assessing Natural Flood Management measures.
Section 4 of the report proposes methods for identifying and appraising Natural Flood
Management opportunities. The methods will provide an overview of the effect of
Natural Flood Management measures that will inform the appraisal process. It is
crucial that these predictions are kept in context. At no point should these estimates
be taken as the correct answer, rather they should be seen as an approximate guide
to what potentially might happen within the bounds of the reported uncertainty.
-i-
2012
SEPA
Section 28 Appraisals
Using output of Screening in combination with
catchment characteristics and further
information to develop a long list of measures
and refinement to short list. More detailed
appraisal producing a list of preffered
measures.
2013 - 2014
SEPA / Responsible
Authorities
2015 16
The relevant
Responsible Authority
It is proposed that the identification and appraisal of Natural Flood Management will
include three phases of work:
1.
Identification phase: A national screening process to identify opportunities for
Natural Flood Management measures. This screening process will identify areas
within catchments with Potentially Vulnerable Areas that have natural flood
management opportunities but will make no consideration of constraints or other
benefits. The screening process will not directly recommend which specific measure
should be implemented where, nor will the screening facilitate the quantification of
the flood risk management benefits of undertaking a specific natural flood
management activity. However the process will facilitate the identification of areas
that are worth further investigations at a later stage. The main output of the
screening process will be six maps showing:
-ii-
It is proposed that these maps, will meet the requirements of Section 20 of the FRM
Act and they will be made publically available along with appropriate supporting
descriptions and guidance.
2.
Appraisal phase: The second phase will seek to bring together the outputs
from the screening process with further information about catchment characteristics
in order to develop better understanding of what may be achievable where within the
catchment. Using the findings of the screening process in combination with other
catchment information the measures will be presented as options to manage the
sources and pathways of floodwaters, as summarised in the Natural Flood
Management Summary Table (Appendix B). This information together will be used to
appraise natural flood management measures alongside other flood risk
management measures. Over the course of this stage the long list will be reduced to
a short list and the short list appraised to identify a basket of the most sustainable
measures. The process will give due consideration to environmental, economic and
social costs and benefits. The end result of this phase is a short list of options that
will be agreed by all responsible authorities.
3.
Further development through local FRMPs: The third phase will include more
detailed assessment of the agreed, measures using existing and new assessment
approaches. This more detailed assessment will be carried out by the relevant
responsible authority as part of a local flood risk management plan.
The literature review has identified that some measures can be assessed using
existing tools and methods. It is therefore proposed that where possible, this further
assessment is carried out using existing assessment tools. However, it is also
proposed that a new hydrological assessment method is developed to facilitate the
quantification of fluvial and pluvial Natural Flood Management measures. This study
has investigated how this assessment could be conducted and the report presents a
proposed method for how this could be accomplished.
The proposed hydrological assessment methodology is a single event spatially
distributed model which builds on the uniformly distributed PDM (Probability
Distributed Model) runoff generation model. In the absence of viable alternatives
many elements of the PDM configuration are based on the ReFH (Revitalised Flood
Hydrograph). The effect of land use change is included within the PDM runoff
generation model through the inclusion of a canopy interception moisture store and
variation of the antecedent moisture condition based on the vegetations moisture
demand. There is also provision to assess the effect of soil degradation on the soil
moisture capacity as a consequence of intensive land management practices.
-iii-
Flow routing within the hydrological model is based on a time to outlet grid which is
generated for a representative bankfull flood event. This grid summarises the travel
time for runoff to travel from any location to the catchment outlet. It facilitates the
generation of a hydrograph at the catchment outlet by the summation of the runoff
generated for the array of PDMs covering the catchment. This approach allows the
effect of any number of diffuse land management changes to be accumulated over a
wide range of catchment sizes. The 1D kinematic wave hydraulic routine within the
flow routing engine permits the quantification of land cover roughness changes prior
to flood water reaching the watercourses in addition to in channel modifications such
as the increase in riparian vegetation or restoration of a watercourses planform.
Although not included within the core of the proposed flow routing method, there is
sufficiently flexible to allow the time to outlet grid to be calculated using an alternative
method where appropriate. This flexibility permits the use of more detailed 1D-2D
hydraulic models to test the time lag effects of floodplain interventions such as the
planting of trees on floodplains to increase roughness.
There is currently no software available to assist in the application of the fluvial
hydrological assessment methodology proposed by this study. Testing undertaken
as part of this study has identified that it would be feasible to construct it as a
Toolbox in ArcGIS or similar. If or when a tool is developed it will be necessary to
undertake a comprehensive calibration and verification process using a large
selection of catchments across Britain. It is essential that the calibration and
verification process considers catchments with a wide range of land use, soils,
rainfall, potential evaporation, size, topography and network shapes. The calibration
process should not solely focus on recreating the reported effects at a small number
of Natural Flood Management test catchments which are likely to only represent the
conditions found within a small proportion of catchments which the methodology
might ultimately be applied to.
-iv-
Table of Contents
Executive summary.................................................................................................. i
1
Introduction ..................................................................................................... 1
Introduction ................................................................................................ 5
2.2
2.2.1
2.2.2
2.3
Natural Flood Management Implementation Learning from Practice
Workshop Report (SNIFFER & SEPA, 2011) ........................................................ 7
2.4
Additional comments.................................................................................. 7
2.5
Summary ................................................................................................. 12
Introduction .............................................................................................. 13
3.2
3.3
3.3.2
Summary.......................................................................................... 72
3.4
3.5
4
Proposed methodology for assessing and considering the contribution of
Natural Flood Management .................................................................................. 87
4.1
Introduction .............................................................................................. 87
4.2.1
Introduction ...................................................................................... 92
4.2.2
4.2.3
Summary........................................................................................ 108
4.3
Quantifying the flood risk management benefits of Natural Flood
Management measures ..................................................................................... 110
4.3.1
4.3.2
4.3.3
4.3.4
4.3.5
4.3.6
4.3.7
4.3.8
Summary........................................................................................ 143
-v-
-vi-
Table of Figures
Figure 1: Outline process for the delivery of Natural Flood Management requirements
under the FRM Act ..................................................................................................... ii
Figure 2: Natural Flood Management techniques (SAIFF, 2011) .............................. 2
Figure 3: Example of the differences between the catchment delineations used within
the Jacobs S20 tool ................................................................................................... 8
Figure 4: Example of a riparian intervention that generated a 0.64% reduction in the 1
in 200yr flood event without any of the proposed riparian woodland being positioned
in the riparian zone. (Proposed riparian woodland shown in pink). ......................... 8
Figure 5: Example of the output from the proposed Jacobs S20 assessment tool
showing the hydrological benefits of restoring the floodplain area (purple) to facilitate
an increase in flood depths of 300mm with a 50% efficiency factor. ........................ 10
Figure 6: Derivation of the risk of compaction classes (Holman et al, 2001) (GLU
Grazing Livestock Unit) ........................................................................................... 33
Figure 7: Mass infiltration rates (Holtan and Kirkpatrick, 1950) ................................ 34
Figure 8: Monitored soil moisture at two adjacent sites in the Allt AMharcaidh
catchment in the Cairngorms: a) mature natural Scots Pine woodland, b) wet heath
moorland. (Dots = averaged field measurement, Unbroken line = HYLUC model
simulation)............................................................................................................... 42
Figure 9: Simulated average seasonal soil moisture deficit for upland grass, heather
and coniferous woodland for the River Calder catchment, Renfrewshire. (Period
simulated: 1983 1993) .......................................................................................... 43
Figure 10: Predicted attenuation of: a) winter, b) summer flood peaks due to a
hypothetical 100% conversion of the Calder catchment (Renfrewshire) from upland
grass to various types of woodland\forest or lowland grass (Price et al, 2000). Given
as a percentage of the peak flows predicted for upland grass. (Prediction uncertainty
band reflects the range of values attributed to the model parameters coming from
different field experiments). ..................................................................................... 43
Figure 11: Modelled potential reduction of flood peaks on the Kamp following
afforestation (increase in pine forest from 47% to 86%), plus the predicted increase
of flood peaks following deforestation of the existing forest (from 47% to 0% forest
cover). [Source: Frances et al (2008)] ..................................................................... 45
Figure 12: Modelled potential reduction of flood peaks on the Iller following a 29%
increase in pine forest cover. [Source: Frances et al (2008)] Note the 25 50 year
event is estimated to be approximately 800 cumecs................................................ 46
Figure 13: The changes to the speed of catchment response parameter (TP) and the
percentage of storm event rainfall forming quick runoff (PR) observed at the Coalburn
experimental peat catchment following intensive upland drainage. (Values calculated
from data presented in Robinson et al, 1998. Speculated canopy cover estimated by
Jacobs based on experience and the limited information provided by Robinson et al.)
................................................................................................................................ 52
Figure 14: Unit costs of implemented realignment plotted against size and year.
Scott et al. (2011). ................................................................................................... 71
Figure 15: Sample output, River Ouse combined sensitivity (Environment Agency,
2008) ....................................................................................................................... 75
Figure 16: Sample output, National assessment combined sensitivity (Environment
Agency, 2008) ......................................................................................................... 75
-vii-
Figure 17: User interface of the Land Management CFMP Tool (Environment
Agency, 2008) ......................................................................................................... 76
Figure 18: High priority areas with the greatest potential for woodland planting to
reduce downstream flooding (Broadmeadows & Nisbet, 2009) ............................... 78
Figure 19: Sample output from the POLYSCAPE tool for the Pontbren catchment
(Pagnell, 2009). ....................................................................................................... 81
Figure 20: Example of current UK farms and likely impacts of land use management
scenarios and how they can be mapped onto the decision support matrix (Hewett et
al., 2006) ................................................................................................................. 82
Figure 21: Four hillslope runoff risk scenarios for the same common land unit
(Hewett et al, 2006). ................................................................................................ 82
Figure 22: Example of a time map generated by OVERFLOW for the Pickering Beck
catchment, in this instance a 14mm per day gross rainfall even, assuming a
percentage runoff of ~55% after adjusting for evaporation and groundwater losses.
The timescale is in hours to the catchment outlet. ................................................... 84
Figure 23: Outline process for the delivery of Natural Flood Management
requirements under the FRM Act............................................................................. 88
Figure 24: Flow chart showing the inclusion of Natural Flood Management within the
FRM Act .................................................................................................................. 89
Figure 25: Variation in percentage runoff with a changing leaf area index using
typical Scottish catchment descriptors ..................................................................... 95
Figure 26: Variation in PROPWET across northern Britain (Bayliss, 1999)............. 96
Figure 27: Variation of PROPWET with SAAR (Data for HiFlows catchments) ....... 97
Figure 28: Potential means of displaying sub-catchment desynchronisation
information .............................................................................................................101
Figure 29: Example output from ST:REAM for the Taff catchment in Wales (Parker, in
press) .....................................................................................................................103
Figure 30: Sample output for the Surge Attenuation Potential ...............................104
Figure 31: Summary of the European Environment Agency wave climate data . Error!
Bookmark not defined.
Figure 32: Pictorial overview of the proposed Natural Flood Management
assessment model .................................................................................................114
Figure 33: Schematic representation of the proposed Natural Flood Management
assessment model .................................................................................................115
Figure 34: Design rainfall profiles for summer and winter, drawn as normalised
hyetographs (Kjeldsen, 2007) ................................................................................116
Figure 35: A comparison of the variation in canopy interception capacity with leaf
area index (LAI) using MORECS (Hough and Jones, 1997) and Hoyningen-Huene
(1981) ....................................................................................................................118
Figure 36: Equal water content ( C ) across stores of different capacity (Kjeldsen,
2007) ......................................................................................................................119
Figure 37: Plot comparing the regression based estimate of PROPWET with the
FEH CD PROPWET ...............................................................................................124
Figure 38: Plot comparing the regression based estimate of PROPWET with the
FEH CD PROPWET ...............................................................................................125
-viii-
Figure 39: An example of a time to outlet grid generated for a sample Scottish
catchment as part of this study ...............................................................................128
Figure 40: Roughness values recommended by USDA (2010) for shallow surface
runoff for flow depths of less than 30mm. ...............................................................129
Figure 41: Example of how the time lag effects of reservoirs, lochs and floodplains
could be included within the assessment method ...................................................132
Figure 42: Summary of the proposed flow routing methodology ............................133
Figure 43: Example of a five band time to outlet grid for a sample Scottish catchment
...............................................................................................................................135
Figure 44: Example hydrograph using five runoff bands for a generalised catchment
...............................................................................................................................135
-ix-
Table of Tables
Table 1: Priority research questions ...................................................................... 20
Table 2: Questions requiring further research......................................................... 20
Table 3: Summary of catchment scale estimates (from monitoring or modelling) of the
effect of afforestation on peak flood flows (Jacobs, 2011). ...................................... 25
Table 4: Summary of riparian woodland studies (Jacobs, 2011) ............................. 26
Table 5: Summary of floodplain woodland studies (Jacobs, 2011) ......................... 26
Table 6: Summary of annual transpiration and interception data for a range of land
cover types (Nisbet, 2005)....................................................................................... 37
Table 7: Wave run-up friction factors (CIRIA, 2007) ..............................................106
Table 8: Summary of the preferred Natural Flood Management screening processes
and the data requirements......................................................................................108
Table 9: Summary of the screening processes which are not recommended without
further scientific support or data .............................................................................109
Table 10: Summary of assessment methods for Natural Flood Management
measures and the identification of areas where there are currently gaps in
assessment capability ............................................................................................112
Table 11: Leaf Area Index values as used in MORECS 2.0 (Hough and Jones, 1997)
...............................................................................................................................117
Table 12: Effect of soil degradation via the use of analogue HOST soil types
(Packman et al, 2004) ............................................................................................121
Table 13: The range and quartiles for variables used in the PROPWET descriptor
regression analysis ................................................................................................123
Table 14: Summary of flow routing used within existing hydrological models and
methods .................................................................................................................127
Table 15: Examples of land cover types and the roughness bands which could be
employed ...............................................................................................................130
Table 17: Example of memory requirements for a range of catchment sizes .........136
Table 18: Summary of the data required for the proposed methodology................137
Table 19: Summary of the data which could improve the quality of the tool ............138
Table 20: Summary of how a range of Natural Flood Management measures could
be assessed using the proposed methodology .......................................................139
-x-
Appendices
Appendix A - Review of Jacobs Section 20 Review (CREW, 2012)
Appendix B - Summary Table of Natural Flood Management Measures
Appendix C - Other Natural Flood Management considerations
Appendix D - Requirements on SEPA in relation to Natural Flood Management under
the FRM Act
Appendix E Runoff generation model examples and initial calibration of flow routing
model
-xi-
Cini
C max
D
DDF
DPLBAR
DPSBAR
DTM
FARL
FEH
FSR
G2G
GLU
HOST
- Duration (hours)
- Depth duration frequency model
- Mean drainage path length (km)
- Mean drainage path slope (m/km)
- Digital terrain model
- Flood attenuation from reservoirs and lakes, FEH descriptor
- Flood Estimation Handbook
- Flood Studies Report
- Grid to Grid Hydrological Model
- Grazing livestock unit
- Hydrology of Soil Types (soil classification)
LAI summer / w int er - Leaf area index, half the leaf area per unit area of ground (sqm/sqm)
LCM
- Land cover map (1990, 2000 or 2007)
Lumped model - A model that treats the whole of a catchment as a single accounting
unit and predicts only values of variables averaged over the
catchment area
MORECS
- Met Office Rainfall and Evaporation Calculation System
PDM
- Probability Distributed Model
PE
- Potential evapotranspiration (reference land cover) (mm/yr)
PR
- Percentage runoff (%)
PROPWET - FEH catchment descriptor, proportion of time when the estimated
SMD was less than or equal to 6mm (1961-1990)
PVA
- Potentially Vulnerable Area
q
- Direct runoff (cumecs)
Q
- Flow (cumecs)
QMed
- Median annual maximum flood
rsc
ReFH
- Revitalised Flood Hydrograph
SAAR
- Standard average annual rainfall (mm/yr) (1961-1990)
SAIFF
- Scottish Advisory and Implementation Forum for Flooding
Semi-physical model - a hybrid model which represents some hydrological
processes in a physical manner and others via mathematical functions
that have no direct physical basis.
SEPA
- Scottish Environment Protection Agency
SM
- Soil moisture depth (mm)
SMD
- Soil moisture deficit (mm)
SNH
- Scottish Natural Heritage
SPR
- Standard percentage runoff (%)
-xii-
SuDS
T
Tp
URBEXT
WFD
-xiii-
-xiv-
1 Introduction
The Flood Risk Management (Scotland) Act 20091 (the 'FRM Act') was enacted in June
2009. The FRM Act places new responsibilities on the Scottish Environment Protection
Agency (SEPA) and other Responsible Authorities in relation to the sustainable
management of flood risk. Part of the sustainable approach is the consideration of the
contribution that Natural Flood Management can make to manage flood risk. This report
concerns the delivery of Section 20 of the FRM Act which stipulates that SEPA must
undertake an assessment of the role that the restoration, enhancement and alteration of
natural features and characteristics can have in managing flood risk in Scotland and
subsequent use of this information through the appraisal process and in local flood risk
management plans. This restoration, enhancement or alteration of natural features and
characteristics is commonly referred to as Natural Flood Management (NFM).
What is Natural Flood Management?
In reflection of the broad-spectrum of measures that have been termed Natural Flood
Management there has been some ambiguity over what it entails despite there being
apparent consensus on what it was trying to achieve. The Scottish Advisory and
Implementation Forum for Flooding (SAIFF, 2011)2 provides the following definition of
Natural Flood Management which has been used by this study:
Natural Flood Management can be defined as those techniques that aim to work with
natural hydrological and morphological processes, features and characteristics to
manage the sources and pathways of flood waters. These techniques include the
restoration, enhancement and alteration of natural features and characteristics, but
exclude traditional flood defence engineering that works against or disrupts these natural
processes.
It should be recognised that alteration has been included within the definition so that
techniques associated with changes in land management and partial restoration are not
excluded from Natural Flood Management. In some instances it will be difficult to draw a
line between what is considered Natural Flood Management and traditional flood
defences. Figure 2 provides additional insight to SAIFF interpretation of what is included
within Natural Flood Management.
-1-
-2-
The SFM guidance published by the Scottish Government encourages that Section 20
assessment also creates information that can be used in the appraisals of FRM options.
This means developing information about the potential benefits and adverse impacts of
Natural Flood Management measures that can feed into the appraisal process.
Introduction to this report
Section 2 of this document provides an overview of the review process for a recent report
commissioned by SEPA on Natural Flood Management (Jacobs, 2011)3. As part of the
quality management process the report was peer reviewed by leading Scottish
academics in the field of flood risk management via the Centre of Expertise for Waters
(CREW) (Werritty et al., 2012)4. This peer review process identified that the Jacobs
literature review was exhaustive, authoritative and up-to-date, leading to only minor
suggestions of additional sources with no notable changes to the outcome of the studys
findings. A proposed methodology for assessing Natural Flood Management measures
was also presented within the report (Jacobs, 2011)3. Unfortunately this methodology is
associated with licensing issues, technical difficulties and a catchment based approach
which did not lend itself to national identification of opportunities led to this tool not being
adopted by SEPA.
The literature base presented by Jacobs has been revisited in Section 3 of this report with
efforts made to expand the depth of the review by bringing in additional literature
recommended by the peer review process. The range of flood mechanisms was also
expanded in an attempt to give due consideration to the role Natural Flood Management
can play in all modes of flood risk, including:
Pluvial;
Fluvial;
Groundwater;
Coastal: and
Urban drainage.
The updated literature review has been divided into three sub-sections so that a specific
focus can be directed at each area in turn:
Having updated the literature review a summary table detailing the key information for a
broad range of Natural Flood Management measures (29 in total) has been prepared.
This table is presented in Appendix B of this report. The table summarises the most
relevant scientific literature for each measure, how the flood risk reduction can be
quantified and the proposed methodology for inclusion within the methodology for the
delivery Section 20 of the Act.
Section 4 details the proposed methodology for the delivery of Section 20 of the Act. This
section details how the delivery will be divided into three phases; The identification (or
screening) phase using national GIS screening; Appraisal phase to assess NFM
measures against and in combination with other measures to manage flood risk, and
finally further development of appraised measures through Local Flood Risk Management
Plans.
-3-
The Screening, process, will be undertaken at a national level to identify opportunities for
Natural Flood Management. The output will be the production of six national maps
detailing the potential within the landscape to use Natural Flood Management measures
to manage flood risk. Each map will detail the opportunities for a specific type of
measure, such that localised opportunities for the source control of runoff or the
attenuation of estuarine surge can be rapidly identified following consideration of further
catchment information as part of the Appraisal process. The scoring system used within
the screening maps will not include non-flood risk constraints or opportunities, nor will the
screening process directly specify which measure would be most suited to address
flooding at a particular location. It is intended that the output of the screening process will
assist in generating ideas at the earliest stages of measures development ahead of more
focused assessment during the Appraisal.
The last stage in the process - quantification of the effects of agreed measures - will be
taken forward as part of local flood risk management plans. The literature in Section 3
identifies that there are currently gaps in the methodologies that can be used to quantify
the effect of Natural Flood Management measures, in particular for fluvial and pluvial
driven flooding. Where there are existing tools and methodologies which can be used to
for assessment then these should be used, however in the case of fluvial and pluvial the
report recommends that a new hydrological methodology is developed. Section 4.4
presents a specification for a proposed semi-physical rainfall-runoff modelling
methodology which could be used to assess fluvial and pluvial measures. It should be
noted that the method is currently only a proposal, and its successful implementation
would require a comprehensive calibration and verification phase prior to it becoming a
valuable hydrological assessment tool.
-4-
Werritty A., Arthur S., Ball T. and Wallerstein N., Review of Jacobs Section 20
Review, CREW-The James Hutton Institute, January 20124.
Informal comments have also been provided by Forest Research via email to the author
of this document, these have been incorporated within the authors comments in Section
2.4.
-5-
Q8) Do you agree with the limitations specified in section 3.7 [of the Jacobs Report] and
recommendations in section 8.8[of the Jacobs Report]?
Q9) Do you have any other observations from Section 3 [of the Jacobs Report] that you
would like to highlight?
Q10) Overall, would you recommend the use of the proposed GIS tool and the rule for
use as the screening tool to provide an indication of the effect of Natural Flood
Management techniques?
Q11) Do you have any recommendations for future work that would be beneficial to
further develop the rule set and the modelling tool?
To avoid duplication of the CREW review it has been appended to this document in
Appendix A.
-6-
Concerns were raised over the validity of the upper/middle/lower division hard
coded within the tool. It was suggested that a more subtle subdivision of
catchments is needed in catchment headwater areas.
It was suggested that the uncertainty may be so large that decision making will
still remain based on local expert judgement.
Many of the encoded Natural Flood Management measures are very generalised
and will need to be improved in the future.
The S20 tool is fast and relatively stable (within the normal expectations ESRIs
ArcGIS platform). It has an intuitive easy to use interface.
The prescriptive structure of the model does retain a good level of control over the
models outputs, this reduces the risk of an inexperienced user applying the tool
incorrectly.
The method is reliant on CEH Flow grid coverage for the entire catchment and the
tool is hard coded on the assumption that the CEH Flow grid is correct. In
essence the tool calculates the change in flows and not the actual flows, therefore
the predictions made by the tool are limited by the underlying CEH Flow grid on
which it has been built. The tool should not be seen as a hydrological model,
rather a method for scaling existing hydrological predictions.
It is unclear what catchment delineation the tool functions with. The sample data
provided by Jacobs for the Allan Water shows two catchment delineations for the
Allan Water which disagree with each other. It is understood that parts of the tool
work with the assumption that the CEH Flow Grid is correct while other elements
work with a user supplied catchment delineation. The ambiguity regarding what
catchment area is being considered is not helpful. An example of the Forglen
-7-
Burn (a tributary of the Allan Water), which drains the loch at the University of
Stirling, is shown in Figure 3.
Figure 3: Example of the differences between the catchment delineations used within the
Jacobs S20 tool
The current configuration of the tool does not permit predefined shapefiles to be
uploaded into the tool nor can manually drawn Natural Flood Management
-8-
outlines be exported from the tool. It is likely that it would be feasible to include
improved import/export functionality in an updated version of the tool.
It is up to the user to classify the function of placed woodland. For example if the
woodland tool is used across a broad area which incorporates floodplain and
riparian areas the tool will only assess the woodland for its runoff reducing
capabilities and not for the riparian or floodplain benefits.
The woodland cover within the LCM2000 dataset is hard coded as the actual
woodland cover. While it is accepted that it is a good representation of woodland
cover at the time the data was acquired, in many instances the LCM2000 data
does not correctly classify the present day local land cover correctly.
The tool does not allow a way for the results to be exported, for example the
revised hydrograph cannot be extracted to allow the flows to be applied to a
hydraulic model.
Currently the tool only allows the assessment of the hydrological benefits for the 1
in 200yr and 1 in 5yr flood events. It is acknowledged that the report indicates
other return periods could be added.
The drain blocking element of the tool does not take account of local soil
conditions, drain geometry or drain direction which are likely to play a significant
role in the both the magnitude and effect (reducing or increasing) flood risk.
(Refer to Section 3.3(D)
The adopted hydrological method does not conserve mass. That is, when the
flood peak is reduced this is achieved by cropping the top off the hydrograph.
This lost flow volume is not reintroduced later in the flood hydrograph. The
method of cropping the flood also introduces the curiosity that floodplain
restoration might have a notable impact at reducing flows around the peak of the
1 in 5yr event, but for a 1 in 200yr event there is no modification to the hydrograph
at that same flow rate, (refer to Figure 5).
-9-
Figure 5: Example of the output from the proposed Jacobs S20 assessment tool showing
the hydrological benefits of restoring the floodplain area (purple) to facilitate an increase in
flood depths of 300mm with a 50% efficiency factor.
The assessment tool is catchment outlet specific, therefore the tool is not
appropriate for national screening of Natural Flood Management opportunities.
In preparing the literature review, Jacobs did not undertake a review of existing
hydrological methods which have been used to represent the hydrological benefits
for Natural Flood Management measures. Nor does the review take account of
the recommendations for future Natural Flood Management assessment tools
such as those by OConnell et al. (2004):
The modelling should be distributed and be capable of running continuous
simulations. It should also be partially or wholly physically based so that the
physical properties of local landscapes, soils and vegetation can be represented,
and it should include detailed modelling of surface water flow so that the effects of
changes can be tracked downstream.
It is unfortunate that the Jacobs S20 is not aligned with these recommendations,
namely it is not a distributed model that is capable of continuous simulation, nor
does it have a physical basis or a detailed flow routing engine to permit the effects
to be tracked downstream.
The hydrological benefit of floodplain restoration does not include any component
to account for the hydrograph volume. The ratio of hydrograph volume to
floodplain volume is a key variable (Section 3.3(L)).
It is not clear how the Jacobs S20 tool can quantify Natural Flood Management
benefits in catchments that have significant lochs or reservoirs, particularly if these
waterbodies are artificially managed or are close to the catchment outlet.
-10-
It is unclear how hydrometric data, such as a real data estimation of QMed or the
catchments observed time to peak can be incorporated within the tool. It would
appear that the tool is hardcoded to remain a no data methodology.
Jacobs recommends that the S20 tool is only used to provide an initial broadscale
identification of possible benefits of various Natural Flood Management measures;
going on to say that measures should be more fully appraised before adoption.
This implies that a different method must be put in place to allow for the detailed
assessment and that the tool will not allow Natural Flood Management measures
to be directly compared against hard engineered measures. A robust hydrological
assessment framework which can be refined as the project develops, and which is
consistent with the underlying principles of current British hydrological practice,
would be more progressive.
The Jacobs S20 tool only provides a means of assessing the hydrological benefits
of Natural Flood Management measures in fluvial systems. Limited discussion of
a potential means to configure a saltflats restoration tool for the estimation of
reduction in estuarine surge is provided however there are notable gaps in the
methodology which is discussed only in broad concept. There is no discussion of
using saltflats to dissipate wave energy and the areas of groundwater and pluvial
flooding are not considered.
In some cases there has been criticism that the Jacobs S20 tool does not give
enough consideration to other benefits. While it would be ideal to have a tool
which would consider all benefits and disbenefits of Natural Flood Management,
to do so would be immensely challenging. For the time being, the focus of energy
should be on the development of a hydrological assessment methodology for
Natural Flood Management techniques thus facilitating the fulfilment of Section 20
of the Act.
There is a need for the tool to consider other forest mechanisms for flood
alleviation, namely infiltration and water use.
-11-
2.5 Summary
It can be summarised that the literature review undertaken by Jacobs is generally well
accepted. Subject to inclusion of a handful of papers, including some recently published,
the literature review should be considered authoritative and up to date.
The Jacobs S20 Tool is based on modifications to the CEH Flow Grid. In developing the
methodology for modelling the effect of Natural Flood Management measures Jacobs
have not drawn on the recommendations of OConnell et al. (2004)83 in regards to how
the model should be configured. Instead the method is centred on pro-rata cropping of
the hydrograph peak via best estimates made from the literature rather than adopting a
more physical based approach to modify parameters contained within the hydrological
model.
There are some concerns over the schematisation of the model, with the most notable
being related to the absence of a hydraulic based flow routing element to the model
(division of catchment into thirds). This formulation prevents accurate accumulation of
hydrological benefits moving down through the catchment towards receptors and the
assessment of delay/attenuation measures (floodplain enhancement and riparian
intervention). There are other areas of concern relating to how the tool accumulates the
benefits of multiple types of interventions, the absence of ground truthing and the
suggested benefits are questioned for some measures (floodplain enhancement).
The requirement to specify a single location as the catchment outlet will lead to
challenges in using the tool to undertake a national screening of Natural Flood
Management opportunities. The tool also lacks functionality to identify flood risk
management benefits for coastal, groundwater and pluvial flood mechanisms.
It is not recommended that the current formulation of the Jacobs S20 assessment tool is
used for the identification or appraisal of Natural Flood Management measures in
Scotland.
-12-
3 Literature review
3.1 Introduction
Natural Flood Management is a broad area covering many of the aspects of hydrology,
coastal science and their related subjects. This literature review will be divided into the
following sections:
A summary of the existing tools for screening and assessing Natural Flood
Management measures (Section 3.4)
reflect the issues in England and Wales. There is a tendency for Scottish Natural Flood
Management literature reviews to not be as scientifically robust as they have not been
subjected to a peer review process, but never the less they do form useful sources.
(A)
This was the first major review of the impacts of land use and management on flood
generation undertaken by Defra and the Environment Agency. The purpose of the study
was to review the factors contributing to runoff and flooding in the rural (managed, not
natural) environment, and to scope the research needed to improve the identification of
the management policies and interventions to reduce the impact of flooding. The study
focused on inland flooding mechanisms such as pluvial and fluvial flooding.
The study was undertaken by a team drawn from the disciplines of agriculture, soil
science, hydrology, hydrogeology and socio-economic science and provides much of our
current understanding of the science with respect to land use management and its impact
on flood generation.
It is worth noting that this review focused on land use and land management techniques,
therefore focusing on runoff generation and thus excluding other Natural Flood
Management techniques such as floodplain restoration, river restoration, washlands
creation and coastal realignment.
The main conclusions of FD2114 are as follows (taken from the Summary of the
document):
Significant changes in land use and management practices in the last fifty years
have resulted in a general intensification of agricultural land use;
There is only limited evidence that local changes in runoff are transferred to the
surface water network and propagate downstream;
Analyses of peak runoff records provide very little firm evidence of catchment
scale impacts of land use management;
There are many measures that can be taken to mitigate local flooding by delaying
runoff;
Rainfall-runoff modelling to predict the effects of changes in rural land use and
management on flood generation is in its infancy: there is no generally accepted
theoretical basis for the design of a model suitable to predict impacts; and
-14-
(B)
Update evidence from Defra Research project FD2114 on the impacts of rural
land use changes and management on flood generation in rural areas, at both the
local and catchment scale within the UK; and
The review was developed from published literature, ongoing studies and additional
information obtained through consultation with a number of key individuals between
January and March 2007. The review reflected the most up-to-date evidence and
provided an indication of where current research is being targeted.
The R&D update focused upon the most important land use management practices, as
identified within FD2114 and subsequent consultation with the Environment Agency.
Unlike FD2114, the update was limited to UK sources and evidence. The findings were
presented according to five main land use or management practices:
Agricultural drainage;
Pasture; and
Cultivation techniques.
For large catchments, existing modelling studies suggest that a large extent of
land-use or land management change is required to produce a relatively modest
reduction or delay in downstream flood peaks.
For particular soil types and geologies (and therefore locations within the UK),
groundwater becomes important for flooding. The effects of land use on
groundwater flooding remains an under-researched area.
Multiple interventions have and continue to hamper our ability to predict the
impacts from specific land-use changes.
-15-
There is strong evidence that upland conifer forests have a negligible effect on the
reduction of flood flows. (This point is subsequently contradicted implying that it
relates to out of date forestry practices.)
There is very limited potential for afforestation to reduce flood risk at catchment
scale. While upland plantations may reduce peak flows for smaller events, they
are thought to have a negligible impact on extreme events. Equally there is scant
evidence that deforestation increases flood risk although there is evidence that
poor logging practices may have temporary negative impacts.
There is evidence that at local scale there is some potential for strategic tree
planting to increase infiltration in compacted soils.
There is a growing body of evidence that suggests that riparian and floodplain
woodland can attenuate flood propagation.
Agricultural practices:
Agricultural policy has been a major driver of land management practices. As yet
there is little evidence of the impact of new agri-environment schemes.
The evidence shows that there is a need to consider the land management history
of a catchment down to the field scale when assessing its flows.
Tools have been and are being developed that help to predict the effect of artificial
drainage or drain blocking on the timing and magnitude of runoff. These should
help target interventions. Catchment scale models also need to be used to take
account of the location of the land-use change within the catchment and its
influence on the timing and magnitude at receptors.
Pasture
The Pontbren study will be important at bridging the gaps between hillslopes,
small catchment and downstream impacts on flood frequency.
-16-
(C)
This study was commissioned as part of joint Defra/EA Flood and Coastal Erosion Risk
Management R&D Programme. Its purpose was to provide an analysis of historical data
on land use and land management to ascertain whether any impacts of land use and
management change on flood generation could be identified. The change identification
methodologies used were Dynamic Harmonic Regression to examine seasonal scale
trends, and Data Based Mechanistic models were used to examine changes in storm
responses.
The study found that, in general the variability between years and inconsistencies in the
rainfall and flow data appear to dominate any tendency of other changes. Only seasonal
flows could be shown to exhibit any significant temporal trend and then only for two of the
nine catchments studied. The assessment of hydrograph responses showed no clear
changes over time. Where tendencies in hydrograph characteristics with time were
evident they were obscured by year to year variability.
The project concluded with a set of policy recommendations:
Both climate variability (particularly rainfall variability) and land use affect flood
runoff. Changes in discharge should not be analysed without consideration of
changes in catchment rainfall inputs.
Adequate information about past land management changes and soil conditions is
not readily available but will need to be collected and made available in future for
different land use categories if improved understanding of the links between runoff
and land management is to be gained and used at catchment scales.
The results of this project show that there will be a real difficulty of estimating the
benefits of land use change measures in respect of any reduction of flood risk.
Further monitoring of studies aimed at reducing runoff should be carried out to
evaluate the effectiveness of different types of measure at the local level in the
context of agri-environment schemes.
The difficulty in identifying consistent changes given the limitations of the available
data means that land management measures cannot be relied on as alternatives
to more proven flood risk management options.
The difficulty in identifying consistent change given the limitations of the available data
does not mean that change is not happening and should not be taken to imply a policy of
doing nothing. Contextually relevant management practice guidelines (linked to land use,
soil type, antecedent condition) should be developed and monitored to deliver multiple
benefits including reduced runoff generation and local flood risk.
-17-
(D)
Delivery of Making Space for Water, The role of land use and land
management in delivering flood risk management (Environment
Agency Halcrow, 2008)8
This study was commissioned by the Environment Agency with the objectives to:
Summarise the strength of underlying science and knowledge gaps with respect
to land-use/management impacts on flood risk;
The lumped sub-catchment hydrological model with 1-D flow routing model
approach used in the Ripon study potentially provides a simple framework for
testing future land use scenarios which could be applied generically to different
catchments. However, this approach is as yet untested on other catchments:
o
Scale issues -Land use change modelling is undertaken at a subcatchment scale and does not take account of localised changes;
The physically based distributed hydrological model used in the Parrett catchment
is considered to be more robust because surface, subsurface and groundwater
are better represented throughout the catchment using a grid approach. This
approach can potentially be more reliably applied to catchments with very different
characteristics and as the Natural Flood Management interventions are modelled
using physically based modifications to the model there is better scope to
accurately represent the accumulative impact of Natural Flood Management
measures at a catchment scale. The study identified:
o
Simple modelling tools (such as the lumped models applied to the Ripon) should
be used wherever only basic level information is available. It should be
recognised that simpler tools will provide simple and more approximate
predictions.
-18-
(E)
The Land Use Jigsaw is an ongoing (as of time of writing) research prioritisation tool
maintained by the Environment Agency that seeks to guide the development of policy
relevant research for Flood and Coastal Risk Management (FCRM) in England and
Wales. The tool, which takes the form of a community maintained register, seeks to
promote current research and determine gaps and overlaps in evidence. It is worth
noting that the Land Use Jigsaw has been focused on the evidence base for individual
measures and has not considered the combined effects of measures on reducing flood
risk. In addition to this it should be noted that it is an Environment Agency tool and
therefore focused on issues faced by the Environment Agency in England and Wales.
While there are marked similarities between hydrological processes across the British
Isles care should be taken in applying the findings in Scotland.
The starting point for the Land Use Jigsaw was the current understanding as presented in
previous studies as discussed above. In summary:
Effects of land use management on flood risk at a large catchment scale are
difficult to determine as they require aggregating many local-scale effects and are
dependent on catchment characteristics;
The lack of understanding does not necessarily mean that there is no catchmentscale effect, but rather that the nature of the effect differs between different
catchments and is difficult to detect;
Although there can be benefits for local flood mitigation, the evidence to date has
not demonstrated the reduction of extreme flood events at the catchment scale
when using most land management interventions. The project will continue to
support research into large catchment scale impacts of land management
changes on flood risk, water quality and resource protection and will actively seek
partnership projects; and
The key priority research questions identified in the 2008 summary report are
summarised in Table 1.
-19-
No.
Evidence
Need
Category
Question
Q020
Evidence
gap
Needed
now
Under
sowing
Q031
Evidence
gap
Needed
now
Q038
Evidence
gap
Needed
now
Q039
Evidence
gap
Needed
now
Q070
Evidence
gap
Needed
now
Wetland
creation
Q088
Evidence
gap
Needed
now
On farm
storage
The most significant gaps in scientific knowledge identified in the 2008 summary report
are summarised in Table 2.
No.
Evidence
Need
Category
Question
Q112
Ongoing
work
Needed
now
Grip
blocking
Q113
Ongoing
work
Needed
now
Grip
blocking
Q121
Ongoing
work
Needed
now
Peat
3a Where in the catchment will this be
restoration most effective?
Q122
Ongoing
work
Needed
now
Peat
3b How long does the intervention last?
restoration
Q139
Ongoing
work
Needed
now
Floodplain
woodland
(F)
This report was commissioned by the National Trust to provide an overview of the current
literature in order that future operational policy is relevant and informed. It focuses on
changing land use impacts upon runoff and the impact of land management upon
catchment level studies (>100sqkm).
The study used a systematic approach to identify published and unpublished research
relating to land use/management and runoff with a strict quality control procedure. In
total 29,772 articles (including duplicates) were indentified by the literature search. Of
these articles, 108 were identified as sufficiently relevant for full text viewing and
-20-
Need to adopt the precautionary principle when assessing the influence of land
management upon runoff;
At a large catchment scale there is some evidence that land use changes impact
runoff;
From the papers reviewed it is not clear what impact land management has on
different return periods; and
-21-
(G)
This report was commissioned by Scottish Environment LINK to inform the development
of flood risk management policies in Scotland. The non-peer reviewed report focuses on
how Natural Flood Management is incorporated within sustainable flood risk
management. The key messages from the report summary include:
Some previous river restoration projects have indicated that there are benefits for
flood management but very often the claim is not substantiated by modelling or
data collection;
There is clear evidence that riparian vegetation will slow down flows in channels
and increase local floodwater storage. There is a need to understand the role of
vegetation type over a floodplain and the ability to temporarily hold back water
(the leaky barrier principle); and
Several studies have been carried out which demonstrate the benefits of some
agricultural practices in flood management. These are mostly related to
increasing infiltration rates into the soils and reducing the amount of artificial
drainage.
(H)
Upland soils;
Upland woodlands;
Upland streams;
Farm management;
Washlands;
The review provides valuable information about the selected Natural Flood Management
measures over a broad front, addressing the following areas:
Hydrological benefits;
Other benefits
The review is less constrained by the hard science approach which forces a more
conservative stance to be reported by government funded studies such as those
-22-
Blocking drainage has complex hydrological response, often having both positive
and negative impact on flood peaks. The restoration of peat is thought to reduce
the 'flashiness' of the catchment, but once saturated, this effect may be reduced.
Planting of coniferous species can have all year round benefits; however, during
and after felling of commercial plantations, peak flows can increase for several
years.
Large woody debris can play an important role in storing and attenuating flood
flows.
Land use and management has significant effect on the soil's capacity to
attenuate water. For example, some arable practices increase the risk of local
muddy floods, and changing the direction of tillage can delay or reduce run-off.
The effect of drainage in lowlands has a complex hydrological effect, and the
response depends upon geology, soil and drainage network. Reducing drainage
capacity on permeable soils can reduce catchment flood risk by slowing and
reducing water export from farmland to the river channel.
Intertidal habitats play an important role in absorbing wave energy and reducing
wave height.
Coastal realignment increases the area of the tidal floodplain, making space for
tidal flooding and thus reduces estuarine surge. The benefits of managed
realignment and tidal exchange have been well demonstrated.
-23-
(I)
Riparian woodland;
Floodplain woodland;
The types of trees planted is important, with annual evaporative losses being
greater from conifer forests than from deciduous;
Seasonality of leaf cover, which may introduce variability in the efficiency of flood
attenuation; and
Other issues that need to be considered relate to physical constraints upon commercial
forest cultivation, land use change and possible effects on sediment deposition and
erosion rates.
The study identified that the 10% attenuation of flood flows from a fully forested
catchment are most likely during a high order flood event and that a 25% reduction is
likely for a low order event.
The literature review considered investigations via monitoring of catchments and by the
use of hydrological modelling. Summary of the reviewed literature is provided in Table 3.
-24-
Riparian Woodland
In terms of flood management benefits, riparian woodlands function in a similar way to
floodplain woodlands by increasing roughness, slowing the flow and increasing storage but on a different, smaller scale. In addition, the presence of large woody debris within
the stream channel can delay flood flow, promote out of bank flow and increase storage.
These riparian 'buffer strips' can also have wider benefits associated with sediment
control and erosion, protection of water quality and maintenance of habitat diversity.
The study reports that the literature indicates that 60 minute delays in flood propagation
are achievable for both upper and middle catchment locations. These benefits were
found to hold true for both low and high order flood events. A summary of reviewed
studies is provided in Table 4.
-25-
Floodplain Woodland
Floodplain woodlands can provide flood risk management benefit by providing a physical
obstacle to the movement of water. In addition, woodlands also enhance soils storage
potential, by improving infiltration and increasing water uptake. The increase in hydraulic
roughness and diverse network of backwaters and ponds characterised in natural
floodplain woodlands increases water storage, retains water on the floodplain and could
therefore have significant potential in flood management.
The two main studies reviewed (the River Carey and the River Laver) suggest that for an
increase in floodplain roughness alone, due to woodland planting, the increase in flood
water level could be up to around 300mm, possibly higher. These studies consider high
order events and though there is a suggestion in some studies that floodplain woodland
may be less effective during extreme events this range is still considered reasonable (on
the basis of the Carey and Laver studies). A summary of these studies is provided in
Table 5.
-26-
(J)
This report was prepared in response to Sir Michael Pitts recommendation for Defra, the
Environment Agency and Natural England to work with partners to establish Catchment
Flood Management Plans. The report provides additional explanation on what natural
processes are in flood management and to assist practitioners in achieving greater
working with natural processes. While the report does provide some commentary on the
evidence base for Natural Flood Management measures, this is considerably lighter than
the commentary provided in other literature reviews. There is some beneficial guidance
on how to maximise project and programme success, the guidance is aligned with the
system in England and Wales and is not applicable beyond general principles in
Scotland. Finally the report provides an overview of current initiatives to support the
increase in flood risk management via Natural Flood Management methods:
The Land Management CFMP Tool The EAs CFMP tool allows the quantitative
assessment of land use and land management changes on the flood hydrograph.
-27-
The tool helps target land management approaches that reduce flooding and
potentially opens up the comparison of Natural Flood Management measures
against traditional engineering.
Appraisal Guidance the Environment Agencys Flood and Coastal Erosion Risk
Management Appraisal Guidance includes greater working with natural processes
for future flood and coastal erosion projects. It encourages multi-criteria analysis
through the assessment of environment, social and economic benefits.
Land management flooding resource pack the Environment Agency and Natural
England are currently working together to develop a training pack intended for
catchment sensitive farm officers, environment officers and agri-environment
advisors.
Flood storage area project to allow for greater use of flood storage areas, and
for greater biodiversity benefits when flood storage areas are constructed, a tool
has been developed to assist in planning and design. It includes a GIS database
of previous flood storage schemes, a cost estimating tool and a design decision
flow chart.
-28-
-29-
(A)
This section discusses land use issues including grazing and land management
practices. Given that there is a notable volume of literature relating to upland drainage
this area is discussed separately in Section 3.3(F).
Upland soils, which are typically peat based in Scotland, are sensitive and can be easily
degraded by land use practices such as overgrazing, drainage or poorly conducted
muirburn. In many areas this has led to vegetation change and retreat, often exposing
peat causing it to desiccate in dry weather, leaving it prone to erosion from both wind and
water. Over the decades this has lead to measurable impact upon the timing and
magnitude of floods (Longfield et al, 1999; Conway and Millar, 1960; Robinson,
1960)131415.
Reducing stocking density encourages regeneration of upland vegetation communities,
including ericaceous plants that tend to have higher rainfall interception values compared
to grasses and bare ground. For example heather provides a consistent year round
interception loss (16 to 19%) although this high interception rate is somewhat offset by a
relatively low transpiration rate (200 to 420mm/yr) (Nisbet, 2005)64. Note that the effects
upon vegetation will be seasonal with the magnitude of these fluctuations being
dependent on the vegetation type.
Lower livestock density can also reduce trampling and subsequent compaction of the soil,
reducing the antecedent water content and increasing the infiltration capacity (Godwin
and Dresser, 2003)16.
The shooting of grouse is one of the main forms of income for the uplands and despite a
recent decline in grouse numbers grouse shooting has continued to remain a popular and
profitable sport in the UK (Davies, 2005)17. The decline has been linked to several
reasons which include the loss of heather moorland and subsequent closure of several
shooting estates, reduction in gamekeepers managing the moor, disease and an increase
in predator species (Cox, 2003; Hudson, 1992)1819.
Prescribed burning of moorland heather (Calluna vulgaris), often referred to as muirburn,
is a traditional management tool to improve grouse densities for sport shooting. Burning
helps prevent establishment of woody species, reduce plant litter and release nutrients
(Gimingham, 1975)20. Such management then stimulates earlier growth of vegetation
and temporarily increases the accessibility, palatability, and nutrient content of forage for
grouse and livestock, thus optimising the populations for maximum profit (Lawton, 1990;
Tucker, 2003)2122. Although most of the burning is conducted on shallow peat and
podzolic soils, it also encroaches onto adjacent areas of deeper peat found on blanket
bog (Yallop et al, 2006; OBrien et al., 2006)2324.
The mosaic of different heather ages created by muirburn is viewed by Scottish Natural
Heritage as offering benefits to native wildlife such as birds, insects and reptiles (SNH,
2012)25. The Single Farm Payment require moorland to be maintained in Good
Agricultural and Environmental Condition, the Muirburn Code (Scottish Government,
2011)26 is used to define the standards expected of land managers.
The replacement of a thick blanket of mature heather and associated understory of litter
with a patchwork of immature bare ground, heather and grasses will reduce the canopy
interception from that reported above for heathers to much lower rates. It is also reported
that the infiltration capacities of cotton grass moorland were as low as 2mm/hr while
cowberry areas have an infiltration rate just less than 20mm/hr (OBrien et al, 2007)27. A
mature ground cover will have a higher hydraulic roughness and will serve to slow
overland flow (Lane et al., 2003)28.
A study by English Nature on the effect of heather burning on soils, hydrology and
ecology summarises the key literature relating to muirburn. Work by Mallik et al (1984)29
-30-
found that infiltration rates on freely drained brown podzolic soils declined by 74% on
burnt heather plots compared to unburnt plots, probably as a result of ash particles
clogging the soil pores in the upper surface layers. Their results concur with similar
studies in other countries, but it is worth noting Kinakos (1975)30 Phd thesis reported that
burning in a heathland in Scotland increased infiltration by 30%. Mallik et al suggest that
Kinakos method of measuring infiltration may have been unsatisfactory, as it was
unsuccessful when tried in their study. Mallik et al (1984)29 also reported that the water
retaining capacity of the burnt podzolic soils was significantly higher than the unburnt
soils, particularly in the upper 2 cm of the soil profile but the difference was detectable to
10 cm. This would have had the joint effect of reducing the rate of water percolation
through the soil and increasing its water retention capacity.
Severely burnt soils would be expected to generate an increased proportion of rapid
runoff, through increased water retention in the soils and the absence of
evapotranspiration and rainfall interception by vegetation (Mallik et al., 1984)29. An
experimental plot scale study on Dartmoor, observed no increase in runoff from an area
of managed burn (Meyles, 2002)31. Further to this no differences in soil moisture levels
were detected by Meyles under wet conditions between burnt and unburnt areas, for up
to two months after burning.
One of the few catchment studies on blanket peat was carried out in the 1950s at Moor
House in the Pennines, the results of which suggested that severe burning of the peat
surface reduced the water storage capacity of the soil and lowered dry weather flows, but
increased peak flows in drainage ditches (Conway and Millar 1960)32. By analysing a
longer sequence of Conway and Millars data, Robinson (1985)33 confirmed their
conclusions. Although moderate burning, without total destruction of the mire community
or the heather, tended to produce a flashier runoff hydrograph, the effect was very much
less than that of drainage by gripping.
A study in the North York Moors, reported via a Phd thesis (Dunn, 1986)34, identified that
land-use change from open heather moorland to burnt ground promoted reductions both
in evapotranspiration levels, especially potential demand and in moisture deficits. The
study identified that complete catchment muirburn could lead to a doubling of the
hydrograph peak predominately because of the increased antecedent moisture condition.
Ramchunder et al (2009)35 highlights the absence of detailed knowledge relating to the
hydrological impact of rotational burning and provides some hypotheses on the effects
that rotational burning might have at a catchment scale. These predictions are
judgement based.
The EMBER Project (Effects of Moorland Burning on the Ecohydrology of River systems)
at the University of Leeds includes increasing the understanding of moorland burning on
hydrology (EMBER, 2012)36. The project is using a paired catchments approach
consisting of five control catchments and a further five were muirburn is undertaken. The
catchments in the Pennines are all less than 3sqkm. Unfortunately to date the project
has not published any material.
(B)
In relation to the grazing of pasture DEFRA research project FD2114 (OConnell et al.,
2004)83 concludes that:
There is evidence from field sites of reduced infiltration and increased field-scale
surface runoff, when some livestock management practices are undertaken in less
-31-
than ideal soil and weather conditions (Holman et al., 2001; Goodwin and
Dresser, 2003; Hollis et al., 2003)394041.
DEFRA research project (BD2304) (Cranfield University, 2007)42 reported that there is a
clearly demonstrable link within the literature between soil compaction and livestock
densities. This finding echoes the concluding message of an earlier Environment Agency
R&D project (Johns, 1998)43 which was predominately based on anecdotal and prima
facie evidence. Compaction damage due to trampling may also be cumulative with time
but this could be dependent on soil texture. Therefore the 2007 DEFRA project adopted
the following approach:
The study identified that the vulnerability of soils to compaction can be estimated using
the Soil Suitability for Grassland method proposed by Harrod (1979)44 with the method
being based on:
Unfortunately it is thought that such a dataset of the Harrod (1979)44 soil compaction
vulnerability does not exist for Scotland nor is it clear if it is technically feasible for an
identical Scottish version to be produced. Further to this, it is not clear if an alternative
approach could be developed. Work by Palmer (2004)45 using a technique developed
during work on flood-affected catchments in 2000 provides the most significant body of
evidence of the extent of compaction but only covers twelve catchments mostly in the
south and west of England (one is from Wales) and the sites cannot be claimed to be
representative of Britain as a whole. By analysis of Palmers published field data the
2007 DEFRA research project identifies a correlation between vulnerability and actual
degradation in so far as low vulnerability soils were found to have predominantly low
levels of actual degradation and, conversely, high vulnerability soils were linked to the
highest proportion of high degradation sites. It also suggest that soils under ley
grassland (younger than 5 years) are more susceptible to compaction that those under
permanent grassland.
Modelling using MIKE SHE of the River Camel catchment (Williams, 2004)46 suggested
that minor changes in the moisture characteristics of soils beneath pastures (12% by
catchment area) produced an increase of approximately 18.75% increase in flood peaks.
These changes are presented within the paper as being representative of degradation
due to intensive grazing. A site based study of a small catchment in Dartmoor identified
that the soil water field capacity is lower for heavily grazed pastures (Meyles et al.,
2006)47. The study also proposes that intense grazing causes improved rapid surface
flow routes due to the presence of trampled livestock tracks. The dominance of the
saturated excess overland flow pathway within grazed grassland is highlighted following
observations within the Pontbren catchment (Wheater et al, 2008)104. Observations at
Pontbren also indicated that, for the soils present in the study area, surface runoff as a
consequence of exceeding the soil infiltration capacity was not detected.
The 2007 DEFRA research project (Cranfield University, 2007)42 assessed the soil stress
class based on a simplified Grazing Livestock Unit (GLU) density (per unit area)
calculation where all cows were assigned a value of 1GLU and sheep a value of 0.1GLU.
Parish livestock statistics were used to gain an estimate of local livestock numbers. It is
-32-
then assumed that soil stress is directly proportional to the GLU density allowing the GLU
density map to be divided into three classes (high, medium and low). The report accepts
that practices such as winter grazing and trafficking are omitted from this approach
however in the absence of viable data it is currently the only means of gaining at least a
slight insight into the compaction caused by livestock. The GLU density was calculated
on a 1km grid for England and Wales for all grid cells with more than 40 percent
grassland. It should be noted that the density was not based on the area of grassland
within the 1km grid cell as this reportedly caused the inappropriate shifting of high stress
areas to the lowlands. Instead the density was based on the number within each 1km
grid cell regardless of grassland proportion.
The spatial estimates of vulnerability to compaction and the stress which soils are under
was then combined to create a measure of the severity of compaction using Figure 6.
42
Following the identification of a method to estimate the spatial distribution and severity of
compaction the 2007 DEFRA research project (Cranfield, 2007)42 identifies the work
undertaken by Packman et al. (2004)48 to quantify the hydrological impacts of
compactions. This methodology is based on the assignment of an appropriate analogue
HOST (Hydrology of Soil Types, Boorman et al., 1995137) class to represent the degraded
soil. The analogue HOST class can then be used to estimate the Standard Percentage
Runoff (SPR) and Baseflow Index (BFI). The rationale for the proposed changes is that
soil structural degradation, in the form of topsoil and upper subsoil compaction or
seasonal capping and sealing of soil surfaces, causes a reduction in the effective soil
storage, which in turn results in increased surface runoff. The method recommended that
the modified HOST is adopted for all areas within the catchment which are cereal or
lowland grass, with all other areas outside this being based on the published HOST class
by Boorman et al. (1995)49.
The DEFRA research project (Cranfield, 2007)42 reported using a modified version of
Packman et al. (2004)48. The study identified an indicative absolute increase in SPR as
an average of less than 6% for the England and Wales study area. The largest increase
-33-
in SPR was identified for catchments with a Boorman et al. (1995)49 SPR of 25 to 45%
(undamaged). Catchments with high SPR showed little impact from soil compaction due
to the lower density of managed grasslands and wetter antecedent conditions. The small
absolute increases in SPR represents relative increases of less than 13% in most
catchments, although a number (mostly in the south west of England) showed indicative
increases of 13 to 41%. It should be noted that the relative increase in SPR are typically
much lower than those observed for plot studies (Heathewaite et al., 1989; Heathewaite
et al., 1990)3738. The DEFRA research project (Cranfield, 2007)42 suggests these
differences are due to the moderating effects of landscape connectivity.
Although modification of HOST class proposed by Packman et al. (2004)48 was initially
developed to be used with the Flood Estimation Handbooks Rainfall-Runoff Method via
the revision of SPR, it is proposed that the same approach could be used to modify BFI
values which are used within the Revitalised Flood Hydrograph Method (Kjeldsen,
2007)50.
(C)
There is a significant body of information reporting that arable land has an impact on
runoff generation (Cranfield, 2007)42. The most relevant literature regarding arable crops
appears to have been completed by Holtan and Kirkpatrick (1950)51. This American
observation based research indicates that long established pasture has a higher
infiltration rate than arable crops, especially during periods of the year when cropped
ground is bare or crushed. It should be noted that the published rates do not account for
soil type or for variation in land management practices. Importantly, the size of
agricultural machinery has changed dramatically since 1950, thus present day soils are
likely to be more trafficked and consequently more compacted (Cranfield, 2007)42.
-34-
51
Soil exposure
Land slope
Soil degradation
Crop cover
Hillslope form
Through assessing the degradation of the land the tool assists in the selection of an
alternate HOST class using the guidance provided by Packman et al. (2004)54. It should
be noted that the guidance reminds the user that the FARM tool functions at a local field
or farm scale while the FEH Rainfall-Runoff Model (and ReFH Rainfall-Runoff Model) is a
catchment scale lumped hydrological model. In applying the methodology it is therefore
necessary to appropriately weight the baseflow index (BFI) or standard percentage runoff
(SPR) value based on the prevalence of the degraded land within the modelled
catchment.
(D)
Field drainage
This section discusses the literature relating to the blockage of agricultural field drainage,
upland drain blocking is discussed in Section (F).
Field-drainage and associated subsoil treatments can increase or decrease peak drain
flows and the time to peak flow by as much as two to three times either way with the
behaviour apparently dependent on soil type and wetness (OConnell et al., 2007)55.
These very variable impacts of field-drainage systems are likely to render flood
management gains from changes to field drainage systems a complex, data intensive
process which retain a high level of uncertainty. It was noted that drainage was found to
alter runoff timing, but it had little impact on runoff volume.
Institute of Hydrology Report 113 (Robinson, 1990)56 reported a comprehensive series of
field studies, catchment studies and regression analysis to investigate the impact of field
drainage on flood flows. The study found that general statements that field drainage
causes or reduces downstream flooding are over-simplifications of a complex process.
At a field scale it was concluded that agricultural drains reduce peak flows originating
from clay soils while having the opposite effect for permeable soils. The type of drainage
was also shown to be significant with moling in combination with pipe drains increasing
peak rates, while open ditches give the highest discharge rates. At a catchment scale it
was identified that it was necessary to differentiate between open channel drainage and
piped drainage. Arterial open drains were found to increase flows downstream due to
accelerated flows and reduced overbank flooding, with the effect increasing with channel
size and flow.
-35-
(E)
Afforestation of uplands
This section discusses the literature relating to the change in runoff rate and timing which
afforesting uplands can introduce. For details of literature relating to the attenuation and
other effects which can be introduce by placing vegetation within flow pathways (riparian,
floodplain and gully woodland) refer to items (K), (L) and (I).
Much of Britains native upland woodland has historically been felled, as well as being
lost to a combination of overgrazing and burning over time. That which remains is often
degraded because natural regeneration is prevented by grazing and browsing pressure
from sheep and/or deer, which have increased in population over the last 50-60 years
and which often use the woods for shelter (Hardiman et al., 2003)57.
DEFRA commissioned a major study (FD2114) (OConnell et al., 2004)83 which reviewed
the impacts of a range of rural land uses including forest on flood generation. In the
reports critical assessment of the monitored evidence for afforestation affecting river
flood flows the following was concluded:
There is consensus within the literature that the introduction of woodland cover can
influence a number of key hydrological processes, namely:
Moisture transpiration
Rainfall interception is the capture and subsequent evaporation of the water before it
reaches the ground surface, it is the difference between the gross rainfall above the
vegetation canopy and the net rainfall at the soils surface. Broadleaved woodland in the
UK typically intercepts 10 to 25% of annual rainfall, compared to 25 to 45% for conifer
stands (Calder et al, 2003; Hall; Nisbet, 2002)585960. Fitting with what might be reasonably
-36-
expected, it has been observed that the greatest interception for broadleaved trees is
when they are in leaf, averaging 40% or more (Roberts and Rosier, 2005)61, in contrast
with 3 to 12% when trees are leafless (Roberts and Rosier, 2005; Harding et al, 1992)6261.
Hardiman et al (2003)57 proposes that this high summer interception rate may become
increasingly important in attenuating local run-off given climate change predictions citing
an increase in the frequency and intensity of summer storms. However the seasonality of
Scottish flooding is reported as being predominately autumn/winter (Reed et al., 1999)63
and the assumption that interception losses within the tree canopy will be present during
the design event peak is possibly overoptimistic. Trees with lighter canopies, such as
ash have been observed to intercept only 10 to 15% per annum compared to 15 to 25%
by oak and beech (Roberts and Rosier, 2005; Harding et al, 1992)6162. It is reported that
the annual interception rates (by percentage) remain remarkably constant over a wide
range of total rainfalls (Nisbet, 2005)64. A summary of the reported interception rates for
conifers and broadleaf species is provided in Table 6.
It is reported that interception loss as a proportion of rainfall reduces with increasing
storm size, reaching a maximum of 6-7mm/day (Nisbet et al, 2011)65. However,
interception varies greatly throughout the year and in particular, declines with the size
and intensity of a given rainstorm. This reflects the relatively small water holding capacity
of forest canopies, equivalent to only a few mm of rainwater. As a result, interception
losses are likely to be <10% for individual major storm events (Nisbet, 2005)64.
It has been reported that transpiration rates vary little between the two forest types, with
annual losses mainly falling within a relatively narrow range of 300350mm (Roberts,
1983)66. Higher values of 390 to 410mm have been derived for native broadleaved
woodland in southern England, with a range of 360mm to 490mm varying by study
(Roberts and Rosier, 2005)61. It can be inferred that if both interception and transpiration
are considered together, and assuming an annual rainfall of 1000 mm, conifers could be
expected to use some 550 to 800mm of water compared with 400 to 640mm for
broadleaves (Nisbet, 2005)64. Other factors influencing use of water are woodland
structure and age, however given the already complex issue between transpiration and
flooding these will not be discussed. A summary of the reported annual transpiration
rates for conifers and broadleaf species is provided in Table 6.
Land
cover
Conifer
Broadleaf
Grass
Heather
Bracken
Arable
Transpiration
(mm/yr)
300 350
300 390
400 600
200 420
400 600
370 - 430
Annual
interception
(%)
25 45%
10 25%
16 19%
20%
-
Winter
interception
(%)
25 45%
3 12%
16 19%
?
Summer
interception
(%)
25 45%
40%
16 19%
50%
?
Table 6: Summary of annual transpiration and interception data for a range of land cover
64
types (Nisbet, 2005)
There is a general consensus in the published literature that, within a British context,
woodlands reduce annual water yield (Calder, 2009)67. The reduction in water yields as a
consequence of the increased evapotranspiration by trees might be of concern for
catchments which are already water stressed.
It should be noted that soil type and condition are significant in determining the pathways
and consequent timing of water draining from the land. Shallow, poorly draining soils are
prone to rapid surface runoff responses, while deep, freely draining soils facilitate much
slower, more attenuated, groundwater pathways. Native woodland rarely involves the
formation of continuous cultivation channels or drainage treatments to aid establishment
-37-
and the lack of disturbance helps to increase soil organic matter and improves soil
structure, resulting in an increase in ability infiltrate and store water (Forestry
Commission, 1999)68. Observations in Pontbren in Wales have found infiltration rates to
be up to 60 times higher under young native woodland than heavily grazed pasture, with
rapid improvement over just two years after stock removal and planting (Bird et al.,
2003)69. More recent modelling work undertaken by the same team suggests that whole
catchment forestation would reduce peak flows by 10 to 54% while being reduced by 2 to
11% via the use of optimally placed shelter belts (Jackson et al, 2008)70. Infiltration and
permeability will be predominately characterised by the soil type however there does
seem to be consensus that forest soils are more permeable than grassland (Aaloui,
2011)71. Therefore, by the conservation of mass they are a reasoned choice when
deciding land cover for flood management purposes.
Forest Research undertook a review of Short Rotation Forestry (SRF), incorporating
Short Rotation Coppice (SRC) (Nisbet et al., 2011)72. This review identifies that the fast
growing species often used in SRF and SRC have a high water demand. This increased
water demand is attributed as a consequence of high transpiration rates with interception
being comparable to conventional forestry. The review focuses on the issue of reduced
water yields particularly in relation to already stressed waterbodies, however the potential
to reduce groundwater levels, and by extension groundwater flood risk, is discussed
using the example of Eucalyptus plantations (Calder, 1992)73. Calder observed an Indian
plantation to have evapotranspiration rates exceeding the available rainfall (3400mm/yr
and 2100mm/yr respectively), it is anticipated that similar effects may be observed using
high water demand native species.
The development of the Flood Studies Report (NERC, 1975) established a link between
catchment wetness (and in particular the presence of any soil moisture deficit) and the
resulting design flood magnitude. It could be hypothesised based on this evidence that as
afforestation leads to an appreciable alteration of the typical seasonal soil moisture
deficits experienced in a catchment the reduction may be sufficient to influence the likely
magnitude of floods. However it should be noted that the Flood Studies Report
concluded that the extent of woodland in the UK is not statistically significant in affecting
peak flows of large flood events and hence the absence of a forest cover catchment
descriptor within the Flood Studies Report and subsequent Flood Estimation Handbook
hydrological methods.
A study at Leadburn, Scotland, implicated pre-planting plough drainage of deep peat as a
major source of local runoff (David and Ledger, 1988)74, whilst other research has
focussed upon the influence of forestry tracks and localised soil compaction. Historically
speaking intensive drainage and plough based planting techniques have been employed
in commercial conifer plantations (Rutter, 1963)75, such techniques are now out-dated
(Forestry Commission, 2011)76. That said, 30 years of research at Coalburn, England,
indicates that even with best practice, felling induces an increase in peak flows for
several years after felling, until new trees are well established (Robinson, 1986; Robinson
et al., 1988; Archer and Newson, 2002)777879. Because of the clear-fell cultivation cycles,
commercial forestry may be expected to have a much more variable medium-term effect
upon runoff and flood generation than native woodland of any type which is likely to be
more consistent over this timescale but may alternatively experience seasonal
fluctuations. This is supported by a broad review of upland afforestation across Europe,
which concluded that complete clear felling of a catchment can increase annual peak
flows by 10 to 20% with complete afforestation having a similar but opposite effect
(Calder et al., 2008)80. Although the cycle varies by location and species among other
variables, typically commercially grown conifers in the upland basins of North West
Europe have a short management cycle of 30 to 60years (Robinson et al., 2003)81.
The influence of native woodland restoration on upland hydrology will thus depend in part
upon what it replaces, and its situation within the catchment. For example, a literature
-38-
review based study for the Woodland Trust (the study itself being peer reviewed)
concluded that restoring native broadleaves from upland conifers could lead to a 10%
increase in localised flooding of minor floods (up to 1 in 5yr) due to lower water use and
greater potential runoff in winter and spring (Calder et al., 2009)67. However, replacing
rough or unimproved grassland with native woodland should have a roughly equal and
opposite effect, with further attenuation in summer from species with a high water use
such as willow (Calder et al., 2008)80. Conversely, research in the UK to date does
demonstrate that changes to hill slope woodland is unlikely to have a measurable impact
upon runoff or flood hydrographs in extreme rainfall events (McCulloch and Robinson,
1993)82.
There is an absence of evidence for afforestation in catchments greater than around
10sqkm (OConnell et al., 2004)83. This is in part due to the complexity of measuring
these effects when compared to other activities occurring within the catchment and
climatic variability. Modelling in Plynlimon in Wales found that it was frequently too
challenging to discriminate the effects of woodlands against, for example, climatic
variation or farming pressures further downstream (OConnell et al., 2004; Robinson and
Dupevrat, 2005; Kirby et al., 1991; Neal, 1997)84838586. This is likely to be a factor of the
very low woodland coverage in the UK. The Woodland Trust review (Calder et al., 2009)67
indicated that unless the coverage is more than 20%, the effects of woodland planting
would be unlikely to have any significance on the flood hydrograph. That said, it should
be noted that a likely mechanism for the achieving the forestation of catchments could be
via a large number of small schemes leading to a more notable net effect rather than a
small number of extensive whole catchment interventions, so it would be unwise to
consider in isolation that a small intervention is too minor. Conversely, decisions to
undertake extensive planting raises implications for other considerations such as habitat
priorities in a given upland area, such as open moor and semi-natural grassland
biodiversity, farming and other land uses. Nevertheless, the relative scarcity of seminatural woodland in the UK, especially in upland areas, may provide a weight for its
expansion.
Furthermore, much of the evidence of woodland effects upon flooding has been focussed
beyond the UK (particularly in the USA), where landscape, forestry practices and
woodland extent and type can be significantly different, making comparisons difficult. The
Centre for International Forestry Research cautions against referring to woodlands as
sponges and suggests that ill-informed policies can have detrimental effects on local
communities (CIFOR, 2005)87. Based on a recent report (CIFOR, 2005)87 the Centre for
International Forestry Research agrees with the UK literature consensus that
predominantly forested catchments often show little evidence of actually preventing large
scale floods. The European Basins (Robinson et al., 2003)88 report concluded that
forestry appears to have a relatively small part to play in managing regional or large scale
flood risk.
Jacobs (2011)3 provides the following concise summary of published catchment
monitoring and catchment modelling studies relating to the impact of forests on flood
flows:
a) Catchment monitoring findings
Catchment monitoring studies in the UK are few. The Institute of Hydrology (now
known as the CEH Wallingford) ran a number of long-term research catchments
studying the effects of commercial coniferous forest on hydrology (viz Plynlimon in
central Wales, Balquhidder in the central southern Highlands of Scotland, and
Coalburn in northern England). The Balquhidder work did not include an assessment
of the effects of forestry on flood flows, whilst the Coalburn study has been more
focused on quantifying the effects of drainage ditches, though Robinson (1986)15
-39-
does speculate that the negative effects of drainage ditches are in part balanced out
by the establishment of the trees.
The Institute of Hydrology used Plynlimon to assess the forest influence on floods
(Kirby et al., 1991)85. The findings of this were that an identifiable effect could be
detected for small floods (<2-year events) but that no effect was discernable for
larger ones. These findings could however be partially affected by the presence of
the old style drainage ditches used in the preparation of the land at the planting
stage.
The findings of other catchment studies were sought in the review of Price et al
(2000)99. Only relevant studies from the USA were identified at the time. These all
considered small (<1.5 km2) upland catchments that were instrumented to assess
the hydrological effects of clearfelling. These studies were relatively short term and
tended to sample events of limited magnitude. Bates and Henry (1928)89 observed a
55% increase in flood peak flows and a 23% increase in flood volumes immediately
after the clearfelling of a natural mixed conifer and deciduous forest cover in
Colorado. Hornbeck et al (1970)90 observed an average 13% and 21% increase in
flood peaks and volumes respectively after a mixed conifer and deciduous vegetation
cover was clearfelled in New England. Rothacher (1970)91 however detected no
significant difference for the largest floods when a Douglas Fir forest was clearfelled
from a catchment in Oregon. Swank and Crossley (1988)92 observed peak flow
increases of 15% and flood volume increases of 10% when a mixed forest was
clearfelled with minimal ground disturbance in North Carolina. Higher increases of
30% (peak flows) and 17% (flood volumes) were observed in an adjacent catchment
when it had 48% of its tree cover clearfelled together with 29% thinned and 23%
undercut. In all, apart from the Swank and Crossley study, the possible compaction
of soil during the felling process cannot be ignored.
Other approaches investigating whether the signature of land-use and land-use
change can be detected within the flood flows of the UK river gauge network have
been tried. Beven et al (2008)7 attempted to identify catchment-scale effects of land
use and land management change using available UK datasets, but failed to identify
a clear relationship between these and flows.
The DEFRA FD2114 study was updated in 20076. This summarised the conclusions
of the original FD2114 project with respect to afforestation/deforestation in the
following additional manner:
-40-
There is only a limited potential for forests to reduce peak flows for larger
flood events, with a much greater effect on small to medium magnitude flood
events.
Their updated (circa 2007) interpretation of the overall role of forest in terms of the
hypothesis that forests reduce flooding was:
In summary, this hypothesis cannot be accepted for the generation of extreme flood
events at a catchment scale on the basis of available evidence. Nevertheless, well
managed forests can help reduce local flooding and the peak flows of smaller, more
frequent events and potentially for infiltration-excess driven extreme summer
events.
b) Modelling studies
An alternative means of investigating the likely influence of forest is to use modelling
techniques that capture the key land-use effects.
Trent, Severn and Thames modelling study:
Naden et al (1996)93 developed a semi-distributed daily rainfall-runoff model
incorporating a soil water balance model, a runoff model, and a flood routing model
to estimate the impacts of both climatic and land-use change on flood response in
the three large (approximately 10,000km2) English basins: the Trent, Severn and
Thames. Projections suggested that completely covering the basins in coniferous
forest would reduce the 50-year flood peaks by an average of 20%, whilst total
conversion to deciduous forest was predicted to reduce peaks by an average of
14%. The authors acknowledged that the although the model had accounted for the
vegetation cover effects via evapotranspiration and water balance of the soil no
account of possible shifts in timing nor the partitioning between fast and slow
components of the runoff were made.
River Calder (Renfrewshire) modelling study:
Price et al (2000)99 investigated the likely influence of a range of Scottish upland
vegetation covers to flood flows in the relatively small (33 km2) River Calder
catchment in Renfrewshire. Vegetation covers considered were: natural coniferous
woodland, broadleaf woodland, conifer plantation, heather moorland, upland and
lowland grass. The Institute of Hydrologys HYLUC model (Calder, 1990)94,
calibrated on Scottish field data (Hall, 198795; Hall & Harding, 199396; Wright &
Harding97, 1993; Haria & Price, 200098), was used to predict the average seasonal
soil moisture status under the varying vegetation covers. An example of the
significant differences that can occur is given in Figure 8. This shows the monitored
difference in soil moisture deficits at adjacent mature Scots Pine woodland and wet
heathland sites at in the Cairngorms, together with the simulated deficits using the
HYLUC model. The simulations for the River Calder catchment were averaged over
an 11-year period (Figure 9) and used to derive the average Flood Studies Report
(FSR) catchment wetness index for different seasons. This was used together with
design rainfall (minus vegetation dependent interception losses) to run the FSR
rainfall-runoff model to test the resulting peak flow sensitivity to varying land covers
during typical winter and the summer periods (Figure 10).
-41-
(a)
Figure 8: Monitored soil moisture at two adjacent sites in the Allt AMharcaidh
catchment in the Cairngorms: a) mature natural Scots Pine woodland, b) wet heath
moorland. (Dots = averaged field measurement, Unbroken line = HYLUC model
simulation)
Figure 8 and Figure 9 indicate that summer season soil moisture deficits below forest
and particularly coniferous trees tend to be greater than under short stature
vegetation such as grasses or heather. Also the replenishment of the deficit during
the late autumn/early winter occurs later by several weeks under the tree cover than
the other vegetation types. Both these aspects are predicted to exert positive benefit
in terms of flood runoff. However during the winter period (December to February)
the deficits have been replenished under all of the vegetation types and by this
measure the catchment flood response is likely to be similar regardless of vegetation
type. This is reflected in the seasonal flood peak reductions which show a likely
greater effect during the summer than the winter (Figure 10).
-42-
20
0
31
61
91
121
151
181
211
241
271
301
331
361
-20
-40
-60
-80
-100
-120
Conifer
Heather
-140
Upland grass
-160
Figure 9: Simulated average seasonal soil moisture deficit for upland grass, heather
and coniferous woodland for the River Calder catchment, Renfrewshire. (Period
simulated: 1983 1993)
a)
b)
Figure 10: Predicted attenuation of: a) winter, b) summer flood peaks due to a
hypothetical 100% conversion of the Calder catchment (Renfrewshire) from upland
grass to various types of woodland\forest or lowland grass (Price et al, 2000). Given as
-43-
a percentage of the peak flows predicted for upland grass. (Prediction uncertainty
band reflects the range of values attributed to the model parameters coming from
different field experiments).
The study by Price et al (2000)99 only attempts to account for likely antecedent
conditions and interception losses but that the effects related to changes to: i) the
soil characteristics (permeability/infiltration rates) and ii) snow accumulation and melt
processes have not been considered due to a lack of sufficient understanding, and
these may also exert an appreciable influence.
The DEFRA FD211483 investigated the modelling approach and concluded the
following in general:
there are serious shortcomings in the rainfall-runoff models and methods
available for the use in the operational assessment of the impacts of land use
change and land management practices. There are three fundamental unresolved
issues: there is no generally-accepted theoretical basis for the design of a model
suitable to predict impacts, it is not known which data have the most value when
predicting impacts, and there are limitations in the methods available for estimating
the uncertainty in predictions. Some general recommendations can, however, be
made for a way forward in rainfall-runoff modelling for predicting impacts. The
modelling should be distributed and be capable of running continuous simulations. It
should also be partially or wholly physically based so that the physical properties of
local landscapes, soils and vegetation can be represented, and it should include
detailed modelling of surface water flow so that the effects of changes can be
tracked downstream. A considerable amount of high-quality field data on impacts will
be needed to support the development of robust methods for predicting impacts.
Since the publication of the DEFRA FD2114 study several detailed modelling studies
that begin to tackle the issue in line with the above modelling recommendations have
been published.
Frances et al (2008)100 report the findings of modelling studies in three catchments
located in Spain (Poyo, 380 km2), Austria (Kamp, 1500 km) and Germany (Iller, 954
km2), with three different climates (semi-arid, sub-alpine and alpine). The models
(different in each case) were all spatially distributed process based models coupled
to hydraulic models of the rivers.
Poyo modelling study:
For the Spanish site a 55% afforestation by pine was predicted to lead to 15% - 38%
peak flow reductions; with the lower reductions more representative of the events
starting with wetter initial conditions and the higher reductions for the drier initial
condition events. Much predicted variability was thought to be due to the
dependence on the combination of the temporal and spatial characteristics of the
storms and the spatial physical catchment characteristics. Peak flows of
approximately the 1 in 200 annual probability (200-year) event were less influenced
by the afforestation with reductions between 3 14%. The largest events (>> 1 in 200
annual probability) had the lowest predicted reductions 0 - 10%. No predicted
change in the timing of the flood peak was reported.
Kamp modelling study:
For the Austrian study pine forest cover was increased from the existing 47% to 86%
of the catchment. A range of reductions to peak flows were reported varying from
very small to 30%. The smaller reductions were predicted for the largest of floods (>
50 year return period), whilst the larger reductions were for the smaller events (<5
year return period). Deforestation of the existing forest from 47% to 0% of the
catchment was also modelled. Very small to 75% peak flow increases were predicted
with again the smallest increases for the largest of floods and the largest increases
-44-
for the small flood events. The asymmetry of afforestation and deforestation
responses was related to where within the catchment the changes are manifest (i.e.
flat deeper soils or shallow steep soils). The modelled changes to peak flows as a
function of both the magnitude of the event and the initial wetness of the catchment
are given in Figure 11. No change in the timing of the peaks was discernable.
Figure 11: Modelled potential reduction of flood peaks on the Kamp following
afforestation (increase in pine forest from 47% to 86%), plus the predicted increase of
flood peaks following deforestation of the existing forest (from 47% to 0% forest
cover). [Source: Frances et al (2008)]
-45-
Figure 12: Modelled potential reduction of flood peaks on the Iller following a 29%
increase in pine forest cover. [Source: Frances et al (2008)] Note the 25 50 year event
is estimated to be approximately 800 cumecs.
The studies at each of the three sites suggest that antecedent soil moisture
conditions are important in determining the effectiveness of the forest to reduce
downstream flood peaks. Each study also suggests that the relative size of the
reduction is a function of the magnitude of the event with very large floods little
affected by the presence of forest cover. The authors link this to the concept of a
retention volume provided by the forest. Their view is that although the retention
volume offered by afforestation is likely to be small, it may be useful to combine with
other flood management techniques to bring about a more substantial reduction in
downstream flows.
The authors of the report indicate that appreciable uncertainty exists in the
description and parameterisation of afforestation hydrological processes and that it is
necessary to view the presented predictions in this context.
In the UK several modelling studies have also been undertaken or are on going:
Parrett modelling study:
Park et al (2006)101 report the findings of an initial investigation in to the likely flood
management potential of afforestation in the Parrett catchment (south west England).
The investigation used a high resolution, fully distributed model based on the MIKE
SHE model which encompasses both the hydrological and hydraulic routing models.
Its operation requires a large number of parameters to be used.
The scenarios modelled were: i) arable land converted to coniferous forest (61% of
total area changed), and ii) grassland to coniferous forest (22% of total area
changed). Only autumn and winter rainfall events were investigated. Overall, there
was no clear evidence to indicate that the introduction of woodland either reduced or
increased flooding in the Parrett catchment to a significant extent, but that these
predictions should be viewed only as initial provisional findings.
Pont Bren modelling study:
A detailed multi-scale investigation has, and continues to be, undertaken at
Pontbren, mid Wales (Wheater et al, 2010)103. The study has investigated both the
effects of soil degradation due to compaction by livestock and the effects of tree
cover in the form of both shelter belts and complete catchment cover on flood runoff.
-46-
The experiments focus on soil properties and runoff processes, based on plot-scale
and hillslope-scale measurements nested within gauged first and second-order
catchments (Marshall et al, 2006)102. Plot studies were undertaken on the varying
land uses from which it was identified that infiltration rates and pore space in the A
horizon soil of the woodland were greater than that from the same soil for grazed
grassland. In addition these properties, at this particular site, had a seasonal
dependence probably related to seasonal wetting and drying and increases in
biological behaviour.
The modelling strategy was to: i) first establish a detailed, physically based local
scale model, ii) use this to train a simpler conceptual model, iii) develop a catchment
scale semi-distributed model such that the effects can be simulated downstream
through the river network. Wheater et al (2010)103 present example output from the
above approach for a 4km2 Pontbren sub-catchment for the following scenarios
compared to the baseline condition of the current-day land use:
a)
b)
c)
The simulated changes in flood peak for these scenarios are: removal of all trees
causes a 20% increase in flood peaks; adding tree shelter belts to all grazed
grassland sites causes a 20% decrease in flood peaks from baseline condition; and
full afforestation causes a 60% decrease in flood peaks from the baseline condition.
The floods simulated are described as relatively frequent events and are
interpreted here as the size of event that is likely to be experienced several times a
year. In Wheater et al (2008)104, the authors also report a speculative simulation for
Pontbren using the extreme rainfall that generated the Carlisle flood of January 2005
(2 day rainfall = 140mm with a rainfall return period of 180 years). The simulated
changes in this extreme case were predicted to be: removal of all trees causes a 37% increase in flood peaks; adding tree shelter belts to all grazed grassland sites
causes a 2-11% decrease in flood peaks from baseline condition; and full
afforestation causes a 10-54% decrease in flood peaks from the baseline condition.
It was also noted that the full afforestation scenario increased the time-to-peak by 30
minutes (though it is unclear whether this may, in part, be attributable to an assumed
deterioration of the field drain system).
The authors emphasise that quite large uncertainty exists for the predictions due to
parameter uncertainty. Additionally attention is drawn in general to the site-specific
nature of these findings. The authors suggested that much work may remain to
provide a more general methodology for national application.
There appear to be an array of local variables (soil, climate, event magnitude, forest
species, forest age, topography, forest extent, catchment shape, other changes in the
catchment etc) which mask the easy development of a simple rule to accurately predict
how the change in forest cover can affect flood flows. While there is consensus in the
literature that the forestation of small catchments can reduce flood flows, particularly for
smaller flood events, based on the available science it is not felt appropriate to directly
relate changes in percentage forest area to a percentage reduction in flow as to do so
would overly simplify what is an immensely complex process. A potential means of
achieving this may be to adopt a more physically based hydrological modelling approach
to represent the role trees play within existing hydrological models. The Leaf Area Index
-47-
There may be potential to use a 1km global data set published on an 8 day basis by
MODIS108 to setup of database of LAI values for Scottish land cover and there may be
scope to make account of cover maturity and season. Alternatively, the methodology
adopted by MORECS (Hough & Jones, 1997)216 for estimating LAI variation with land
cover and the effect of LAI could be applied.
As previously discussed, forests can also serve to reduce antecedent moisture
conditions. A range of potential methods of quantifying this effect are discussed in
Section 4.4.2(D).
(F)
This section discusses the literature relating to upland drain blocking in headwater areas
of catchment which are typically moorland or peat bogs. For details of the literature
relating to agricultural field drains refer to Section 3.3(D).
The majority of catchment monitoring studies took place prior to 1990 and may consider
the impact of now outdated drainage practices. There are a number of monitoring studies
which have investigated the impact of introducing drainage while qualitative monitoring
information is sparse when it comes to blockage information. Adopting the simplistic
approach of assuming no long term damage has taken place to the soil and that simply
blocking the drainage channel will return the soil to an identical condition to its predrained state may be overly optimistic.
A literature review undertaken during the work on peatland hydrology by Armstrong et al
(2006)123 noted that grips are typically 50cm deep and 50-70cm wide, and have been
installed in a large proportion of UK peatlands. Their work was on upland blanket peats,
and they note that drainage of lowland raised bogs has included drains both significantly
smaller and larger. Drainage has been shown to influence both the properties of the peat
and the runoff characteristics of outflowing streams. Ramchunder et al (2009)110 noted
impacts on peat shrinkage and consolidation, microbial activity and decomposition, all
influence hydraulic conductivity and water storage capacity as well as flow rates and
processes and susceptibility of the peat to erosion.
Price et al (2003)109 identified that the efficiency of drainage depends on the depth of the
ditch, the ditch spacing and the hydraulic conductivity of the peat. Quoting several studies
they proposed that water tables might be drawn down up to 50m from the ditch in fibrous
peat, but hardly affected at all in decomposed fen peat. On gentle slopes, closer drain
spacing increases the likelihood that water tables will be depressed over large areas of
the landscape. In addition to drawing down the groundwater by their effect on subsurface
flows, ditches can have a significant effect on peatland by interrupting water which would
naturally flow over or near the surface of the peat. Ramchunder et al. (2009)110 noted that
-48-
many blanket peatland drains were excavated so that they run across the slope, typically
following contours and so interrupting natural flow routes.
In other cases, however, grips are often channelled directly into an adjacent watercourse
and in some cases herringbone patterns either side of watercourses have been created.
Ditches running up and down slope are typically thought to produce rapid flow velocities
and have a very different effect on water tables from ditches following the contours, which
have been shown to produce the asymmetric patterns of water table upslope and
downslope of the drain. Holden et al (2004)121 also noted the importance of the position of
grips in the catchment, because ditches serve to lower the watertable (and so may
increase storage and reduce flow peaks in the stream) but they also increase the velocity
of flow once it reaches the drain, and may thus increase the peak flow by speeding water
from one point to another.
As reported in Labadz et al (2010)111 although large amounts of water are held within the
peat of a blanket bog, it is important to understand that little of the water received as
rainfall is retained. Peat has extremely high water content and, in an intact peatland, most
of the water storage capacity is already full. Even intact blanket peat is highly productive
of rapid (rainfall event) runoff and, by contrast, generates little long-term baseflow during
dry periods. For example, in the Feshie catchment in the Cairngorms, Soulsby et al
(2006)112 showed streams draining blanket bog are often ephemeral in their flow regime,
ceasing to flow in driest periods. Many bogs do not act to delay flow into streams (Rydin
and Jeglum, 2006)113. The oftenquoted idea of a peatland as a sponge that soaks up
rainfall and then releases it slowly into rivers is therefore viewed as erroneous.
Ramchunder et al (2009)11041 noted that although drain blocking prevents efficient delivery
of water through the artificial network, and alters hydrological routing to give noncontinuous flow, there is little evidence as yet of larger scale impacts. Grayson et al
(2010)114 have also noted that although carbon storage and flood mitigation are
increasingly used to justify the expenditure on peatland restoration, there is a lack of
reliable evidence of impacts on the flood peak downstream of grip blocking and
revegetation of bare/eroded peat.
Ballard et al (2010)124 applied a physically-based modelling approach to predicting the
effect of drain blocking on peak flows and found that it may depend on site specific
conditions, sometimes increasing and sometimes decreasing peak flows. The study
suggested that the greatest benefits for flood management would be achieved by
blocking drains on steep slopes that are poorly vegetated, but acknowledged that there
are many uncertainties and suggested that better characterisation of overland flow and
drain roughness is required. Subsequent study using the same small scale physically
based hydrological model identified that the drainage of peatlands will increase peak
flows but that drainage blockage will not necessarily reduce flood peaks (Ballard et al,
2011)115. The research suggested that drainage blockage can be prioritised to steeper
and smoother channels but it concludes that field studies are needed to provide data for
future modelling.
A concise summary of the literature relating to upland drain blocking catchment
monitoring and modelling studies has been provided by Jacobs (2011)3 as presented
within this breakout box:
a) Catchment monitoring studies
Perhaps the earliest study in Britain was by Lewis (1957)116 of the Alwen and Brenig
catchments in north Wales. Although principally concerned with the water yield from
upland catchments he noted that forestry drainage operations resulted in a more
flashy runoff response, but gave no further details.
Conway and Millar (1960)14 compared the flows from adjacent upland drained and
-49-
undrained small upland peat catchments in the northern Pennines and found that a
higher flood response came from the drained catchments. Their results and further
data subsequently collected were examined by Robinson (1985)117 which confirmed
their main conclusions regarding flood flows.
The following three experiment sites\catchments offer a more detailed and quantified
understanding of the effects of upland drainage:
Blacklaw Moss (Lanark, Scotland):
The 7 ha experimental site was instrumented for 5 years from 1959-1964 by the
Macaulay Institute and the Hill Farming Research Organisation (Robertson et al,
1968)118. The site is described as a raised basin mire used for rough grazing with
deep peat (average depth 5.8 m) soils that are almost permanently waterlogged. The
vegetation was dominated by heather (Calluna vulgaris) and cotton grass
(Eriophorum species), with a discontinuous carpet of mosses including Sphagnum
species. After a 3-year calibration period the land was drained by cutting open
ditches about 40cm wide and 36cm deep at 9 metre spacings.
There was little difference between the two periods in the characteristics of the
storms, but there was a large increase in the observed peaks (Robinson, 1990)56. In
part this was due to an increase in the percentage of rainfall that formed runoff, but
mainly due to an increase in the flashiness of the site. These findings are numerically
reported in terms of how the standard unit response of the site changed using the
parameters of the Flood Studies Report136 Unit Hydrograph rainfall-runoff model. The
time-to-peak decreased to just 0.33 of it pre-drainage value; the percentage runoff
increased from 46% to 58%, and the peak of the unit hydrograph increased by a
factor of 2.6.
The increased in responsiveness of the site was interpreted as the result of the large
increase in the channel network speeding up flows by shortening the slower flow
paths through the soil to the channels, and (bearing in mind the high water retention
of the peat, together with capillary rise above the water table, the soil remained wet
for long periods after drainage) there was very little compensating increase in the
available storage capacity of the soil (giving limited additional runoff attenuation).
Llanbrynmair (central Wales):
An instrumented catchment study investigating the effects of forestry on hydrology
and water quality (Leeks and Roberts, 1987)119 comprised both a 3 km2 and a nested
0.34 km2 peat moorland catchment that were planned to be afforested. The
catchment was progressively drained over a 4-year period until 70% of the area was
affected (the residual area was either too steep or not to be planted). Robinson
(1990)56 undertook similar analysis on the smaller of the two catchments as
undertaken for Blacklaw Moss. The comparison of unit hydrographs before and after
the drainage indicated similar hydrological effects to those at Blacklaw, namely open
drainage resulted in a much peakier storm flow response.
Drainage of the main catchment started on the higher ground most distant from the
outlet gauge, and then moved to the valley bottom near to the gauge. Drainage of the
higher land (affecting 45% of the catchment area) gave rise to a much peakier runoff
response recorded at the outlet gauge, but when the valley bottom was subsequently
drained (15% of catchment) there was no further increase in peaks, although the
response time of the catchment was shortened. This was interpreted as the speeding
up of flow from the areas near the gauge, building up the early part of the runoff
hydrograph; any increase in the peak response was offset due to this water leaving
the catchment before the arrival of flows from the more distant parts of the
catchment. Subsequent drainage of a further 10% of the area near the centre of the
catchment produces no further increase in peaks. In fact the maximum flows slightly
-50-
reduced and this was thought potentially connected to the earlier drains becoming
hydraulically less efficient.
The effect of location of drainage was also reported by Acreman (1985)120 for the
extensive pre-planting upland drainage that occurred in the Ettrick catchment in
southern Scotland. Initially most of the ploughing occurred near to the catchment
outlet and then later in the headwaters. The result was a small decrease in peak
flows when the lower areas were ploughed, but as the drainage extended into the
headwaters the peak flows increased. As for Llanbrynmair this increase reduced
within a few years as the drainage was speculated to have become less effective.
Coalburn (northern England)
Robinson et al (1998) report the hydrological findings of the Coalburn study from its
inception in 1967 to 1996.The catchment comprises a 1.5 km2 upland peat
catchment (altitude range 270 330 m AOD) that has a mean annual rainfall of about
1350mm. 75% of the catchment is covered by peat bog, and the remaining 25% by
peaty gley soils. The peat depth is generally within the range of 0.5 3 m, though the
maximum recorded depth is 10 m. Prior land use was for rough grazing (vegetation
comprised: Molinia grassland and bog species including Eriophorum, Sphagnum,
Juncus and Plantago). Some old relic grip drainage existed from the 1940s and
1950s, but these were considered to have become largely inoperative. Hydrological
data was collected for 5 years before the whole catchment was subject to the
ploughing of open drains in 1972. Spacing of drains was about 5 m apart and they
were aligned with the ground slope. The long continuous channels were 80 90 cm
deep and over 100cm wide. Water from these drains was either intercepted by
deeper drains or allowed to connect directly to the natural water course. This is an
example of the old style environmentally insensitive upland drainage and can be
considered to be a relatively aggressive form from that era.
As for Blacklaw Moss the characteristic response of the catchment was investigating
using the parameters of the Flood Studies Report Unit Hydrograph rainfall-runoff
model. The key parameters for the model were established before drainage and at
subsequent periods after. In the 5-year period after the drainage the time-to-peak
parameter reduced on average by 22%. The effect diminished over the following 20
years (Figure 13), though the apparent effectiveness of the drainage may be affected
by the establishment of the forest cover. It is therefore not possible to reliably
assume that the effectiveness of the drainage would have been the same had the
trees not been planted. The authors suggest the increase in catchment flashiness to
be a result of a greater density of drainage channels which speeds up the removal of
surface waters. The diminution of the effect over time is suggested to be a result of a
reduction in the hydraulic efficiency of the drains due to the furrows becoming
colonised by vegetation and later by litter from needle fall.
-51-
100%
-25%
80%
-20%
60%
-15%
40%
-10%
-5%
20%
0%
0%
1970
1975
1980
1985
1990
-30%
1995
Figure 13: The changes to the speed of catchment response parameter (TP) and the
percentage of storm event rainfall forming quick runoff (PR) observed at the Coalburn
experimental peat catchment following intensive upland drainage. (Values calculated
from data presented in Robinson et al, 1998. Speculated canopy cover estimated by
Jacobs based on experience and the limited information provided by Robinson et al.)
The reduction in the proportion of rainfall forming fast flow is small over the first ten
years and is not statistically significant. Later estimates maybe suggest a reduction,
but again this may be related to the establishing tree cover.
Based only on the change in flashiness (TP) design flood flows (using the FSR Unit
Hydrograph method) were predicted to have been increased by 15 20% in the first
5 years after the ploughing, after which the effect reduced. Importantly the authors
also note:
This is not, however, to say that the increases indicated here would be the same for
extreme events: the unit hydrograph analysis was based on events that were
generally smaller than the mean annual peak. In extreme (and rare) events, system
threshold may be surpassed. For instance, if a storm exceeded the interception
capacity of the vegetation and the available depression storage of the many partly
infilled and blocked drains, then the extensive artificial channel network could
potentially produce much higher peak outflows than the original moorland That is the
direct consequence of the much greater velocities of open channel flow than
subsurface or overland flow.
Holden et al (2004)121 present an overview of studies that have examined the
hydrological response to artificial drainage in peat. The majority of the studies noted
an increase in annual flood peaks following drainage, although some showed a
decrease. They deduced that flood responses decrease in cases where the peat,
topography and ditch characteristics cause water tables to fall. This may affect both
the saturation excess overland flow and the occurrence of soil pipe/macropore flow.
-52-
(However they also note that a fall in watertable may lead to peat decomposition and
subsidence, so that the temporary increase in storage is lost and runoff becomes
more flashy again). In most peats, however, they indicate that the soils have low
hydraulic conductivity, so in most cases the effect on watertable is limited to a very
localised zone along the ditches. They therefore see that the primary effect of ditches
is to speed up the rate that water leaves the catchment, which increases flood peaks.
They also recognise that changes in the timing of runoff may alter downstream flood
peaks in different ways, depending on the phasing with flow contributions from other
sub-catchments.
Holden et al (2006)122 review the long-term hydrological impacts of artificial drainage
of peat at two north Pennine catchments drained in the 1950s compared to two
control catchments at Moor House. These include the catchments considered by
Conway and Millar (1960)14. Their findings were:
River Ashop catchment in Derbyshire grip blocking trial for Severn Trent
Water, the National Trust and Natural England involving monitoring before
and after blockage of eroded gully channels (OBrien et al, 2004)
Sapling Clough in the River Hodder headwaters in north west England - grip
blocking study as part of United Utilities Sustainable Catchment Management
Programme (SCaMP).
The focus of these studies is primarily the impact on colouration (of particular interest
to the water companies) but the studies also have the potential to provide valuable
insight into the impact upon the hydrological behaviour during flood events. Findings
from these monitoring studies have not become apparent to this review, but SEPA
are recommended to seek the findings in the future.
b) Modelling studies
Modelling studies have largely been focused on trying to predict how effective
peatland drain blocking will be, as opposed to the catchment studies that have
focused on the impact of installing drainage.
The SCIMAP (Sensitive Catchment Integrated Modelling and Analysis Platform)
study (Lane et al, 2003)28 investigated drain blocking using a model that linked a
-53-
hydrological model to a detailed digital elevation model. From this they concluded
that grips affect flood runoff more via the increased speed of water shed from the
catchment than by the extra soil moisture storage that results from the drainage. In
terms of grip blocking they concluded that a catchment scale model that represents
the spatial arrangement of drains and their connectivity to the drainage network is
needed to a) determine the catchment scale impact of drainage or drainage blocking
on downstream runoff, and b) to identify which channels would be most effective to
block.
Ballard et al (2010)124 used a simplified physics based model to simulate the flood
response of a 200m x 200m plot of upland peat. The study assumed that no channel
structural changes and no vegetation changes occurred after drain blocking and that
no bypass flow around drain dams occurred. Drain depth was fixed at 0.6m. Storm
events from 2006 were used to drive the model using a 15-minute time step for i) no
drain blocking, and ii) drain blocking over a range of: drain spacings, surface slopes,
drain angles, and hydraulic roughness values representing different vegetation
covers and channel surfaces. The simulations suggested that flood peak response
was sensitive to the characteristics of the storm and the antecedent conditions.
Sensitivity analysis suggested that one average drain blocking leads to the greatest
reduction in flooding for sites with larger drain spacing, steeper drain angle, steeper
slope, rougher plant cover, smoother drains and a thin acrotelm (surface layer).
However there was much variability in the outcome with increases in peak flows also
predicted as well as decreases depending on the event and parameter set used. The
authors conclude that their work suggests that drain blocking may not necessarily
always reduce flow peaks, with some cases showing negligible changes in runoff and
other cases actually indicating an increase in flow peaks. But that the results from
their preliminary study could be used to prioritise works for drain blocking. Drains that
are steeper and smoother are most likely to show the greatest reduction in flood
peaks following blocking.
Geris et al. (2010)125 describe preliminary work involving numerical modelling of grip
blocking at Sapling Clough (1.7km2), a headwater of the River Hodder catchment in
north west England. However the model is in the stages of early development so no
direction regarding the effectiveness of blocking was presented.
JBA (2007)126 report upon a modelling study that considered the potential benefits of
upland drain blocking upstream of Ripon in North Yorkshire. The catchment down to
Ripon has an area of 120km2 (comprising the rivers Laver, Skell and Kex Beck), an
altitude range of 30 410m, average annual rainfall of 900mm with the headwaters
receiving about 35% more rain than the easterly lowlands). The soils in the higher
rainfall, upland headwaters (approximately 30% of the overall catchment area) are
dominated by various types of deep and surface peats overlying mineral soils. The
land cover in these upland areas comprises moorland grass and heather. During the
1970s this area was subject to partial upland grip drainage (67km in length; equating
to about 22% of the overall catchment to Ripon). In the lower portions of the
catchment the soils are a mixture of free draining and relatively impermeable mineral
soils that are used for pasture, arable crops and some forest\woodland cover. In the
model the overall catchment was split into 10 sub-catchments. The hydrological
response from each was modelled using the Probability Distributed Moisture (PDM)
rainfall-runoff model, and linked to an ISIS hydraulic routing model to simulate the
passage of the river flows through the catchment down to Ripon.
The effect of drain blocking in the headwaters was simulated by increasing the
various timing parameters in the PDM hydrological model by 1 hour which has the
effect of reducing the speed of sub-catchment response. No evidence was given to
justify this size of parameter amendment. Its selection was indicated to be reliant on
-54-
(G)
Constructed farm wetlands can be used to treat lightly contaminated surface water runoff
from farms and they are a means of managing diffuse pollution (Scottish Government,
2009)128. The Constructed Farm Wetland Design Manual for Scotland and Northern
Ireland (Carty et al., 2008)129 provides a best practice manual for the construction of farm
wetlands; however the manual focuses on designing wetlands to manage diffuse pollution
and only mentions in passing that the constructed ponds may also provide flood
attenuation benefits. There is no literature specific to the flood attenuation benefits of
constructed farm wetlands which have been designed for the treatment of diffuse
pollution. It is speculated that the wetlands perform in a similar manner to wetlands
which are used in SUDS. The SUDS Manual (CIRIA, 2007)130 identifies that it is
-55-
necessary to provide additional storage volume over the permanent retained volume and
an outlet control which will ensure that this attenuation volume is utilised effectively during
flood events. The design of flood attenuation features is a complex undertaking and it is
considered unlikely that farm wetlands will provide any significant attenuation due to both
the requirement to provide additional storage over the permanent treatment volume
advocated within current best practice and the need to provide a carefully tuned flow
control at the outlet to ensure the volume is utilised at the flood peak, not before or after
it.
The use of an array of offline and online flood storage areas, which are referred to as
Runoff Attenuation Features (RAFs), is presented for the town of Belford by Wilkinson et
al (2010)131. The paper reports that a recent flood alleviation scheme feasibility study
identified that it was not viable to protect the 31 properties in the town which are
estimated to be at risk of flooding during the 1 in 200yr event. However it is noted that
the Environment Agency has recently undertaken minor works to improve the situation for
a handful of worst affected properties affected in the town by increasing the channel
capacity and constructing a short flood wall (Environment Agency, 2011)132. Wilkinson et
al (2010) presents the use of four RAFs providing a total of 2,800cu.m of storage within
the 6sqkm catchment, these are online and offline flood storage areas which are intended
to primarily attenuate floods rather than address diffuse pollution problems. By a
simplified comparison of the hydrograph volume and the available storage it is proposed
that the current flood storage reduces the 1 in 5yr flood peak by approximately 8%,
however this assessment takes no account of any inherent inefficiencies in utilising the
storage (Ackers and Barlet, 2010)133. The paper presents the protection of the town from
flooding during a 1 in 48yr rainfall event in 2008 as evidence that the measures are
effective, however the question of whether the town would have experienced flooding
prior to the installation of the RAFs for this 36 hour 96mm event is not answered
conclusively. It is noted, that in the paper the duration of flooding is approximately two
hours, therefore the 1 in 48yr, 36 hour rainfall event is not representative of the flow
return period. Elsewhere the paper estimates the attenuated 2008 peak flow as 2.1+/0.4cumec, putting the event as substantially less than the 4.56cumec (1 in 5yr) threshold
of flooding once the optimistic 0.4cumec reduction of flood is accounted for. Finally it is
reported that during the 2008 event all the flood storage cells were full around the time of
the flood peak, implying that during a larger event the storage would be utilised ahead of
the flood peak and thus not attenuating the hydrograph peak.
It is accepted that small scale flood storage areas, such as those in use in Belford, can be
an effective flood risk management method, however there remain a number of issues
which need to be addressed:
For the RAFs to be effective they all need to be designed to throttle the flow
effectively so that the flood storage they offer is efficiently used to reduce the peak
of larger floods and not the leading part of the hydrograph (Hall et al., 1993)134.
The large number of sites, often located on private land in remote parts of the
catchment, introduces maintenance challenges. It is unclear if RAFs can be relied
upon to function in the desired way without frequent routine maintenance,
especially considering the relatively short design life of the structures, the
blockage risk to small openings used to throttle flows and the multi-year periods
between the sites being required to perform their function.
The large number of sites leads to a large number of in-channel structures, each
forming a potential barrier to wildlife movement and a morphological pressure on
the watercourse.
As with all flood storage, care must be taken to ensure that flood peaks on flashier
sub-catchments are not delayed so that they coincide with other sub-catchments
-56-
although the same effect can be applied to desynchronise peaks (Hall et al.,
1993)134.
(H)
Two thirds of studies have concluded that wetlands increase average annual
evaporation or reduce average annual river flow. It should be noted that a
minority conclude the opposite.
Two thirds of studies conclude that wetlands reduce dry weather flows.
Many bogs do not act to delay flow into streams (Rydin and Jeglum, 2006)113. The Winter
Rainfall Acceptance Potential (WRAP) was developed as part of the Flood Studies
Report (NERC, 1975)136 uses five WRAP classes to inform the spatial percentage runoff.
Soil and land cover types normally indicative of wetlands are typically in WRAP class 5,
this class has the highest percentage runoff within the Flood Studies Reports RainfallRunoff Method (Boorman et al., 1995)137. The underlying messages are that each
wetland is unique and that many wetlands are saturated at the start of flood events
therefore any incident rainfall is rapidly converted to runoff.
The SEPA Wetlands Inventory (SEPA, 2012) presents a useful spatial database of all
known wetland habitats in Scotland. It is based on the accumulation of a range of data
sets and wetlands are subdivided into classes based on wetland type. This data set
represents a useful resource for a national overview of wetlands in addition to more site
specific information.
(I)
There is limited published literature on the specific benefits of gully planting within a
Scottish flood risk management context. There are a few papers on the subject of the
formation of gullies within loess soils (alluvial silts) in other parts of the world. These
gullies are notably smaller than the gullies or cleughs that are typically found in
Scotland, with depths, widths and lengths on the order of a few meters rather than tens of
meters often encounter in places such as the Scottish Borders.
It has been reported in a number of locations across Scotland that high rates of
deposition may be a major contributor to flooding due to a loss of channel capacity as
observed at Selkirk (Long Philip Burn), Hawick (River Teviot) and Marykirk (Balmaleedy
Burn).
-57-
Poesen et al. (2003)138 reports that gully erosion represents an important sediment
source in a range of environments and gullies are effective links for transferring runoff
and sediment from the uplands to valley bottoms. Poesen et al identified that most
research effort in recent decades has concentrated on sheet (interrill) and rill erosion
processes operating at the plot scale with few studies investigating gully erosion.
A study of the Upper Wharfedale by Lane (2006)139 identified that the forested
headwaters of the catchment produced comparatively little coarse gravels while tributary
gills (terminology used for gully in some areas of England) were a major source of the
problematic gravels which were causing the channel aggradation and subsequent
flooding further downstream. It was identified that the gills were cutting down into old
glacial deposits. Lane comments that in this case it was most appropriate to address the
sediments at source rather than seeking to convey the sediment through the problematic
reach (which was in an essentially natural state). Through modelling he proposed that
very localised planting of gully slopes could reduce the delivery of coarse sediments by
85%.
A historic mapping/aerial photography/digital terrain model study of a 25sqkm catchment
in Spain identified a sediment production rate in gullies of 1322+/-142 ton/Ha/yr
140
(Martn
ez-Casasnovas, 2003) . While it would be unwise to adopt anything more than a
sense of possible magnitude from this data it does give an indication of the scale of the
sediment source gullies represent. A catchment study in Yangjiagou identified that
manmade forests have reduced sediment yields by as much as 92.5% (Zhuo, 1992)141.
Raven et al (2010)142 provides a detailed overview of the issues associated with the
management of gravels in upland rivers. The paper concludes that there is a need for a
shift in how rivers are managed based around catchment scale approaches. The
management of sediment sources on the hillslopes may provide long term mitigation, but
it will be necessary to manage sediments which have already entered river systems. In
many cases the use of bank protection prevents the process of safely deposing coarse
sediments at the inside of bends in exchange for the mobilisation of fine sediments. The
uncertainties imply that the best long term management strategy may be for people to
learn live with bank erosion and occasional flooding.
In preparing this section it is noted that:
The steep topography of a gully is likely to make any intensive land use difficult if
not impossible
The steep sides of a gully would infer very rapid hydraulic connectivity for any
runoff that is generated within a gully.
-58-
(J)
Floodplain reconnection
This section refers to the literature relating to the physical re-introduction of flow to the
floodplain during flood events. Refer to item (L) for literature relating to using vegetation
to slow out of bank flow.
A modelling study of the Cherwell catchment was undertaken to assess the hydrological
impact of restoring the connection of floodplains within the catchment (Acreman et al.,
2002)143. The study employed a 1D hydraulic model and a semi-distributed hydrological
model to assess the impact of restoring entrenched river reaches to a natural state. The
natural conditions were based on the findings of a geomorphological study. The study
found that the complexity of floodplain hydraulics presented as considerable challenge to
the modelling work. The results showed that where embankments separated the river
from the floodplain the peak flow rate could be increased by 50 to 150% (represented by
adding embankments to reaches presently without embankments) however it reports that
restoring the channel through the floodplain to its pre-engineered dimensions could
reduce flood flows by 10 to 15%. Within the body of the report it is presented that some
floodplain reconnections have a negligible impact, hence it may be more representative
to report a 0% to 15% reduction in flows where the watercourse is entrenched and for
cases where floodplains are completely disconnected the benefits may be greater. The
report does discuss the variation on time of peak for some of the test reaches, in one
case it is reported that the peak could be brought forward by 40 hours should
embankments be introduced and delayed by 17 hours if the channel was restored to a
natural condition. Unfortunately the study does not provide a time to peak or provide an
estimation of the change in flood wave velocity. It should be noted that the 1D modelling
approach did not allow for the conveyance of water on the floodplain (extended crosssections within ISIS using a Mannings value of 0.0), while this approach is a commonly
adopted method for modelling floodplains it is likely to result in the overestimation of flood
attenuation. The increasing usage of 2D hydraulic modelling techniques which would
avoid the need for extended 1D river sections render this study somewhat outdated.
While this study adds to the case that floodplain reconnection can be used to attenuate
flood flows the quality of this study is insufficient to allow it to be applied elsewhere.
The benefit of floodplain restoration is discussed qualitatively by Blackwell et al (2006)144
and includes the presentation of a number of projects where floodplain reconnections
have been implemented across Europe. Unfortunately the report does not provide a
summary of the hydrological benefits. It does report that reduction in flood risk is
dependent on local conditions.
A modelling study on the River Suir in south-east Ireland (Ahilian, 2009)145 reported that
storm duration is a significant factor in the attenuation of the flood wave. To this end it
was identified that the ratio of the storage volume available on the floodplain to the
hydrograph volume is significant. The study also found that the floodplain width and
floodplain slope were also significant.
A comprehensive literature review of the hydrological role of wetlands was undertaken by
Bullock and Acreman (2003)146. The review collated 439 published statements on the
role of wetlands within the hydrological cycle. Part of the review focused on the function
of floodplain wetlands in reducing flood peaks. The review found that most (23 of 28)
show that floodplain wetlands reduce or delay floods. The study concluded that there is
no simple relationship between wetland types and the hydrological function they perform
and that apparently similar wetlands can be driven by different hydrological processes.
For this reason the papers authors recommend the development of an improved wetland
classification system which makes better account of function.
Jacobs (2011)3 proposes the application of a tool which functions by scaling the
generated floodplain area by model results generated for a donor catchment. The
suggested methodology includes an allowance for slope (1:250 to 1:1000), the location
-59-
within the catchment (upper, middle and lower) and an efficiency factor. The method
implies that a reduction in flood peak of approximately 0.0% to 4.6% per 10,000cu.m of
created storage is achievable. The method does not take account of the ratio of
floodplain volume to hydrograph volume which has been found by others to be of key
significance Ahilan (2009)145 and Potter (2006)147. It is accepted within the Jacobs (2011)
study that this is an attempt at simplifying a complex process which varies markedly
between locations. As this proposed methodology is heavily based on the findings of one
donor catchment and the judgement of its author it is difficult to recommend its
application.
Within the scientific literature there is evidence that floodplain reconnection can in some
circumstances increase the downstream flood risk (Bullock and Acreman, 2003)146. This
is because improving the connection to a floodplain could increase the conveyance
capacity offsetting the increase in storage in a similar fashion to the effect of removing
constrictive bridges and culverts which is well documented. Particular care and attention
should be made when removing flood embankments, especially where they form a
constriction at the downstream end of an existing functioning floodplain.
It should be noted that some means of reconnecting a floodplain (e.g. reducing channel
entrenchment) can result in an increase in water level. The backwater effect should be
investigated as discussed in Appendix C.
As presented by Jacobs (2011)3, it is difficult to draw general conclusions on the
magnitude of the benefits as these are invariably case specific and depend on the nature
of the enhancement (compared to the existing condition), the specific flood event and the
proximity to the flood receptor. Going a step further, it is probably impossible to make a
generalised statement about the effect which floodplain reconnection can have and that
the only means of making a meaningful assessment is to use a flood routing element
within a hydrological model. As presented in Ahilan (2009) the ratio of floodplain storage
generated to hydrograph volume is a useful indicator of the potential impact. In the
absence of sufficient data to develop a regression based analysis of the link between the
floodplain volume to hydrograph volume ratio in addition to other key variables the only
viable means of assessing the flow attenuation capabilities of floodplain reconnection is
to construct a hydraulic model.
(K)
The term riparian is relatively loose and there are a number of different definitions. The
UK Forestry Standard Water Guidelines (Forestry Commission, 2011)76 defines the
riparian zone as:
the area of land adjoining a river channel (including the river bank but not the wider
floodplain) capable of directly influencing the condition of the aquatic ecosystem (e.g. by
shading and leaf litter input).
Elsewhere, the riparian zone is defined by the Environment Agency (1998)148 as:
the land either side of the river channel, extending to the limits of associated floodplain
wetland, or 50m, whichever is greater.
In reviewing the literature it is necessary to take due consideration of the varying
definition of riparian.
The main functions of the riparian buffer are considered by Broadmeadow and Nisbet
(2004)149 to be:
-60-
From a flood risk management perspective riparian woodland can alter the following
processes:
Provide a supply of large woody debris therefore facilitating the natural formation
of large woody debris dams leading to the slow of flows; and
Provide a supply of organic material which can cause the blockage of downstream
structures. The elevated risk of blockage to structures is discussed in Appendix
C.
Nisbet et al. (2011)65 presents a review of the potential benefits of riparian woodland
referencing earlier work which demonstrated that woody debris dams can significantly
delay flood peak travel time and lengthen catchment response times. It is suggested that
to be effective at a larger catchment scale would require extended reaches of riparian
woodland and the subsequent formation of large woody debris dams along tributary
streams. It is also noted that questions remain over the ability of riparian woodland (and
floodplain woodland) to capture or release large woody debris. The risk of woody debris
blocking pinch points at downstream flood receptors and increasing flood risk at such
locations also needs to be considered.
The Robinwood study (Forest Research, 2008)150 into the effect of large woody debris in
watercourses indicated that large woody debris can slow in-channel flows and thus help
to attenuate floods. Modelling (1D hydraulic) undertaken as part of the project indentified
that during a 1% AEP event each large woody debris dam would delay the flood peak on
the River Y Fenni by an average of 2 to 3 minutes. It was found that the effect of large
woody debris on flow attenuation diminished with event magnitude as the features would
become drowned out during high flows. The report indicates that larger benefits can be
achieved when large woody debris dams improve the floodplain connection by raising inchannel levels and thus facilitating flood water to enter the floodplain earlier in the
hydrograph. It is suggested that large woody debris dams in headwater streams are
expected to be most effective primarily because of the increased chance of an effective
blockage forming in small constrictive channels. For this reason large woody debris
interventions are best suited for middle and upper reaches of a catchment.
Odoni and Lane (2010)198 undertook a modelling study of the Pickering Beck catchment
using OVERFLOW to assess the effect of catchment riparian intervention measures
(CRIMS) and woody debris dams. The CRIMS are described as 30m wide riparian
woodland corridors on either side of the channel. The woody debris dams were assumed
to form along the channel at a natural spacing of 7 to 10 times the channel width, based
on work by Linstead and Gurnell (1998)151. The woody debris dams and riparian
woodland were represented within the distributed hydrological model by localised
increases to the Mannings roughness values. The study represented large woody debris
dams via a Mannings value of 0.18 and riparian woodland via a value of 0.2. The
authors acknowledge that the selected Mannings values are at the upper limit of what
can be supported within the literature. As discussed in item (L) there is extensive
published guidance relating to the selection of Mannings numbers. The modelling
identified that the installation of 100 debris dams could reduce a flood event from
29.5cumecs to 27.3cumecs (7.5% reduction). It was identified that the magnitude of flow
-61-
reductions increased with event magnitude, thus suggesting that the effectiveness of
riparian woodland and large woody debris dams improves for the larger more extreme
events which cause flooding. This increase in effect goes against the findings of the
Robinwood study (Forest Research, 2008)150 and Gregory et al (1985)152. As with other
flood measures which desynchronise flows there is no hard and fast rule of thumb to
dictate where slowing flood flows will reduce downstream flood risk, indeed, many
potential locations for placing large woody debris were found to increase rather than
decrease downstream flood risk. Like the Robinwood study it was identified that the best
locations for riparian interventions tend to be in the upper catchment.
Gregory et al. (1985)152 studied vegetation dams in a small watercourse in the New
Forest over a 12 month period. The paper reports that 36% of the debris dams changed
location or were destroyed during this period, however the overall number remained
relatively constant, with a small increase being attributed to stormy weather. It was
estimated that the flood wave travel time at bank full condition was delayed by 10 minutes
over the 4028m study reach and that the delay in propagation was inversely proportional
to flow. It was reported that it was necessary to reduce the Mannings roughness value of
the reaches containing the debris dams to replicate the reduction in flood wave travel
times with a roughness of 1.02 being representing of lows flows and 0.31 being
representative of high (bank full) flows. This paper introduces terminology which
subsequent British literature has adopted:
Active Debris Dams dams that introduce a step in water elevation during low
flows
Complete Debris Dams dams which span the entire channel but do not generate
a step in water level during low flows
Partial Debris Dams dams which do not span the full channel width
Modelling of the River Laver by Nisbet and Thomas (2008)155 to assess the effect of
floodplain woodland on reducing downstream flood risk also included riparian woodland.
Unfortunately it is not possible to distinguish the effect of riparian woodland as all
presented model results include both riparian and floodplain woodland. Like the
Pickering Beck study the in-channel Mannings roughness values were increased (by
comparable amounts) in areas of riparian woodland to represent the presence of bank
vegetation and large woody debris within the channel. The findings of the River Laver
study were broadly inline with the other riparian woodland studies.
Modelling of proposed large woody debris dams in the Great Triley Wood was
undertaken by Forest Research to assess the impact of the placed structures on flood
propagation (Thomas et al., 2012)153. This study represented the large woody debris
dams within a calibrated InfoWorks RS model by using the Channel Blockage function
(application of 70% channel blockage) rather than increasing roughness. The approach
of applying a channel blockage allows the diminishing effect of large woody debris dams
with flow to be represented, with the study citing Gregory et al (1985)152 as evidence for
this reducing effect. Large woody dams were placed at an interval of 7 to 10 times the
channel width, consistent with Linstead and Gurnell (1999)151. The study identified the
large woody debris dam on average increased the reach time of travel by 2-3 minutes per
dam but that there was only a very minor change to flow. The suggestion that the large
woody dams facilitated the reconnection of the floodplain was seen as important. The
reconnection was partly influenced by the deposition of sediments upstream of the debris
dams which raised the bed elevation and proved a desirable location for natural
vegetation to colonise thus further increasing channel roughness.
The general consensus is that the creation of riparian woodland serves to slow in-channel
flow through the presence of bank vegetation and in-channel vegetation and that the
decrease in flow velocity can be represented via an increase in the Mannings roughness
-62-
value but recent work has been undertaken by partial channel blockages. It is agreed
that riparian woodland has the largest role to play in the upper and middle catchment
where watercourses are narrower and more prone to effective blockage by large woody
debris. While reductions of up to 10% in flood peaks are reported in some instances
these reductions appear to be subject to local conditions and the modelling technique
used. It is felt that the most appropriate means of accounting for riparian woodland is to
specifically apply a localised blockage as undertaken by Thomas et al., (2012)153 within a
hydrological model that includes a physically based flow routing engine. The alternative
is to use elevated Mannings roughnesses, however this approach will suggest that the
effect of large woody debris increases with flow and hence may require the Mannings
roughness to be varied with flow.
(L)
The complete forestation of the floodplain increased the floodplain storage volume
by 71% and delayed the flood peak progression by 140 minutes (baseline case
~3hrs).
The use of a 50Ha woodland block in the middle of the test reach increased
floodplain storage volume by 15% and delayed the travel time by 30 minutes
(baseline ~3hrs).
Peak water depths within the forested areas were found to be 50 270mm
deeper. This was based on increasing the Mannings roughness values for the
out of channel areas from 0.04 to 0.15.
The backwater effect of the 50Ha forest block (raised water levels by 120 to
180mm) was identified to extend nearly 400m upstream.
A detailed analysis of the flood hydrograph would identify where the restoration of
floodplain woodland would have the greatest benefits in terms of desynchronising
sub-catchments. Desynchronising, however, could extend the flood hydrograph,
with possible implications for longer duration or consecutive events.
Although it is very unlikely that floodplain woodland on its own would be able to
provide complete protection, it could make a valuable contribution along side
existing defences. Similarly, it could help manage smaller-scale flooding
problems where the high cost of constructing hard defences cannot be justified.
The paper does not directly report the change in velocity of the flood wave, however
using the graphs provided it can be estimated to be reduced from 0.73km/hr for the
baseline to 0.42km/hr and 0.63km/hr for the complete floodplain and 50Ha block
respectively. Unfortunately the paper doesnt specify the gradient of the watercourse,
however using the published hydraulic long section the gradient can be estimated as
0.00045m/m.
-63-
The paper does not report the reduction in flood flows as a result of the introduction of
floodplain forestation, however it is questioned quite how a percentage reduction for a
single hydrograph could be used since the increase in travel time for the reach would
likely increase the time to peak of the catchment. The longer storm duration required to
generate the critical storm would result in a marginally higher volume flood hydrograph,
thus any reported percentage reduction for a single storm hydrograph may be misleading.
Nisbet and Thomas (2008)155 undertook a subsequent study of the River Laver. The
modelling based study was undertaken to inform the placement of 40Ha of proposed
floodplain woodland. The study used 1D hydraulic model (InfoWorks RS) and a donor 1
in 100yr hydrograph. The modelling study identified that:
At each site the planting of woodland raised water levels over a backwater
ranging from 120m to 330m long. It was identified that the extent of the
backwater was a function of the river gradient and the increase in flood depth
caused by the planting.
There was a negligible impact on the peak flow rate at the downstream end of the
modelled reach with a maximum reduction of only 0.3% being identified.
The lag for each of the four sites considered ranged between 15 minutes and 20
minutes over the baseline case. The lag for the scenario where all four sites was
considered was estimated to be approximately 55 minutes.
The lag per hectare of floodplain planting was estimated to range between
1min/Ha to 3.5min/Ha, with the greatest lag per unit area being generated by the
smallest interventions. This reduction in effect suggests a law of diminishing
returns.
It was proposed that flooding in the town of Ripon could be reduced by the
strategic planting of the River Laver which joins the similarly sized River Skell just
upstream of the town. By delaying the peak on the River Laver and thus
desynchronising flows within the River Skell it was hypothesised that the flood
peak in Ripon could be reduced by 1 to 2% with the possibility of much greater
benefits if the extent of the planting was increased.
Roughness values of 0.03 and 0.05 were assigned to the channel and floodplain,
respectively, to represent the bed roughness associated with the nature of the existing
river channel and the baseline grassland or arable land cover. The establishment of a
cover of native floodplain woodland was represented by increasing the channel
roughness to a value of 0.10 and the floodplain roughness to 0.3. The report authors
accept that the adopted value for floodplain woodland is considered to be at the upper
limit of possible values for floodplain woodland and to attain such a high roughness the
woodland would need to be particularly dense with plenty of undergrowth, low branches
and fallen woody debris. The high channel roughness was justified as being a
consequence of an increased supply of large woody debris leading to an increase in large
woody dam formation.
Based on the provided river long section within the report body, the hydraulic gradient of
the watercourse in the vicinity of the floodplain roughening was identified to be typically
0.006m/m.
Initially 40Ha of woodland was considered feasible and a subsequent baseline monitoring
programme was put in place to collect valuable real data. Unfortunately landowner
interests resulted in the extent of the woodland planting being limited to an 8Ha plot at a
location prescribed by the landowner. This resulted in the baseline monitoring
programme being dropped.
-64-
Sensitivity testing of partial afforestation of the floodplain sites suggested that the most
effective placement for delaying flood propagation was within the lowest lying parts of the
floodplain, suggesting that corridors of planting along the sides of the river may exert
greater than average effect i.e. riparian woodland. However in preparing this review it is
considered possible that this increase may be a side effect of the 1-D modelling approach
and it may be more appropriate to conclude that the effectiveness of increasing
roughness is dependent on flow velocity (which will be highest in deepest areas of a 1-D
river cross-section). The areas of high flow velocity will be typically found in the lowest
areas of the floodplain so there is no issue with the overriding conclusion.
Floodplain woodland is presented within Woodland for Water: Woodland measures for
meeting Water Framework Objectives (Nisbet et al., 2011)65 via reference to the Cary and
Laver studies (also completed by Forest Research). The report proposes that floodplain
woodland might have a decreasing effect with more extreme floods.
Staffordshire Wildlife Trust has led a project in the River Sow and Penk (Jones, 2010)156.
A review of the project report indicates that the study focused on ecological service and
that while the implemented measures will have hydrological benefits no modelling was
undertaken to assist in the design of the works. It is understood that minimal floodplain
woodland has been introduced however a number of study sites involve the creation of
wet meadows with higher roughness introduced by ground scrapes and longer grasses.
A hydrological monitoring program is in place however it is apparent that this focuses on
the functioning of the ecological features and together with an absence of baseline data
may not be appropriate for assessing the hydrological impact of the measures during
extreme events.
Verbal communication with Tom Ball of University of Dundee has identified that as part of
the Eddleston Water project it is understood that floodplain woodland is proposed on the
Longcote Burn. The primary function of this floodplain woodland is to attenuate flood
flows. It is understood that a 1D modelling study will be used to inform the placement of
woodland and to assess its effectiveness. There is an extensive hydrological monitoring
programme in place which will provide much needed real data.
A study has been undertaken at Trier University in Germany to investigate the function of
forest management on the flood retention characteristics of floodplain forests (Sartor,
unknown year)157. The unpublished study presents the 2001 reforestation of 2.2% (23Ha)
of the River Blies in Germany floodplain, a catchment of approximately 1800sqkm. Using
hydraulic modelling it was estimated that extreme events (1 in 75 year) would be reduced
by 1%. The paper reports that another investigation in 2005 confirms the authors own
findings, unfortunately this supporting paper is in German so has not been accessed
(Aatz & Musong, 2005)158. It is reported that the complete forestation of the floodplain
was assessed to have the potential to reduce the peak flow rate by up to 15%.
Forest Research undertook a review of Short Rotation Forestry (SRF), incorporating
Short Rotation Coppice (SRC)(Nisbet et al., 2011)72. This review proposes that the
greater planting density and faster growth of SRF could be expect to provide a rapid
means of roughening floodplains in addition to achieving roughnesses greater than
convention woodland. As a further benefit the report highlights that the lack of large
woody debris generated by SRF crops [not mentioned: the ability to entrap floating debris
via the close plant spacing] would reduce downstream blockage risks (discussed in
Appendix C). Importantly SRF may be more attractive to landowners than conventional
forestry. Care would be needed to avoid SRF in water stressed catchments as discussed
in item (E).
There are a number of sources which can provide guidance on the selecting of Mannings
roughness values and this area can be considered well documented. It is beyond the
scope of this literature review to provide guidance on how to select appropriate Mannings
values for channel and floodplain flow, however further reading sources include:
-65-
(M)
Washland creation
Washlands are areas of managed floodplain that are allowed to flood, or are deliberately
flooded, this can be for a range of reasons including flood attenuation or habitat creation
-66-
(Hardiman et al., 2011)169. Typically, manual interventions are used to control how water
enters or exits the washland storage area so that the available storage is reserved for the
flood peak or flood waters are artificially held back to be discharged after the flood peak
has passed. It has been demonstrated that washlands generally provide a higher level of
flood attenuation than more natural floodplain systems (Shultz and Leitch, 2003)170. As
washlands are artificially altered floodplains much of the literature relating to floodplain
connections discussed in item (N) are applicable to washlands however given the
increased variability due to manual intervention the need for specific modelling of the
system is likely to be even greater than floodplain systems.
A review of washlands is provided within Morris et al. (2004)171. The English Nature
commissioned report defines washlands as an area of the floodplain that is allowed to
flood or is deliberately flooded by a river or stream for flood management purposes, with
potential to form a wetland habitat. The report identified that most washlands have
either been used mainly for flood storage or for wetland habitat and few attempts have
been made to integrate the two objectives. There are some conflicts between ecological
and flood risk management considerations including the need to keep washlands wet in
the interflood period to sustain ecology, consequently reducing the volume of storage
which is available to attenuate floods. Washlands vary between the duration of flooding,
the inlet and outlet mechanisms, and the habitat types found within them. Some
washlands are configured so as to reserve the washland flood storage for the flood peak
thus ensuring the maximum benefit in attenuation can be drawn from the available
storage volume. The outlet at some washlands can be manually controlled, thus there is
the potential to delay the release of flood water until the flow in the receiving watercourse
has receded.
(N)
There is limited literature relating to the use of placed large woody debris dams and
boulder placements. Odoni and Lane (2010)198 present the use of woody debris dams in
the Pickering Beck catchment to reduce downstream flood risk by slowing in channel
flows and thus creating additional flood storage volume by raising in channel water levels
and increasing floodplain connectivity. It should be noted that this approach requires a
large number of discrete interventions which has the potential to be expensive. There are
no hard and fast rules of thumb to dictate where the placement of large woody debris
would have the desired impact of reducing flood flows, indeed, many potential locations
for placing large woody debris were found to increase rather than decrease downstream
flood risk. Modelling using OVERFLOW identified that the installation of 100 debris dams
could reduce a flood event from 29.5cumecs to 27.3cumecs (7.5% reduction). The study
found that the large flood events experienced a greater percentage reduction than
smaller events however it should be noted that the placed woody debris was modelling by
localised elevation of Mannings roughness values which cannot take account of the
woody debris becoming drowned.
There is significant cross over between artificial large woody debris and riparian
woodland hence section (K) should be read for further information.
No literature relating to the Natural Flood Management benefits of in channel boulder
placements could be identified. It is assumed that these would have a similar, but notably
smaller impact than placed large woody debris.
(O)
River restoration
Many river restoration projects cite flood risk management benefits in support of the
restoration project, however it has been highlighted that there is an absence of data
detailing the nature of the flood risk management benefits these projects can have (MNV,
2008)11. In many cases it is evident that river restoration is undertaken to address
-67-
specific local issues. River restoration can have a role to play in delivering Natural Flood
Management by three means:
1.
2.
Extending the flow path length through the reintroduction of meanders thus
reducing the channel slope, reducing in-channel flow velocity and increasing
travel time.
3.
(P)
Beach recharge
(Q)
Defra FD1302 (Pye et al, 2007)176 provides an authoritative overview of the use of sand
dunes for flood risk management. Part 2 of the report indicates that dune systems which
-68-
are both wide and high provide the greatest flood defence value. It is noted that narrower
systems do provide a reduced benefit however it is common for these systems to be
receding due to erosion. It is reported that dune systems in the Netherlands have been
assessed to offer a 1 in 10,000yr standard of protection to large areas. Part 3 of the
report describes dune restoration activities which predominately include fencing, control
of visitors, construction of board walks and revegetation. A guide to managing coastal
erosion in beach/dune systems (SNH, 2000)188 provides detailed guidance for sand dune
restoration projects.
(R)
-69-
The area of saltmarsh immediately landward of the permanently vegetated edge is most
significant in terms of attenuating wave energy, so vegetation changes in this area are
likely to have the greatest influence. Mller and Spencer (2002)179 observed a correlation
between lowered wave attenuation and seasonal decreases in vegetation density over a
gradually sloping saltmarsh edge. The same study observed no such relationship over a
steep eroding saltmarsh margin, suggesting that the dominant effect on wave dissipation
at this type of site is the eroding topography rather than the vegetation.
The vegetation type can also influence wave attenuation rates, due to the different
physical structure of the plants. Across Dengie, Mller et al. (1999)178 identified a possible
link between vegetation density, wave attenuation and the type of vegetation, with flexible
grasses in the upper saltmarsh having less effect on wave attenuation than woody plants.
(S)
Reed (2009)182 provides a concise overview of the function of estuarine tidal flats in
relation to flood risk by highlighting their value in damping storm surges from the ocean or
by providing storage of riverine floodwaters.
A review of coastal managed coastal alignment practices in England and Germany was
undertaken in 2002 (Rupp and Nicholls, 2002)183. The comparative study involved the
interview of six key managed realignment experts, three in each country. The general
consensus was that coastal realignment is correctly viewed as a cheap and
environmentally friendly option, provided the practical conditions on the ground are not an
insurmountable barrier. In both countries it was identified that at that time (2002) there
was a shortage of well documented managed retreats, with three pilot sites in the
Blackwater estuary in Essex being the best documented English site. All tidal retreat
schemes in the UK to that date amounted to little more than 100Ha. In most cases
managed retreats have been undertaken primarily for habitat creation purposes. It is
reported that there is agreement between the experts that the additional flats created by
retreat will assist in attenuating wave energy during storms. It should be noted that it was
identified by the probably more reliable work by Mller (2001)181 that the effect of wave
attenuation decreases with water depth and thus attenuation might be minimal during
extreme tidal surge events when it is most needed.
A published modelling study was undertaken to assess the feasibility of adopting
managed realignment as an adaptive management to sea-level rise (French, 2008)184.
The study uses the Blyth estuary in Suffolk as a case-study, reporting the estuary to be a
morphologically complex and heavily engineered. The reported model results indicate
that a sea-level rise of 0.3m (central estimate for 2050) would necessitate extensive
upgrades to existing tidal floodplain defences. Realignment (or managed retreat) of
these defences can reduce local flood defence costs by eliminating unsustainable
seawall but needs to be evaluated in the light of wider impacts. Modelling of hypothetical
realignment scenarios shows that restoration of tidal exchange to the largest flood
compartments could have an immediate effect on outer estuary hydrodynamics that is
larger than worst case scenarios for half a century of accelerated sea-level rise. More
generally, incompatibilities are apparent between flood defence and habitat restoration
objectives, such that the appropriateness and feasibility of large-scale flood defence
realignment should not be assumed in estuarine contexts.
As part of flood risk management within the Humber Estuary the Environment Agency is
utilising coastal realignment as a tool within a sustainable flood risk management
strategy. To date four coastal retreat sites have been created (OMReg, 2012)185:
It is understood that the Paull Holme Strays, Chowder Ness and Welwick sites where
undertaken for habitat compensation purposes. It is reported that the Alkborough site
increases the level of flood protection to an area stretching from the Humber Bridge to
Goole up the tidal River Ouse and as far as Keadby Bridge on the tidal River Trent. The
scheme features include a 20 metre-wide breach of the existing flood bank, a 1,500
metre length of lowered embankment or spillway and a new section of floodbank to
protect assets at the edge of the site. For a 1 in 200yr event the scheme at Alkborough
will reduce extreme water levels by more than 150 mm (EA, 2006)186.
A directory of managed coastal realignment projects is hosted by ABP Marine
Environmental Research Ltd (OMReg, 2012)185. The database includes detailed
information on 51 managed breaches, regulated exchanges and defence removal
projects across the UK since 2001. A review of the data held within the database was
presented to the ICE by Scott et al (2011). In total 1,300Ha of coastal retreat has been
created since 1990, and as a result there is an increasing knowledge base relating to the
subject. The paper identifies that the scale and number of coastal retreat projects have
both increased in recent years as shown Figure 14. The paper also provides a useful
insight into the cost of implementing coastal realignment projects and finishes with the
closing proposal that it may be time to consider very large scale retreat projects
(substantially greater than 1000Ha) to improve the prospect of reaching national
ecological targets. CIRIAs C628 (Legget et al., 2004)187 provides a useful overview of
the issues associated with managed realignment projects.
Figure 14: Unit costs of implemented realignment plotted against size and year. Scott et al.
(2011).
(T)
-71-
Chasten, M.A., Rosati, J.D. Mc Cormick, J.W., Randall, R.E. (1993) Engineering
design guidance for detached breakwaters and shore stabilisation structures.
CERC-93-19, US Army Corps of Engineers Waterways Experiment Station,
Vicksburg, Mississippi.
CIRIA, (2007) The Rock Manual, Second Edition, CIRIA C683. ISBN 0-86017683-5.
The application of biogenic reefs to act as breakwaters in the Mobile Bay, Gulf of Mexico,
indicates that oyster beds can be used to protect an eroding shoreline (Cyphers, et al.,
2011)189. The study sites used loose shells held temporarily in place by mesh until
adequate oysters are recruited to cement the loose shell substrate in place. This
naturally cemented substrate then can provide an effective protection to medium energy
eroding sites. It was found that the biogenic reef slowed the rate of erosion by 40% when
compared against a control site. The study accepts that this technique is not viable for
high energy sites where due consideration may need to be given to more traditional
measures.
(U)
It is understood that West Sands Partnership in Fife is trialling techniques which will
improve the erosive resistance offered by the strandline and that St Andrews and Abertay
Universities are undertaking research to assess its effectiveness (pers comm). The trial
includes the hand picking of litter rather than using mechanical collectors with the
intention of leaving natural material to form a natural strandline. With time it is hoped that
the approach may improve the natural colonisation of the standline by vegetation allowing
the seaward extension of the dune system. Early indications are that the modified
maintenance regime is proving very effective at stabilising the seaward edge of sand
dunes.
3.3.2 Summary
The literature relating to a broad range of Natural Flood Management measures or
techniques have been presented in this Section of the report. The quality and extent of
literature relating to the 21 techniques is variable, however it is apparent that all
measures have a role to play in the future sustaibable management of flood risk. Some
measures have been shown to be effective in most cicumstances while evidence
suggests that other measures can be counterproductive if implemented in the wrong
-72-
-73-
(A)
This is a national screening tool which has been developed for the Environment Agency.
The tool is GIS based and employs commercially available national data to calculate
Combine Runoff Sensitivity on a 1km grid for the whole of England and Wales. The tool
considers the following aspects:
Soil class, based on NATMAP1000 (there may be scope to use HOST data),
scored 1 to 4
These scores are then combined via the following equation to produce a national grid of
sensitivity to land use change:
Sample output from the tool are shown in Figure 15 and Figure 16,
The tool is configured for English and Welsh type catchments and would require some
minor modifications before it could be applied in Scotland, i.e. it the rainfall banding runs
from <600mm a year (Very low) to >900mm a year (High).
The tool does not include a flow routing element for flood flows and only looks at runoff
generation potential. The incorporation of a flow routing element together with
incorporation of known downstream flood receptors is a recommendation for a future
extension of the tool.
-74-
Figure 15: Sample output, River Ouse combined sensitivity (Environment Agency, 2008)
-75-
(B)
The purpose of this bespoke tool was to assist the Environment Agency in the
implementation of a delivery plan for the Catchment Flood Management Plans (CFMP) by
assessing the effect of rural land use and land management on the flood hydrograph.
The tool employs the WaSim daily lumped soil water balance model to simulate the
effects of climate, soil type, field condition and land cover on daily runoff depth and
antecedent moisture conditions. The changes in maximum daily runoff depth have then
been used in the Revitalised Flood Hydrograph (ReFH) model to assess the change in
peak flow.
The Land Management CFMP Tool allows users with local knowledge to explore how
realistic changes in land use can quantitatively affect the downstream hydrograph. Within
the Excel interface the user has the option to change the proportion of the following broad
land cover classes:
Managed grassland;
Cereals;
Horticulture/non-cereal crops;
Semi-natural; and
Woodland.
The effects of land management can be explored through changing the distribution of
field condition which encapsulates both the condition of the soil and the presence of land
management practices which alter runoff transmission.
The report stresses that, despite the inevitable assumptions and simplifications
necessary to develop the tool for national applicability, the scenarios suggested changes
in peak runoff which are reasonable within the context of the uncertainty of current
scientific understanding. The methodology is reported to be repeatable, academically
sound and supported by the Environment Agency.
Figure 17: User interface of the Land Management CFMP Tool (Environment Agency, 2008)
-76-
(C)
A range of GIS data sets were used to locate land vulnerable to flooding and unaffected
by constraints to woodland planting. The approach built on previous work in the
catchments of the River Parrett in Somerset and Bassenthwaite Lake in Cumbria. The
main output is a series of maps showing opportunities for planting floodplain and riparian
woodland for flood mitigation within the region. Areas were prioritised according to the
scope for planting to generate added value for nature conservation and water quality.
Floodplain sites were ranked by their potential to create an extended forest habitat
network, while riparian zones took into account the risk of the adjacent land delivering
sediment to watercourses.
The project also assessed opportunities for woodland planting to assist flood and water
pollution management within the wider catchment. This used data sets that classified the
catchment soils by their vulnerability to generate rapid surface runoff, degrade structurally
and/or deliver sediment to watercourses. The result is a map ranking areas as low,
medium or high priority for woodland planting.
Significant opportunities exist within the region to restore floodplain woodland for
sustainable flood management. A total of 35,328 ha of the fluvial floodplain is potentially
available for woodland planting, including 168 major sites with an individual area of >50
ha. The majority of these lie within the catchments of the River Derwent, River Swale,
River Ure and central section of the River Ouse, where there is potential to help reduce
the flooding of small towns and villages, as well as major towns such as York.
There are also many opportunities across the region for using riparian woodland to help
reduce flood flows. A total of 18,730 ha are available for planting in the upper reaches of
most rivers, including those catchments in the Southern Pennine Fringe at high flood risk
but with limited potential for planting floodplain woodland. Around 2,562 ha is adjoined by
land at high risk of sediment delivery, where woodland creation could potentially benefit
both flood risk management and diffuse pollution control. Some 997ha lay within ECSFDI
priority catchments, although most of this is located in the catchment of the River
Derwent.
Finally, 65% of the land in the wider region is potentially available for woodland creation
for multiple benefits, including flood reduction. Some 1,538sqkm or 16% of the region is
mapped as high priority for woodland planting on the basis of soil propensity to generate
rapid runoff, soil sensitivity to structural degradation, and/or high risk of sediment delivery
to watercourses. This land tends to be concentrated in the upper parts of river
catchments and is most extensive within the River Ure, Swale, Derwent and Nidd
catchments.
It is hoped that the opportunity maps will be used by regional stakeholders to promote the
use of woodland in sustainable flood management and in so doing, help to meet the
following objectives:
Delivery of reduced flood risk through effective and better integrated Catchment
Flood Management Plans
Delivery of the regional Biodiversity Habitat Action Plan targets for wet woodland
and an enhanced forest habitat network
Figure 18: High priority areas with the greatest potential for woodland planting to reduce
193
downstream flooding (Broadmeadows & Nisbet, 2009)
Limitations:
(D)
The tool provides an overview of opportunities within the catchment, it does not
quantify the benefits which might be achieve from implementing that measure nor
does it provide an indication of which measure is most beneficial
There is not assessment of time to outlet hence the tool does not allow
assessment of the contribution that the identified opportunity might have to the
flood peak
A single event distributed hydrological model which estimates runoff rates using the
American Soil Conservation Services Runoff Curve Number methodology. The ArcGIS
model includes overland and channel flow routing. Channel flow routing is via a modified
normal depth routine (kinematic wave). The method uses a grid smoothing technique to
-78-
provide a means of inferring the backwater profile, while being less accurate than a
diffuse wave method the approach is numerically robust and computationally efficient.
The model predicts channel dimensions and roughness values based on estimates of
median annual flood and channel slope. The channel routing routine includes
functionality to represent floodplain flow however floodplain width is based on typical
floodplain widths (based on valley slope) found within the catchment. Floodplain
roughness values are based on a user supplied grid and the threshold for spill of water
from the channel to the floodplain can be user defined (site reconnaissance or review of
the morphological pressures database). The transition from overland runoff to channel
flow can be user defined, i.e. using a mapped river network or once a catchment area
threshold is exceeded. Overland flow routing is based on a simple normal depth routine
which assumes a constant depth over the cell width, the roughness used for this routine
is via user supplied grid which can be based on landuse.
Vegetation interception and initial ground infiltration is based on the simplified initial
abstraction provided within the SCS methodology. It should be noted that this has been
found to be dependent on local antecedent moisture conditions and should be reviewed
to maintain a Scottish context.
The method does not simulate the whole storm hydrograph, instead the hydrograph peak
is estimated by estimating the time that the flood peak would occur at and then
accumulating the runoff generation from across the catchment that would arrive at the
catchment outlet at that time.
The model does not include groundwater flow and it is assumed that all infiltrated water is
lost from the system. It would also be desirable to improve the channel routing routine to
be based on the more accurate diffusive wave.
The model has not been formally tested however tests on the Allan Water indicate time to
peaks similar to estimates made using the FEH time to peak by catchment descriptors.
The model is encoded as a ToolBox within ArcGIS in an incomplete state; that is no clear
user interface, user prompts or help files. The incomplete state also means that applying
the model is user intensive requiring output from one routine to be manually passed to
the next routine.
(E)
The Natural Flood Management tool proposed by Jacobs as part of a SEPA funded
project to develop a methodology to apply Section 20 of the Flood Risk Management
(Scotland) Act is presented in detail within Section 2 of this report. In summary the
ArcGIS tool allows the user to scale hydrograph peaks based on land use changes within
the catchment. The percentage changes included within the tool are based on published
literature and uncertainty is managed by the tool via the use of qualitative uncertainty
bands pessimistic, central and optimistic. The tool incorporates functionality to
estimate the percentage change in flows for the 1 in 5yr and 1 in 200yr flood events for
the following land use change types:
Forestation;
Riparian woodland;
Drain blocking.
The S20 GIS tool uses the Automated Flow Grid produced by CEH as the baseline
hydrology, this gives the tool good national coverage but introduces a challenge for
distributing the tool due to licensing. As presented in Section 2 of this report the tool
suffers from a number of fundamental technical issues including how the hydrological
-79-
benefits of the measures, particularly floodplain enhancement, are assessed, and the
assumptions based around the division of the catchment into thirds.
(F)
Farm impact:
o
Flood mitigation:
o
It is evident that the tool is centred around the principle of optimally placed
shelter belts based on the research conducted by H. Wheater at Imperial
College London, but there is no published information on how this is
achieved.
Tradeoffs:
o
Tool explores trade offs, multiple benefits and opportunities for change
-80-
Figure 19:
Sample output
from the
POLYSCAPE tool
for the Pontbren
catchment
(Pagnell, 2009).
Previous consultation in 2010 by Scottish Borders Council with the author of the toolbox
indicated it was unclear if he was going to continue the development of the tool which
was at the time in an unfinished state. There is no evidence that the development of the
tool has continued since 2010, hence this tool does not represent a viable option for the
screening or assessment of Natural Flood Management measures in Scotland.
(G)
FARM is a decision support matrix designed to allow farmers and land use planners to
investigate land use interventions at a field scale. The tool is written in Excel and is
currently available for download.
The tool is aimed at farmers and local flood risk managers who have a high level of local
knowledge but less awareness of what constitutes a high or low runoff risk. As the
FD2114 Impact Study Report has stated, the factors affecting catchment scale runoff are
difficult to quantify and are still open to some debate, hence, the FARM tool has been
deliberately aimed at the farm scale where runoff processes can at least be observed and
are backed up by the ISR findings. Aggregation to the catchment scale can be performed
via a FEH toolkit (Packman et al., 2004)48.
The tool functions by asking the user a series of questions to identify the soil infiltration,
storage and tillage regime in addition to the flow connectivity. Throughout the process
the user can see the impact of their answers through the animated highlight of the matrix
(Figure 20). Examples are provided to assist the user in answering the questions (Figure
21).
-81-
Soil and
management
Connectivity
Figure 20: Example of current UK farms and likely impacts of land use management
scenarios and how they can be mapped onto the decision support matrix (Hewett et al.,
2006)
Figure 21: Four hillslope runoff risk scenarios for the same common land unit (Hewett et al,
2006).
-82-
Given the simplicity and field scale of the tool, it is considered that FARM is most
appropriate for educating land managers. Other tools such as the Land Management
CFMP Tool (Section (B)) will be a more valuable resource for professional practitioner
use.
(H)
It is understood that the model incorporates the following assumptions (Odoni & Lane,
2010)198:
Rivers are assumed to form when the flow exceeds 0.06cumecs, with watercourse
dimensions being estimated using generic channel width and depth equations
such as those published by Leopold and Maddock (1953)201.
In channel and out of bank flow resistances are estimated using representative
catchment wide Mannings roughness values. It is noted that there is potential to
vary Mannings values in any way that the user chooses provided there is
adequate data and justification to do this.
With assumptions made concerning input rainfall, baseflow, runoff percentage, channel
geometry and Mannings, the model has all the components needed in order to run. A
flow accumulation and routing algorithm calculates how rain falling on the catchment is
made to flow through the landscape, with the assumption that the catchment is saturated
and the water that contributes to the flood wave moves primarily by overland flow. Flow
paths are allowed to vary as a function of rainfall depth, as well as in situations where
some of the flow is out of bank and routed across floodplains. The flows are converted to
inferred flow depths and velocities via Mannings thus allowing an estimation of travel
time. By accumulating these along flow paths, starting at the outlet, a flow time map is
generated for the rain rate in question (Figure 22).
-83-
It is understood that the tool does not include a means of representing spatially varying
rainfall or percentage runoff across the catchment although in the available material it is
stated this is something that the author is working towards. There is no discussion on
how the attenuation effect of lochs and reservoirs might be incorporated within the model.
OVERFLOW has to date been only applied to assess CRIMS (Catchment Riparian
Intervention Measures) such as placed large woody debris and riparian/floodplain
planting. Should the model be improved to include modules to assess the spatial
variation in runoff generation (based on land use and soil types) and to allow the variation
in rainfall across the catchment then the modelling method could be of use for assessing
Natural Flood Management measures in Scotland.
It is unclear if the tool would be commercially available for the Section 20 assessment as
previous approaches by Scottish Borders Council have found that the developer is not
keen to release the model to industry at this time requiring any assessment to be
undertaken by staff at Durham University.
(I)
The Diffuse Pollution Screening Tool presented within report WFD19 (ADAS & MLURI,
2006)202 includes a water balance model. The model functions on a 1km grid and
produces monthly estimates of the generation of overland flow, sub-surface flow and
drainage to groundwater. The model makes allowance of the spatial variation in
precipitation, evapotranspiration, land use and soils and has been used to produce
monthly estimates of overland flow, sub-surface flow and ground water for the 10 year
period from 1989 to 1998. It should be noted that the tool was not developed for flood
risk management purposes but it may present a means of quantifying likely antecedent
moisture conditions following land use changes in addition to useful data sources.
-84-
(J)
There are nearly as many hydrological models as there are hydrologists (Beven, 2012)203,
with many hydrologists keen to develop and promote their own hydrological model. Many
of these hydrological models are physically based distributed models such as:
CRUM3
Physical model
- a model that
seeks to represent hydrological
processes in a physically realistic
manner.
Semi-physical model
-a
hydrbid model which represents
some hydrological processes in a
physical manner and others via
functions that have no direct physical
basis.
Lumped model - A model that treats
the whole of a catchment as a single
accounting unit and predicts only
values of variables averaged over the
catchment area.
Beven (2012)203 advocates the use of semiphysical models which lie in a grey area
between physical and conceptual lumped
models, two such models are GRID-TO-GRID
(G2G) and TOPMOD. These semi-physical
models have a much reduced parameter set
when compared to physical models, therefore
the process of supplying data and calibration is made much more straightforward.
Normally the parameters sets have an underlying physical basis to key hydrological
processes but there is no attempt to represent every process that occurs within the
catchment. He notes that even the current state of the art distributed physical models do
not present a significant improvement in accuracy over simple lumped models and he
warns that an increase in complexity does not necessarily lead to an increase in
accuracy.
There are currently a number of research initiatives seeking to improve the conditioning
of rainfall-runoff hydrological models; two recent publications include Bulygina et al
(2009)205 and Moore et al (2007)206.
(K)
It is understood that a planning tool has been developed as part of the Louisianas
Coastal Management Plan development which is being developed in the wake of
hurricane Katrina. The tool functions around an index which is based on the location and
amount of land in proximity to flood receptors, vegetation and elevation. The index also
includes the effect that changes might have on the attenuation of waves and surges.
The approach has allowed 400 project options, considering of terabytes of data, to be
assessed in a user friendly manner covering areas such as costs, duration, landscape
and stakeholder preferences. The planning tool is not designed to make decisions but
instead it provides the information to assist in the decision making process of grouping
projects together to achieve coast wide goals.
-85-
-86-
To take into account the identified Natural Flood Management potential when
setting objectives and identifying measures in flood risk management
strategies and plans (FRM Sections 27, 28 and 34).
The methodologies proposed in this report aim to deliver requirement under Section
20 to produce maps of Natural Flood Management potential, whilst also proposing
how this information can be used in the objectives setting and identification of
measures in flood risk management strategies. Figure 23 summarises the overall
approach to delivering Natural Flood Management requirements.
-87-
2012
SEPA
Section 28 Appraisals
2013 - 2014
SEPA / Responsible
Authorities
2015 16
The relevant
Responsible Authority
Figure 23: Outline process for the delivery of Natural Flood Management requirements
under the FRM Act
-88-
Baseline
national
pluvial, fluvial
and coastal
flood maps
Appraisal
methodology
and additional
guidance
Catchment
characteristics
and local
information
Implementation of
measures
Figure 24: Flow chart showing the inclusion of Natural Flood Management within the
FRM Act
It is proposed that the identification and appraisal will be divided into three main
phases:
1.
A national screening process to identify opportunities for Natural Flood
Management measures. This screening process will identify areas within catchments
with Potentially Vulnerable Areas that have natural flood management opportunities
but will make no consideration of constraints or other benefits. The screening
process will not directly recommend which specific measure should be implemented
where, nor will the screening facilitate the quantification of the flood risk management
benefits of undertaking a specific natural flood management activity. However the
-89-
process will facilitate the identification of areas that are worth further investigations at
a later stage. The main output of the screening process will be six maps showing:
The proposed methods for constructing the GIS based screening tools which will be
used for this assessment are presented in Section 4.3.2. Please note that subcatchment desynchronisation will not be undertaken due to unacceptable uncertainty
in the underlying science and data constraints.
It is proposed that these maps, will meet the requirements of Section 20 of the FRM
Act. These maps will be made publically available along with appropriate supporting
descriptions and guidance.
2.
The second phase will seek to bring together the outputs from the screening
process with further information about catchment characteristics in order to develop
better understanding of what may be achievable where within the catchment. Using
the findings of the screening process in combination with other catchment information
the measures will be presented as options to manage the sources and pathways of
floodwaters, as summarised in the Natural Flood Management Summary Table
(Appendix B). This information together will be used to appraise natural flood
management measures alongside other flood risk management measures. Over the
course of this stage the long list will be reduced to a short list and the short list
appraised to identify a basket of the most sustainable measures. The process will
give due consideration to environmental, economic and social costs and benefits.
The end result of this stage is a short list of options that will be agreed by all
responsible authorities.
3.
The third phase will include more detailed assessment of the agreed,
measures using existing and new assessment approaches. This more detailed
assessment will be carried out by the relevant responsible authority as part of local
flood risk management plans.
The Section 3 literature review has identified that some measures can be assessed
using existing tools and methods, while the assessment of other Natural Flood
Management techniques require the development of new tools. It is proposed that
the assessment is undertaken using existing tools when available. In the case of
fluvial and pluvial flood risk there is a general absence of existing tools and methods
which can be used to quantify the effects of Natural Flood Management measures. It
is therefore proposed that a new hydrological method is developed to facilitate the
quantification of fluvial and pluvial Natural Flood Management measures. Section
4.4 presents a proposed specification for how this could be accomplished using a
single event spatially distributed model based on the uniformly distributed PDM
(Probability Distributed Model) runoff generation model.
-90-
-91-
This screening process aims to produce a national dataset of runoff potential with the
intention of providing an insight into which areas contribute most to the generation of
fluvial and pluvial flows. With this information, it will be possible to review the existing
-92-
landuse, then in conjunction with the information contained in the Natural Flood
Management Summary Table in Appendix B and the local constraints or
opportunities an alternate reduced runoff landuse can be identified. In this way runoff
reducing activities can be targeted to the areas where they will be most effective
before a quantitative assessment of the hydrological effect is undertaken. A number
of means of completing the screening for high runoff areas have been identified:
1. Using the Environment Agencys Method for the identification of catchments
sensitive to land use change (JBA, 2008)191
This previously applied method creates a Combined Sensitivity Score via a method
described in Section 3.4(A). It takes into account land cover, soil, slope and rainfall,
but does not apply a weighting to these factors. There is also a data availability issue
regarding a degraded soil dataset used by the method. The method is built around
qualitative judgement based scores which are representative of conditions found in
England and Wales, thus application in Scotland would require additional study.
Equation 2 summarises how the land cover, soil, slope and rainfall classes are
combined to generate a combined score.
Equation 2: Combined runoff sensitivity score used within the Environment Agencys
identification of catchments which are sensitive to land use change.
The supporting report for the method indicates that it could be improved by weighting
the effect of the model variables but it does not recommend a means of achieving
this.
2. Using output from the Diffuse Pollution Screening Tool (ADAS and MLURI,
2006)202
The existing monthly overland and sub-surface flow data produced as part of this
study could be used to identify areas of high runoff generation. It should be noted
that the tool was not developed with the intention of assessing flood risk and
therefore it may be inappropriate to use the output in this way. The model makes
allowance of the spatial variation in precipitation, evapotranspiration, land use and
soils. It is also not clear if only monthly averages were produced or whether the
totals for each month in the 10 year period are available thus allowing an analysis
based on annual maximums.
As this method was not developed for the purpose of assessing flood events it is
considered inappropriate to use it for this purpose.
3. Using a modified version of the QMED by catchment descriptor equation
(Kjeldsen et al., 2008)208
The median annual maximum flood ( QMED ) for a catchment can be calculated
using Equation 3.
0.8510
0.1536
1000
SAAR
FARL3.44510.0460 BFIHOST
208
The catchment area ( AREA ) and catchment storage ( FARL ) elements could be
dropped out of the equation to produce a measure of the median annual flood
generated per unit area ( q ) as detailed in Equation 4.
-93-
q 8.3062 0.1536
1000
SAAR
0.0460 BFIHOST
Some land use pressures could be represented using the method of degraded soils
via analogue HOST classes proposed by Packman et al (2004)48. However the
method would not permit the inclusion of topography or the representation of existing
land use activities.
The alignment of this approach with current British flood hydrological methodology
and its simplicity make it a good alternative should the preferred (next) method not be
feasible.
4. Combination of the Environment Agencys Method for the identification of
catchments sensitive to land use change (JBA, 2008)191 and the modified QMED
by catchment descriptor equation (Kjeldsen et al., 2008)209
There is potential to produce a hybrid method based on options 1 and 3. This
approach would facilitate the quantitative of method 3 to be modified by factors
relating to land cover and slope. It would be necessary to undertake further study to
develop what weightings might be appropriate for the various components. A
potential relationship could take the form of Equation 5, where local topographic
gradient ( g ) and leaf area index ( LAI ) are modified by yet to be defined coefficients
( , , , , ).
q 8.3062 0.1536
1000
SAAR
slope _ weight 1 g
Initial suggestions for the slope coefficients could be in the region of 0.1
and 1 , these values would limit the effects of slopes to increasing the runoff
generation score ( q ) by 15 to 25%. It should be noted that these factors are
suggested based on speculation only. The factors must be selected and robustly
justified as part of the screening process. The variation of soil moisture capacity with
slope, as discussed in Section 4.4.2(C), could provide a scientific basis to this
weighting factor.
In regards to land use effects, there may be scope to track through the effect of LAI
on percentage runoff (Equation 26) by applying a range of LAI to Equation 31
together with typical Scottish SAAR, Potential Evapotranspiration (PE) and bulk
(canopy) resistance ( rsc ). Figure 25 shows one example of tracking the effect of LAI
through using some typical Scottish values, this suggests the relationship for the land
use weighting as detailed in Equation 6. Again it is essential that these factors are
reviewed and robustly justified as part of the screening process.
-94-
0.69
Percentage Runoff
0.68
Median LAI for Scotland
is ~4 (4^4.7 = 675)
0.67
Generated using a
50mm rainfall depth
0.66
0.65
PR = -0.00002680(LAI^4.7) + 0.68854927
0.64
0.63
0
200
400
600
800
1000
1200
1400
1600
1800
2000
LAI^4.7
Figure 25: Variation in percentage runoff with a changing leaf area index using typical
Scottish catchment descriptors
landuse _ weight
0.0000268LAI
4.7
summer
0.672
4.7
summer
Equation 6: Potential method for the inclusion of land use effects within the runoff
generation screening process
Equation 6 does not include any effect of soil degradation or the effect of canopy
interception. There is the potential to include the effect of soil degradation via the
analogue degraded HOST class (Packman et al., 2004)48 discussed in Section
4.4.2(C). The effect of canopy interception could be included through reducing the
rainfall as discussed in Section 4.4.2(B), however the effect of canopy interception is
anticipated to be negligible on anything but the smallest rainfall events as canopy
interception is likely to be completely utilised ahead of the hyetograph peak.
This is currently the preferred method due to the speed which it could be
implemented in combination with the useful insights the data could provide. It is
accepted that further work is required to develop a more robust means of including
land use effects and topography.
5. Using a distributed PDM model (modified version of the ReFH)
This approach could estimate the total runoff depth at field scale using the modified
ReFH losses model as presented in Section 4.4.2(C). This model would facilitate the
inclusion of soil type, land cover, topography and rainfall depth within the
assessment. It would be necessary to undertake the assessment for a standardised
rainfall return period and duration, this could be based on a carefully selected return
period and duration (i.e. 3hr 1 in 50yr rainfall) requiring the calculation of a national
grid of values.
Importantly the method would permit the existing runoff generation scenario to be
compared to an idealised land cover scenario thus facilitating an insight into the
value of altering land use within that area. This approach could also used to assess
the effect of degradation linking to the requirements of Section 19.
-95-
While BFIHOST can be calculated from vector HOST data to approaching field scale
resolutions via the method laid out in IH126 (Boorman et al., 1995)137 generating a
grid of PROPWET values to a similar resolution would be more problematic. Section
4.4.2(D) presents a number of potential means of improving the spatial resolution for
estimates of PROPWET . It is noted that the small power applied to PROPWET (0.24) has the effect of largely negating its influence on CMax as the range of
PROPWET across Scotland typically varies from 0.5 to 0.8 as shown in Figure 26.
As demonstrated within Figure 27 SAAR can be used as a reasonable predictor
of PROPWET , hence Equation 28 could be replaced with Equation 7 without a
detrimental loss of accuracy.
0.24
PROPWET
Figure 26: Variation
in PROPWET across
northern Britain
223
(Bayliss, 1999)
-96-
Figure 27: Variation of PROPWET with SAAR (Data for HiFlows catchments)
The SEPA Wetlands Inventory (SEPA, 2012) presents a useful spatial database of all
known wetland habitats in Scotland. It is based on the accumulation of a range of
data sets and wetlands are subdivided into classes based on wetland type. This data
set represents a useful resource for a national overview of wetlands in addition to
more site specific information. It is recommended that consideration is given to
whether an element of this database could be presented alongside this screening
process. In presenting this data it is important to acknowledge that the analogy of
wetlands as sponges is technically incorrect as presented in Sections 3.3(F) and
3.3(H).
(B)
-97-
Qn
y
5
3
1 3
1
1
SP
5
S
S
S6
Equation 8: Proposed equation for generating a measure of indicative Storage
Potential
This is not the preferred method, as it overlooks existing landuse and therefore would
require an alternate method for Section 19.
2. Increased Storage Potential based on hydraulic slope and the current landuse
roughness
By rearranging Mannings Equation to the form of a wide channel (where channel
width equals wetted perimeter) and constraining flow to 0.1cumecs/m (as it is
unknown), setting the maximum floodplain roughness (possible range 0.075-0.2) and
estimating the existing land roughness based on land coverage, the Increased
Storage Potential can be defined as detailed in Equation 9 based on the estimated
hydraulic gradient ( S , m/m).
5
3
Qnexisting 3
Qn
y max
S
n
ISP max
S
0.1nmax 0.1nexisting
nexisting
5
S
S6
5
3
5
3
5
3
This method will enable a consistent screening method to be used in Section 20 and
Section 19. Equation 10 has been presented in case it is desirable to reduce the
complexity of Equation 9.
ISP
nexisting
max
-98-
Land capability maps (forestry and agriculture) could be used to inform the selection
of maximum roughness values to improve the quality of the estimation of the
capability to store water at a given location, however the existing land cover and land
capability must be of comparable qualities to avoid the introduction of uncertainty.
3. Increased Storage Potential based on hydraulic slope, flow velocity and the
current landuse roughness
This method builds on the previous screening method but introduces the use of flow
velocity which can be based on the data provided by the national flood mapping
study. It is necessary to use the estimated velocity and slope for both the existing
and tested scenario. Equation 11 presents the proposed method.
3
2
3
2
max
3
4
v n
S
3
2
v n
3
2
existing
3
S4
3
2
3
3
v 2
2
4
S
HV T p QMed
Equation 12: Proposed equation for generating a measure of indicative Hydrograph
Volume
Hence the Storage Potential (or Increased Storage Potential) score could be divided
by the indicative Hydrograph Volume to generate a Storage Value ( SV ) score as
detailed in Equation 13.
SV
SP
1
HV ST p QMed
-99-
The use of QMed and T p from the nearest watercourse may invalidate this method
as it would be more applicable to use the QMed and T p values for the location that
experiences the flood, therefore the method is not preferred however if time permits
there may be merit in undertaking additional study in this area.
It would be beneficial to capture knowledge of the morphological condition of the river
channel as an enlarged channel is likely to have a reduced connection to its
floodplain. While it is attractive to attempt to somehow weight the above Storage
Potential score using the morphological condition of the watercourse, this would
place an overemphasis on the quality of the morphological data. It is proposed that
selected elements of the Morphological Pressures Database (channel realignments
and embankments) are laid over the above score so that the user can quickly link the
potential to store water in that area with the morphological condition of the
watercourse.
The SEPA Wetlands Inventory (SEPA, 2012) presents a useful spatial database of all
known wetland habitats in Scotland. It is based on the accumulation of a range of
data sets and wetlands are subdivided into classes based on wetland type. This data
set represents a useful resource for a national overview of wetlands in addition to
more site specific information. It is recommended that consideration is given to
whether an element of this database could be presented alongside this screening
process to assist in the identification of areas which have previously been classified
as having habitat types which are indicative of good channel-floodplain connectivity.
Consideration should be given to the likely backwater extent generated by floodplain
roughening activities (and other activities which might raise water levels) so that
upstream receptors do not experience an increase in flood risk due to elevated water
levels. It may be feasible to relate the likely extent of a backwater based on the
valley slope however this will require numerical analysis of the backwater extent
equation (Appendix C).
(C)
T p ,min
TS
T
p ,max
Equation 14: Proposed method for identifying the Time to Peak Similarity of
catchments
-100-
A similar means could be used to quantify a second score to represent the value of
desynchronisation by replacing T p with the estimated Median Annual Maximum
Flood ( QMed ).
The largest challenge will be displaying the data in a concise and coherent manner.
One potential means may be to present the watercourse network as polylines and to
adjust the colour of the lines upstream of confluences depending on whether it has
the longer of shorter time to peak. At the junction a point of variable colour could be
used to visually present the Time to peak Similarity Score ( TS ) with the size of the
point being based on the Median Annual Maximum Flood Score ( QS ).
TS=0.96
-101-
(D)
The aim of this screening process is to identify potential locations where the removal
of unnatural hydraulic constrictions might achieve a local reduction in flood risk. A
constriction might be caused by a confined river corridor, culvert, bridge or weir. The
restoration of the river reach or the replacement with a higher capacity structure
might offer a significant reduction in flood risk to neighbouring flood receptors.
One means of identifying conveyance improvements could be by overlaying the
latest fluvial flood maps with the locations of hydraulic structures identified within
SEPAs Morphological Pressures Database. However this approach will provide no
further information beyond simply overlaying an Ordnance Survey map with the
mapped flood extent as the Morphological Pressures Database was generated using
OS Master Map data.
A more relevant means of undertaking this screening process would require the
national fluvial flood mapping to be undertaken using an additional alternate scenario
representing a naturalised catchment case. The naturalised scenario would be
schematised by removing all hydraulic structures. This would facilitate a comparison
of the with and without hydraulic structures flood extents or flood depths to identify
areas that would benefit from the removal of hydraulic structures. It is understood
that the Section 19 assessment of the role existing artificial structures play in
reducing flood risk will already be undertaking a very similar assessment, therefore
this screening process should be aligned with the method adopted for Section 19.
(E)
-102-
The first issue of model results is programmed for summer 2012. It is recommended
that once data is available it is reviewed to ensure that it does meet the aim of this
screening process.
(F)
The aim of this screening process is to identify areas where coastal realignment
could be undertaken for the purpose of reducing estuarine surge.
It is proposed that this screening process would be based on identifying land which
could be allowed to flood during a tidal surge event. It will be assumed that the
benefit offered will be directly proportional to the depth of water during a coastal flood
event. That is, the Surge Attenuation Potential ( SAP ) will be the estimated 200yr
water level ( y 200 ) minus the local ground level ( Z ) and the mean sea level ( y ) as
shown in Equation 15.
SAP y 200 y Z
Equation 15: Estimation of the Surge Attenuation Potential
-103-
The location of known direct coastal defences, as identified by Section 19, will be
presented to facilitate the visual identification of areas which could potentially be
allowed to flood in the future following a managed realignment. There is no
consistent national data set detailing the condition and height of coastal defences
therefore it will not be possible to identify land which is currently protected within the
national screening process.
The SEPA Wetlands Inventory (SEPA, 2012) presents a useful spatial database of all
known wetland habitats in Scotland. It is based on the accumulation of a range of
data sets and wetlands are subdivided into classes based on wetland type. This data
set represents a useful resource for a national overview of wetlands in addition to
more site specific information. It is recommended that consideration is given to
whether an element of this database could be presented alongside this screening
process to assist in the identification of salt habitats which are already likely to
experience coastal inundation.
The location of artificial drainage ditches or channels which may also serve to
accelerate water getting onto tidal marshes will also be presented as part of this
screening. The assessment of the effect these flow conduits might have on the
movement of surge water is anticipated to be very complex requiring the use of a
hydraulic model. For this reason the effect of flow conduits will not be included in the
screening output, but the presence of conduits will be presented.
(G)
The aim of this screening process is to identify areas where saltmarshes or other
measures such as artificial reefs could be used to reduce the wave energy arriving at
the shore. The attenuation of wave energy could be used to supplement existing
direct defences to either extend their lifespan or to increase the standard of
protection that they provide.
It is proposed that this screening process would be based on estimating the Incident
Wave Power ( IWP ) using Equation 16 from knowledge of the significant wave height
( H s ) and the representative wave period ( T ). The best source of wave climate data
identified to date is the Atlas of UK Marine Renewable Energy Resources (DTI,
2008)211 which provides higher resolution than Hydrodynamics and Sea Level Rise
(European Environment Agency, 2005). This data is freely available and details the
mean annual wave power for non-estuarine coastal waters.
-104-
IWP
g 2
0.5H S2T 0.5H S2T
64
The Space for Attenuation ( SA ) of wave energy could be estimated based on the
intertidal slope which, to avoid the problematic calculation of local tide levels, could
be taken as the horizontal distance between Mean High Water Springs and Mean
High Water Springs ( x ).
SA x MHWS MLWS
Equation 17: Space for Wave Attenuation
Subsequently the Capability of Wave Attenuation ( CWA ) could be estimated for the
entire coastline by multiplying the Incident Wave Power by the available Space for
Attenuation as shown in Equation 18.
CWA IWP SA
Equation 18: Capability of Wave Attenuation
This is the preferred means of undertaking the wave attenuation screening process
due to its simplicity.
Figure 31: Annual mean
significant wave power
(kW/m of wave crest). Data
from the Atlas of UK Marine
Renewable Energy
Resources, Commissioned
by DTI and produced by
ABPmer, The Met Office and
Proudman Oceanographc
Laboratory (2008,
www.renewables-atlas.info/)
-105-
A more detailed means of calculating the wave run-up or defence overtopping could
be undertaken using the guidance contained within the CIRIA Rock Manual (CIRIA,
2007)212. While potentially offering more accurate output the methods also requires
additional data. It is felt that an attempt to quantify overtopping would be misguided
and would give the impression of a greater level of knowledge than is actually
available. Therefore it is proposed that the estimation of wave run-up would be a
more valuable descriptor to be used for the screening process. According to the
C683 the wave run-up can be estimated to within 2% by Equation 19 with knowledge
of the significant wave height ( H m 0 ), Iribarren Number using the energy wave period
( m 1.0 ) and a friction factor ( f ).
Ru 2%
2.5H m 0 f
m 1,0
The Iribarren Number can be estimated using the shore slope ( ) and the fictitious
wave steepness ( S 0 ) as shown in Equation 20.
m 1,0
tan
S0
The fictitious wave steepness can be calculated using Equation 21 based on the
significant wave height ( H m 0 ) and the energy wave period ( Tm,1, 0 ).
S0
2H m 0
gTm21, 0
Shore type
Friction factor ( f )
1.0
Pitched stone
0.80-0.95
0.7
0.55
The calculation of wave run-up is a data intensive and complex process. The
additional complexity associated with the method will not bring a notable
improvement in the quality of the screening process and therefore it is not preferred
over the Capability of Wave Attenuation score presented in Equation 18.
It should be acknowledged that the Flood Risk Management (Scotland) Act makes no
reference to erosion therefore the inclusion of a measure of potential coastal erosion
-106-
-107-
4.3.3 Summary
The methods outlined in Table 8 are recommended as the most valid for undertaking
the Natural Flood Management screening process.
Preferred Method
Data requirements
q 8.3062 0.1536
1000
SAAR
0.0460
BFIHOST 2
3
2
3
3
v 2
2
S4
Areas of wave energy attenuation potential: (to be Mean high and low water springs OS
confirmed)
Master Map
Wave Power and the Space for Wave Attenuation
Table 9 outlines the screening method that will not be undertaken due to low
scientific uncertainty in the underlying Natural Flood Management method.
-108-
Method
Data requirements
T p ,min
TS
T
p ,max
Table 9: Summary of the screening processes which are not recommended without
further scientific support or data
-109-
Capability of existing
assessment techniques
Conifer woodland
No existing tools
Weak literature base in place
No existing tools
Very weak literature base in place
No existing tools
Very weak literature base in place
No existing tools
Weak literature base in place
No existing tools
Weak literature base in place
Localised effects can be estimated
using existing hydraulic models
when used in conjunction with
existing hydrological techniques
Localised effects can be estimated
using existing hydraulic models
when used in conjunction with
existing hydrological techniques
Localised effects can be estimated
using existing hydraulic models
when used in conjunction with
existing hydrological techniques
Broadleaf woodland
Good muirburn practice
(compliance with muirburn
code)
Reducing grazing pressure
on pasture
Creation of cross slope tree
shelter belts
Creation / restoration of nonfloodplain wetlands
Reducing soil compaction in
arable areas, improving soil
texture and reducing bare
earth in wetter seasons
Changing agricultural field
drainage
Upland drain blocking
Floodplain reconnection
Creation of washlands
-110-
Natural Flood
Management measure
Capability of existing
assessment techniques
Afforestation of floodplains
Use of SUDS
Managing channel
instabilities
-111-
Natural Flood
Management measure
Capability of existing
assessment techniques
Table 10: Summary of assessment methods for Natural Flood Management measures
and the identification of areas where there are currently gaps in assessment capability
In summary there are adequate assessment techniques in place for the quantification
of coastal, groundwater and urban Natural Flood Management measures. However
there are the following caveats to this conclusion:
Unfortunately there is a large gap in the assessment methodologies that can be used
for quantifying the hydrological benefits of fluvial Natural Flood Management. To
allow fluvial measures to be compared against more traditional engineered it will be
necessary to develop a methodology which can close this gap. A proposed
hydrological assessment methodology to do this is presented in Section 4.4.2.
-112-
-113-
Rain storms
assumed to be
standard FSR storm
profile and
instantaneously
cover the entire
catchment
Figure 32: Pictorial overview of the proposed Natural Flood Management assessment
model
-114-
Rainfall
Evapotranspiration
FEH DDF
MORECS
Interception
Land use
Antecedent
moisture
Based on modified
ReFH estimation of
Cini by modifying
PROPWET with
land use.
Based on MORECS
leaf area index
Drainage network
Baseflow model
Soil type
Based on ReFH
estimate of the initial
baseflow (no variation
during event)
HOST (and
degraded HOST)
OUTFLOW
(A)
Rainfall estimation
The design rainfall for storms will be derived in a similar manner to the ReFH and
FSR/FEH rainfall-runoff models using the Depth Duration Frequency (DDF) model as
presented by Faulkner (1999)214. The Area Reduction Factor (ARF) used to transfer
a design point rainfall depth to a catchment representative depth would also be a
logical inclusion. It is understood that SEPA already has access to a GIS toolbox
and parameter set to facilitate point rainfall estimates for varying return periods and
storm durations. It is anticipated that this toolbox can at least be included in the build
of the hydrological model to be used internally by SEPA. Outside SEPA, other
organisations may need to secure appropriate licences or revert to a manual query of
the FEH CD-Rom.
The ReFH introduced a Seasonal Correction Factor (SCF) to improve the models
calibration to the statistical analysis of flow records. It should be noted that the SCF
is a calibration coefficient which is specific to the model, following testing of the
-115-
model it may become apparent that a similar calibration coefficient may be required.
It is not recommended that a SCF is included within the initial models construction.
It should be noted that the DDF, ARF and Design Storm Profiles are generalisations
of complex systems. Kjeldsen (2007)50 provides an overview of some of the
limitations of these generalisations, and it should be remembered throughout that
these methods provide a means of making informed estimates, not a quantification of
the actual amount, and that these estimates are commonly accepted in the absence
of alternatives.
Following the derivation of the design storm rainfall depth grid (and after subtraction
of the canopy interception as presented in (B)) it is proposed to use the same 75%
winter and 50% summer rainfall profiles originally used in the FSR/FEH rainfall-runoff
method. Both are symmetric single peak profiles, and their shape does not vary
between storms (spatially, event magnitude or duration). As the tool will be applied
to predominately rural catchments (URBEXT <0.125), then the inclusion of the winter
profile will be most important within the tool (Institute of Hydrology, 1979)215.
Figure 34: Design rainfall
profiles for summer and
winter, drawn as normalised
hyetographs (Kjeldsen,
50
2007)
-116-
SAAR
D T p 1
1000
(B)
Interception model
As presented in Section 3.3, the literature indicates that the interception of rainfall by
the vegetation canopy is a process that reduces the runoff generation and
subsequent magnitude of flood flows. It is proposed that a basic canopy interception
model is included within the model. A simple model based on the Leaf Area Index
( LAI ) is used within the MORECS model (Hough and Jones, 1997)216. The
MORECS system places an upper threshold on daily interception ( I ) as 0.2mm per
unit of leaf area index as defined by Equation 23.
I 0.2LAI
Equation 23: Maximum daily canopy interception
A more complex means of estimating the canopy interception capacity is to use the
method proposed by von Hoyningen-Huene (1981)105 as presented in Equation 24.
Land cover
Summer LAI
Winter LAI
Grass (General)
Improved grassland
4*
1.5*
5*
2*
Cereals
0 (Bare earth)
0.5 (Winter cover)
Oil-seed rape
0 (Bare earth)
0.5 (Winter cover)
Potatoes
0 (Bare earth)
Sugar beet
0 (Bare earth)
Deciduous trees
0 (Bare earth)
Conifers
Orchards
Upland (unimproved
grassland)
3.5
2*
Upland (heather)
3.5
3.5*
Table 11: Leaf Area Index values as used in MORECS 2.0 (Hough and Jones, 1997)
(* Are inferred from the available data)
-117-
216
A comparison of the two methods for estimating the canopy interception capacity is
presented in Figure 35. This shows that the Hoyningen-Huene (1981)105 method
produces considerably higher estimates of the canopy interception capacity. It is
noted however that Hoyningen-Huene (1981)105 describes this as the maximum
capacity, whereas the threshold used within MORECS (Hough and Jones, 1997)217
was intended to be representative of a typical daily maximum. It is considered
probable that the true answer for the interception capacity during storm conditions
lies somewhere between the two methods. As the MORECS method was used by
the Met Office to represent typical conditions found in Britain after a robust model
development process, and because it produces a more conservative result, this
method is recommended for initial adoption within the method. In adopting this
assumption it should be noted that work by Calder (1990)94 suggests that canopy
interception by coniferous forest (LAI=6 to 10) can be up to 6 to 7mm/day. While the
variation between the two methods to the face of it appear very notable, it should be
remembered that the total interception during a flood event will be relatively small,
less than 10% of the total event rainfall (Nisbet, 2008)218 tied to the consideration that
flooding tends to occur during leafless months. In addition, this interception will occur
at the start of the event leading to the suggestion that canopy interception will have
little effect on the flood peak.
6
0
0
10
Hoyningen-Huene (1981)
Figure 35: A comparison of the variation in canopy interception capacity with leaf area
index (LAI) using MORECS (Hough and Jones, 1997) and Hoyningen-Huene (1981)
For simplicity it is proposed that the default maximum daily canopy interception will
be taken to equal the maximum interception for a single event (as calculated using
MORECS). It is recommended that the Hoyningen-Huene (1981)105 method is
included as a user selectable alternative to permit upper bound sensitivity testing,
however the MORECS method should be the default.
The calculated interception will be subtracted from the gross rainfall at the start of the
hyetograph. Once the accumulative gross rainfall exceeds the canopy interception
capacity no further rainfall interception will occur. The intercepted rainfall will be lost
from the model with no further consideration of evaporation etc from the interception
store. It is accepted that this approach may underestimate the effect of canopy
interception at the hyetograph peak and over estimate the effect at the start of the
-118-
(C)
The role of the runoff generation or loss model is to estimate the proportion of rainfall
that will be converted to direct runoff, with the remainder being passed to soil
moisture. It is proposed that it will be based on that used in the ReFH model
(Kjeldsen, 2007)50 which is in turn based on the Probability Distributed Model (PDM)
developed by Moore (1985, 2007)219220. PDM is widely used for hydrological
applications across the UK, including in SEPAs Flood Early Warning System as well
as investigating the impact of climate change.
Conceptually, the PDM assumes the catchment consists of a number of individual
storage elements, each of a soil moisture capacity ( C ) arising from a statistical
distribution. Assuming a uniform distribution of soil moisture capacities, if the storage
elements are arranged in order from the highest ( CMax ) down to zero, the resulting
PDM distribution of soil moisture capacity is shown in Figure 36. It is further
assumed that storage elements interact so that soil moisture is constant for all
elements of capacity greater than ( Ct ), and full for those which are smaller. Incident
rainfall increases the depth of water in each storage element and when rainfall
exceeds the storage capacity, direct runoff is generated. For the short duration of
storms being considered, the effects of evapotranspiration and drainage out of the
soils will not be included.
Figure 36: Equal water content ( C ) across stores of different capacity (Kjeldsen,
50
2007)
Therefore, a pulse of rain ( P ), on the soil gives 100% runoff from the stores which
are full and increases the moisture content in all other areas. The excess amount of
rainfall converted into direct runoff ( q ) can be estimated using Equation 25.
-119-
PC
P2
CMax 2CMax
( C CMax )
( C CMax )
qP
By dividing the rainfall series into timesteps the runoff at any timestep ( t ) can be
calculated using Equation 26.
Pt Ct 1
Pt 2
qt
CMax 2CMax
( Ct CMax )
qt Pt
( Ct CMax )
And from continuity the stored soil moisture can be calculated using Equation 27.
Ct Ct 1 Pt
Equation 27: Calculation of soil moisture based on previous soil moisture and incident
rainfall
While BFIHOST can be calculated from vector HOST data to approaching field scale
resolutions via the method laid out in IH126 (Boorman et al., 1995)137 generating a
grid of PROPWET values to a similar resolution is more problematic. Section (D)
presents a number of potential means of improving the spatial resolution for
estimates of PROPWET .
It is envisaged that the runoff generation model will be constructed on a 25-100m
grid, it is important that this grid is representative of the resolution of the land cover
and soil data. It is not thought that there is a strict requirement for the runoff
generation grid to be the same as that used for the flow routing, however a variation
in grids could lead to unforeseen errors.
A method for modifying the BFIHOST to account for the effect of soil degradation
was proposed by Packman et al (2004)48. The rationale for the method is that
degraded soils suffer from structural degradation, compaction of the topsoil and
-120-
upper subsoil and seasonal capping. Together these effects reduce the effect soil
moisture capacity leading to increased runoff generation. A central theme of the
method is that the hydrogeological component remains constant while the soil
component is degraded by a class. Kjeldsen (2007)50 speculates that the method
proposed by Packman et al (2004)48 is a potential means of representing the effect of
soil degradation within the ReFH model which is the underlying basis of this loss
model. The analogue HOST classes proposed by Packman et al (2004)48 are
presented in Table 12.
HOST Class
Analogue HOST
for degraded soils
Un-degraded BFI
Degraded BFI
1.00
0.90
1.00
0.90
7?
0.90
0.79
6?
0.79
0.65
7?
0.9
0.79
0.64
0.56
7?
0.79
8 level ground
0.56
Na level ground
9 level ground
0.73
Na level ground
10
10 level ground
0.52
Na level ground
11
11 level ground
0.93
Na level ground
12
12
0.17
NA
13
3?
1.00
0.90
14
24
0.38
0.31
15
15?
0.38
16
18?
0.78
0.60
17
18
0.61
0.52
18
20
0.52
0.40
19
22
0.47
0.32
20
20?
0.52
21
23
0.34
0.22
22
27
0.32
0.26
23
25
0.22
0.17
24
25
0.31
0.21
25
25?
0.17
26
26
0.24
NA
27
27?
0.26
NA
28
0.58
NA
29
0.23
NA
Table 12: Effect of soil degradation via the use of analogue HOST soil types (Packman
48
et al, 2004)
There is the option to reduce the maximum soil moisture storage capacity ( CMax ) to
account for variations in topographic slope following the method presented by Bell
and Moore (1998)221 and later used in the G2G model Bell et al. (2007)222. The
reduction in maximum soil moisture capacity ( CMax ) is on the basis that steep
topographic slopes limit soil moisture to shallow levels, whereas flat areas can permit
the build up of deep soil moisture stores. It is unclear how much allowance
topographic slope is inbuilt within the method of estimating CMax based on HOST
-121-
data and therefore it is recommended that the variation in slope is not included until
after an initial modelling testing period. Equation 29 shows the potential basis of
reducing CMax based on the ratio of average grid cell topographic slope ( g ) over the
gradient at which the soil moisture storage capacity would reach zero ( g Max ). The
maximum gradient ( g Max ) varies between soil types and is related to the depth of the
soil.
g
C Max
C Max,rev 1
g
max
Equation 29: Method of reducing the soil moisture capacity based on local
221
topographic slope (Bell and Moore, 1998)
The calculation of the initial soil moisture content C ini is presented in Section 4.4.3(D)
below.
(D)
Following the ReFH the initial soil moisture content ( C ini ) for a winter storm can be
estimated using Equation 30.
C ini
C Max
1.20 1.70BFIHOST 0.82PROPWET
2
-122-
Value
Min
25%
Median
75%
Max
SAAR
762
1016
1334
1649
2848
PE
428.5
474.9
489.2
511.0
556.6
LAI Summer
1.917
3.656
3.943
4.330
4.911
rsc
60.699
69.521
70.769
74.607
94.006
BFIHOST
0.292
0.369
0.411
0.469
0.609
PROPWET (FEH)
0.400
0.565
0.630
0.715
0.850
Table 13: The range and quartiles for variables used in the PROPWET
descriptor regression analysis
-123-
Regression PROPWET
0.70
0.60
0.50
y = x - 5E-09
R2 = 0.8276
0.40
0.30
0.20
0.10
0.00
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
FEH PROPWET
Figure 37: Plot comparing the regression based estimate of PROPWET with the
FEH CD PROPWET
-124-
Figure 38: Plot comparing the regression based estimate of PROPWET with the
FEH CD PROPWET
3. The most detailed means would be to gain access to the original Soil
Moisture Deficit Data generated using MORECS for the 1961 to 1990 period
for all land cover types and to reanalyse it to produce estimates of
PROPWET for each land cover type on the 40km grid. These estimates
could then be weighted based on the scenarios land cover to produce higher
resolution estimates of PROPWET as part of the hydrological modelling
process.
4. Consideration has been given to the use of soil moisture data estimated using
MOSES (Hough, 2003)224 to in turn estimate PROPWET, however it was
decided that the different output of MOSES would result in a non-comparable
PROPWET. This would lead to the need for alternative Cini and Cmax
estimator equations needing to be developed.
-125-
(E)
The ReFH model uses an instantaneous unit hydrograph approach which is founded
on the estimation of the catchments time to peak. For observed catchments the time
to peak can be estimated by inspection of observed flow and rainfall records,
however for ungauged catchments it is necessary to use a regression based
estimate of the time to peak. Unfortunately neither of these approaches allow for the
effect of changes in the catchments drainage network (i.e. increase in length by
meandering or increase in roughness by riparian planting) or the variation in time of
contribution for different regions of the catchment to the downstream hydrograph.
It is understood that G2G undertakes flow routing by estimating the velocity and
distance water takes as it passes through the catchment. From these estimates an
indication of the time of travel can be inferred. It is understood the model uses
constant channel and land flow velocities which can then be varied in order to
calibrate the timing of the runoff peak within the hydrological model where observed
data is available. The use of spatially constant velocities will prevent a meaningful
estimate of the impact of changes to the catchments drainage network (i.e. increase
in length by meandering or increase in roughness by riparian planting).
OVERFLOW and the Allan Water Natural Flood Management study method both
estimate the velocity of overland flow to the watercourses based on a hydraulic
calculation, hence the velocity is dependent on topographic slope, runoff rate and
ground cover roughness. For in channel flows OVERFLOW uses a more advanced
diffuse wave hydraulic routine which allows for flood water to be spilled onto the
floodplain, while the Allan Water Natural Flood Management method uses a steady
state kinematic wave routine. It should be noted that the simplistic kinematic wave
theory is built on the assumption of all flow being at normal depth and it therefore
prohibits the assessment of backwater effects. Both methods assume that the
channel dimensions are based on approximated dimensions based on an index
flood. The OVERFLOW model assumes that watercourses form when the flow
exceeds a threshold (0.6cumecs) whereas the Allan Water Natural Flood
Management method uses mapped watercourse network or a threshold catchment
area. The flow routing engine used by OVERFLOW is computationally expensive,
and the one employed by the Allan Water Natural Flood Management study should
be viewed as halfway between G2G and OVERFLOW.
-126-
Land flow
velocity
Channel flow
velocity
Computational
expense
Suitability
ReFH
NA
regression
based
NA
regression
based
Very low
Gauged catchments
which are not subject
to changes
G2G
Spatially and
temporally
constant
velocity
Spatially and
temporally
constant
velocity (1D)
Low
Gauged catchments
which are not subject
to land use changes
Allan Water
Natural Flood
Management
study
Steady state
kinematic
wave (normal
depth)
Steady state
kinematic wave
(normal depth)
(1D)
Medium
Ungauged catchments
with (limited) in channel
and diffuse land use
changes
OVERFLOW
Time varying
velocity based
on kinematic
wave
hydraulics
Time varying
velocity based
on diffuse wave
hydraulics (2D)
High
Ungauged catchments
with floodplain, in
channel and diffuse
land use changes
Research
institution
distributed
models e.g. SHE
Full 2D St
Venants
Equations
Full 2D St
Venants
Equations
Very high
Ungauged catchments
with floodplain, in
channel and diffuse
land use changes
Table 14: Summary of flow routing used within existing hydrological models and
methods
It is recommended that an approach similar to that used in the Allan Water Natural
Flood Management study is included within the initial S20 assessment procedure. It
is felt that this approach offers a good balance between computational expense and
functionality by enabling the accumulation of effects from diffuse land use changes
and the limited assessment of in-channel interventions. However a flexible approach
should be adopted to allow either a decrease or increase in computational expense
by permitting the use of user supplied constant velocities (G2G) or a more detailed
computational engine.
It is proposed that a single time to outlet grid is calculated for the catchment which
will be representative for all flood return periods, in a similar fashion to how the ReFH
relies on a single time to peak for all return periods. The approach of using a single
time to outlet grid is described by Beven (2012)203 as a linearity assumption, and he
reports that the method has been found to work surprisingly well with any
inaccuracies due to the linearity assumption being generally less than those
associated with the estimation of runoff. This time to outlet grid can be recalculated
for different Natural Flood Management scenarios. It is envisaged that the time to
outlet grid would be calculated at a grid resolution of 25-200m depending on data
availability, computation constraints and catchment size. A flow chart summarising
the proposed method is shown in Figure 42. An example of a time to outlet grid for a
sample Scottish catchment is shown in Figure 39.
-127-
Figure 39: An example of a time to outlet grid generated for a sample Scottish
catchment as part of this study
The time to outlet grid would be calculated for a representative bankfull event, with
the assumption that a bankfull event is equal to the 1 in 2yr flood event. Due to the
requirement for the time to outlet grid to extend to all parts of the catchment, right
down to the scale of a single grid cell, it is proposed to use a 1 in 2yr rainfall event of
a duration equal to the FEH recommended storm duration (Equation 32) which in turn
requires knowledge of the catchments time to peak. At this stage the time to peak
can be estimated using Equation 33.
SAAR
D T p 1
1000
3.34
DPSBAR 0.28
-128-
A user supplied threshold (possibly ~0.1cumecs) will be used to define the transition
between land flow and channel flow. The threshold will be set at a value to achieve a
similar river network extent as seen on high resolution Ordnance Survey mapping
such as the OS 25k. It may be appropriate to supplement the river network by
loading in a mapped artificial drainage network. For all cells which have been
identified as watercourses, a channel width can be calculated using the relationship
identified by Robson and Reed (1999)225 (or similar i.e. Leopold and Maddock
(1953)201) which is based on QMed, as shown in Equation 34. The bankfull flow
estimated earlier will be used as surrogate for QMed.
b 0.182QMed 1.98
Equation 34: Estimation of channel width based on QMed
The next step is to calculate a grid of topographic gradient ( S ) for the entire
catchment using the updated ground surface model. It is recommended that a
smoothing process is undertaken along watercourses to overcome adverse slopes.
Using land cover, estimate roughness values ( n ) for the land flow and channel flow,
this will result in the generation of two different grids with the land flow roughness
being generally higher than channel flow. It should be noted that the land flow
roughness should be representative of the impedance that flow will experience in that
locality, be it as overland flow or shallow groundwater flow. Recognition of local
influences on flow path length should also be given consideration to (i.e. the direction
of tillage). The Mannings roughnesses presented for channels and floodplains will
not be applicable for the land flow. Although not directly applicable, some guidance
on the selection of shallow overland roughnesses is provided by USDA (2010)226, as
summarised in Figure 40. It should be noted that shallow overland flow considered
by the USDA does not include shallow groundwater flow and therefore its validity
may be questionable.
Figure 40: Roughness values recommended by USDA (2010) for shallow surface
runoff for flow depths of less than 30mm.
The roughnesses for river channel should take account of the riparian vegetation and
can theoretically be selected from published values for channels and floodplains.
-129-
Low
Medium
Improved pasture
High
Semi-improved pasture
Very high
Table 15: Examples of land cover types and the roughness bands which could be
employed
Now there is sufficient data to estimate the normal depth ( y n ) using Mannings
equation, and hence the velocity for the land and channel flows. For the land flow it
is assumed that the flow is spread evenly over the whole width of the grid cell so that
the wetted perimeter ( P ) is taken to equal the analysis cell size. For the channel
flow it is assumed that P is the bankfull channel width ( b ). The slope ( S ) would be
divided by a universal sinuosity factor for both the land ( SFLand ) and channel flow
( SFChannel ) so that the effect of assessing flow path lengths on a coarse grid does not
result in an overestimation of the slope. There is scope for the channel sinuosity
data set held by SEPAs Hydromorphology team to be used to inform local channel
sinuosity, this data could be overwritten should a channel planform restoration project
be considered.
5
3
1 A
n P 23
3
QnP
S
SF
v
Q
S
SF
3
5
Equation 35: Use of Mannings equation to calculate the flow velocity during normal
depth flow conditions
The calculated land and channel velocity grids can then be merged (mosaiced)
together to form a single flow velocity grid for the entire catchment. The final step is
to calculate the travel time for each grid cell ( t ) based on the calculated flow velocity
( v ) and the grid size ( l ) multiplied by the relevant sinuosity factor ( SF ). The
estimated grid cell travel times can then be accumulated working upstream, so that
the time to outlet for each grid cell can then calculated.
-130-
Calibration of the flow routing model and the application to real catchments:
It is not a priority for the initial model, however the lag effect of reservoirs and
lochs could be included via a modification to the time to outlet grid should it
be required.
Initial testing of the flow routing model has been undertaken for a collection of
Scottish gauged catchment with reported observed catchment lag times in the Flood
Estimation Handbook vol 4 (1999)227. The results of this initial testing are presented
in Appendix E.
Although not implicitly within the proposed method, there is scope to create a time to
outlet grid by any means that the hydrologist feels appropriate for the catchment in
question. For an initial investigation it may be decided that it is appropriate to use
spatially constant estimates for channel and land flow velocity as assumed within
G2G. Conversely, where there may be attenuation from a reservoir or loch, it may be
seen as appropriate to calculate a lag time for the waterbody and add that lag (as a
constant) to the catchment situated upstream of the waterbody. In a similar fashion a
time lag for a floodplain could be assessed. This time lag could be calculated by
assessing the flood peak travel time using a localised 1D-2D hydraulic model of the
area where the changes will be made, or it could be estimated via a much more
simplistic means depending on what is appropriate to the circumstances. Although
the proposed method is based on a single time to outlet grid for all flood return
periods, this assumption may be questionable for the lag effect generated by open
waterbodies and could be replaced with an alternate approximation as the
hydrologist sees fit. An example of how a time lag effect could be added to the time
to outlet grid is shown in Figure 41Error! Reference source not found..
-131-
Lag time
0.1
Flow
0.08
0.06
0.04
0.02
Time
Figure 41: Example of how the time lag effects of reservoirs, lochs and floodplains
could be included within the assessment method
-132-
RAW DTM
SAAR GRID
DRAINAGE SHAPEFILE
FEH DDF
SAAR
D T p 1
1000
3.34
DPSBAR 0.28
Calculate grid of
topographic
gradients
Using the
hydrologically
corrected DTM,
calculate a grid of
topographic
gradients
CHANNEL ROUGHNESS
Based on presence of
riparian vegetation
LAND FLOW
ROUGHNESS GRID
CHANNEL
SINUOSITY
UNIVERSAL LAND
FLOW SINUOSITY
Calculate grid of land
flow velocities
Time to outlet grid
Using Mannings
-133-
(F)
Baseflow model
It is anticipated that baseflow will not form a significant component of flood flows
within typical Scottish catchments, it will be the responsibility of the hydrologist to
verify this assumption prior to the application of this method. It is proposed to adopt
a simplistic approach for baseflow by using a constant flow rate. In the absence of
observed that the estimate of initial baseflow applied within the ReFH as detailed in
Equation 36 or the FEH Rainfall Runoff Method detailed in Equation 37.
As discussed in Section (A) the winter event has been adopted as it is envisaged
that the Natural Flood Management assessment tool will be predominately used for
rural catchments.
(G)
It is proposed that the flood hydrograph will be generated by dividing the time to
outlet grid (as generated in Section 4.4.2(E)) into a number of bands of equal travel
time, or hydrochrones. Following the principles of the FEH Rainfall-Runoff methods,
it is recommended that an odd number of hydrochrones are used, and seven to
fifteen equal bands may present a good balance between accuracy and
computational expense. The design rainfall depth would be divided into the same
number of bands using the FSR storm profile as discussed in 4.4.2(A) before being
weighted across the catchment and used to estimate the runoff generation for each
grid cell via the loss model (Section 4.4.2(C)) after taking account of canopy
interception (Section 4.4.2(B)). The flow at the catchment outlet for timestep ( Qt )
would then be calculated based on the summation of the area of the hydrochrone
band ( AN ) times the average runoff rate for the band after being delayed by its runoff
band number ( q N ,t N ) using Equation 38. As discussed in Section 4.4.2(F), a
constant baseflow will be assumed throughout the event. It should be noted that
moisture accounting and the subsequent runoff generation via the losses model
would be undertaken for each grid cell and not via the use of average moisture
conditions at the hydrochrone band scale.
Qt AN q N ,t N Qbaseflow
N
N 1
Equation 38: Generation of the flow at the catchment outlet based on the summation
of the delayed runoff from each hydrochrone band
-134-
Figure 43: Example of a five band time to outlet grid for a sample Scottish catchment
700
2
600
4
500
400
10
300
12
Flow (cuemcs)
Band 4
Band 3
Band 2
Band 1
Band 0
Baseflow
Rainfall
200
14
100
16
18
0
10
11
12
13
14
Time step
Figure 44: Example hydrograph using five runoff bands for a generalised catchment
-135-
(H)
Model construction
It is proposed that the model is constructed within GIS as this fits well with the spatial
nature of the data that will be used. To comply with SEPAs existing software
licences the most applicable GIS product is ArcGIS 9.3. The tool could feasibly be
constructed as a series of Toolboxes, however the use of geoprocessing scripts
would bring significant computational efficiencies for the more computationally
expensive processes. The most time consuming process will be the iterative running
of the loss model and to a lesser extent the hydrograph accumulation and normal
depth hydraulics.
The file sizes of the grids associated with the method is an important consideration
as this will dictate the speed of the process and there will be significant efficiencies if
all grids can be held in memory rather than needing to written to the hard drive during
the processing. To be efficient this will require multiple grids to be held in memory at
any one time. Table 16 provides a summary of the memory requirements for a range
of analysis grid resolutions for a sample of catchments.
Catchment size
(bounding grid)
Example
Potential
grid
resolution
32bit raster
grid
5 x 10km
Minor
watercourse
25m
0.33MB
25 x 12km
River Carron
50m
0.95MB
60 x 25km
River Earn
50m
5.84MB
120 x 65km
River Tweed
100m
5.66MB
-136-
Preferred
Alternative
Comments
Topographic
ground
surface
Filtered
LiDAR DTM
NextMap DTM
Rainfall
SAAR 1km
SAAR 5km
FEH DDF
parameters
FEH DDF
parameters
on 1km grid
HOST data
Vector
HOST data
250k
Land
coverage
Vector Land
Cover Map
Potential
evapotranspiration
MORECS
40km grid
It may be possible to
generate a
regression based
equation for PE if
MORECS data cant
be supplied.
Table 17: Summary of the data required for the proposed methodology
-137-
Data
Comments
Mapped river
networks
This data could be used to ensure that flow paths identified within
the routing model are correct. The model can function without it and
it would only be of particular use when channel planform restoration
is being considered.
Mapping of soil
degradation
Mapped
drainage
network
This data can only be used to extend the channel network if the
analysis grid is relatively fine. It could be used to reduce the land
flow roughness or the antecedent moisture condition following
further investigation.
Channel
sinuosity
Table 18: Summary of the data which could improve the quality of the tool
Existing C Max
Existing rsc,summer
Existing nchannel
Existing nland
Existing velocity grids (channel and land combined) for a selection of RMed
durations (3hr, 6hr, 12hr & 24hr) giving potential for the estimation of
intermediate durations via interpolation or extrapolation.
-138-
Means of representation
Not included
Runoff reduction
due to conifer
woodland
Runoff reduction
from deciduous
woodland
Runoff effect of
widespread
muirburn
Reducing soil
degradation due to
overgrazing
Reducing soil
degradation due to
arable land use
Effects of
agricultural
drainage
Effects of upland
drainage
Hydraulic
constrictions
Not included
Table 19: Summary of how a range of Natural Flood Management measures could be
assessed using the proposed methodology
The proposed methodology does not contain a means of estimating the effect of
drainage, be that piped drainage typically found within farmland or open ditches often
-139-
found in upland areas. A discussion of the literature relating to farmland and upland
drainage is provided in Sections 3.3(D) and 3.3(F). In summary, the effects of
drainage is complex with the impact being related to soil type, land slope, drain
slope, drain capacity and an array of other considerations. It currently pushes the
capability of academic plot scale models to make meaningful predictions about small
areas. It may be feasible to develop a means of estimating the impact of drainage on
the antecedent moisture condition via regression, or another means, of the significant
contributing factors. It is likely that considerably more data will be required to
achieve this. As the drainage network is likely to be less than the relatively coarse
analysis grid size proposed by this fluvial assessment methodology the only viable
means to assessing the accelerated runoff pathway that drainage represents could
be by reducing the land covers Mannings roughness value. This modification of
roughness value would need to make consideration of drainage direction (cross or
downslope), drainage capacity, drainage condition, soil type and the impedance to
flow caused by the surrounding land cover. In combination with a modification to the
land roughness it would also be necessary to modify the initial soil moisture and the
soil moisture capacity to account for the reduced antecedent moisture condition and
the reduced capacity to store water.
-140-
Method for including the time lag effect of lochs, reservoirs and floodplains
answer, rather the output should be seen as an approximate guide to what potentially
might happen.
-142-
As presented in Sections 3.3(B) and 3.3(C), the effect of soil compaction has been
identified to be a cause of increased runoff generation by Packman et al (2004)48.
Along with other remedial measures, it is anticipated that the conversion of landuse
to woodland will help reduce soil compaction thus serving to bolster the flood risk
management benefits of woodland.
The ability of vegetation to impede the flow of water over the land and hence
attenuate flood water upstream of flood receptors has been partly included within the
method. The 1D kinematic wave model proposed within the core of the flow routing
method will allow the impact of channel, bank and riparian interventions to be
investigated on a broadscale through the application of Mannings roughnesses. A
method for quantifying the effects of floodplain roughening using woodland has been
presented in Section 4.4.4 using a 1D-2D hydraulic model (or alternative hydraulic
calculation) to quantify the lag time generated by the intervention. The proposed
method for assessing the effect of vegetation impeding the movement of flood water
and hence attenuating flood flows is the same as that applied by previous studies
(increased Mannings roughness). The delay effects will be related to the impedance
that the vegetation creates, with vegetation that has a high Mannings roughness
creating the largest effects. On this basis, lightly managed woodland and short
rotation forestry, might be expected to be good at maximising the attenuation effects.
As discussed in Section Error! Reference source not found., consideration should
also be given to the potential supply of debris to the watercourse which could block
constrictive structures downstream.
4.4.8 Summary
Following the recommendation of DEFRA FD2114 (OConnell et al., 2004)190 a
hydrological assessment methodology has been presented which is:
1. Distributed;
2. Capable of tracking soil moisture conditions;
3. Partly physically based; and
4. Capable of tracking effects downstream.
The method is currently not capable of continuous simulation however the PDM
principles which it is founded on do not preclude it. Although a vast simplification of a
complex set of interrelated processes, the method avoids the need for continuous
simulation by estimating the antecedent moisture conditions using the ReFH same
methodology applied within the ReFH. It is felt that this approach is in the spirit of
what was recommended by DEFRA FD2114 (OConnell et al., 2004)190.
It should be noted that this is a proposed hydrological method and that as of the time
of writing there is no software to facilitate the application of the method. Once
software has been created it will be necessary to calibrate and verify the model.
-143-
Where feasible, existing tools should be used for the assessment of Natural
Flood Management measures as detailed in Table 10. Guidance should be
-144-
The inclusion of physically based flow routing via a time to outlet grid
permits the effects of diffuse land use changes to be accumulated for
a broad range of catchment sizes in addition to quantifying the effect
of in channel modifications
The calibration and verification phase will enable the assessment of model
uncertainty. When quoting results based on this method it is essential that
they are quoted alongside the evaluated uncertainty.
-145-
References
1
Scottish Government, Flood Risk Management (Scotland) Act, Scottish Parliament, June
2009. http://www.legislation.gov.uk/asp/2009/6/pdfs/asp_20090006_en.pdf (Accessed 13
February 2012)
2
SAIFF, What is meant by restoration, enhancement, and alteration under the Flood Risk
Management (Scotland) Act 2009. V2.0, SAIFF policy discussion paper. 2011.
3
Jacobs, Development of Methodology for Assessment Required under Section 20. SEPA,
2011.
4
Werritty A., Arthur S., Ball T. and Wallerstein N., Review of Jacobs Section 20 Review,
CREW-The James Hutton Institute, January 2012
5
SNIFFER & SEPA Natural Flood Management Implementation Learning from Practice
Workshop Report, November 2011
6
Atkins. R&D Update review of the impact of land use and management on flooding.
Environment Agency, 2007.
7
Beven K., Young P., Romanowicz R., FD2120: Analysis of historical data sets to look for
impacts of land use management change on flood generation. DEFRA & Environment
Agency. 2008.
8
Halcrow, Delivery of Making Space for Water; HA6 & HA7: The role of land use and land
management in delivering flood risk management. Environment Agency, 2008.
9
Environment Agency, The Land Use Jigsaw, Prioritising scientific research on land use
management and FCRM. 2008.
www.google.com/documents (insert username: land.use.jigsaw password: evidence)
10
Haycock Associates, Evidence Based Review: Does Land Management Attenuate Runoff?
National Trust, 2008.
11
MNV, The Way Forward for Natural Flood Management in Scotland A report for Scottish
Environment LINK. 2008.
12
Environment Agency, Greater working with natural processes in flood and coastal erosion
risk management, A response to Pitt Review Recommendation 27, Environment Agency,
2012
13
Longfield S.A. Macklin M.G., The influence of recent environmental change on flooding and
sediment fluxes in the Yorkshire Ouse Basin Hydrology. Hydrological Processes 13 105166,1999
14
Conway, VM. Miller, A. The hydrology of some small peat covered catchments in the
Northern Pennines Journal of the Institute of Water Engineers 14 415-24. 1960.
15
Godwin R.J. and Dresser M.L. Review of soil management techniques for water retention in
the Parrett catchment Environment Agency, Almondsbury, Bristol. 2003
17
Davies R., Predation and the Profitability of Grouse Moors. British Wildlife, 16, 339-347.
2005
18
Cox R Parasites and Red Grouse. Are parasites of red grouse becoming resistant? The
Game Conservancy Trust Review, 35, 48-49. 2003
19
Hudson P., Grouse in Space and Time: The population biology of a managed gamebird.
Game Conservancy Ltd, Fordingbridge, Hampshire. 1992
20
Gimingham C.H., An Introduction to Heathland Ecology, Oliver & Boyd, Edinburgh. 1975
-146-
21
Lawton J.H., Red Grouse Populations and Moorland Management, British Ecological
Society, Shrewsbury. 1990
22
Tucker G., Review of the Impacts of Heather and Grassland Burning in the Uplands on
Soils, Hydrology and Biodiversity. English Nature Research Report Number 550. English
Nature Peterborough, 147. 2003.
23
O'Brien H.E., Labadz J.C. and Butcher D.P., An Investigation of the Impact of Prescribed
Moorland Burning in the Derwent Catchment Upon Discolouration of Surface Waters. Report
to Moors for the Future. Nottingham Trent University pp.85 2006
25
Scottish Natural Heritage, Fire: Friend or foe? http://www.snh.gov.uk/land-andsea/managing-the-land/upland-and-moorland/fire-friend-or-foe/ (referenced 25 January 2012)
26
27
OBrien H., Labadz J C & Butcher D P, A review of Blanket Bog Management and
Restoration, Technical Report to DEFRA. CTE0513, Nottingham University, DEFRA, 2007.
28
Lane, SN. Brookes, CJ. Hardy, RJ. Holden, J. James, TD. Kirby, MJ. McDonald, AT. Tayefi,
V. Yu, D. Land management, flooding and environmental risk: new approaches to a very old
question Proc. Of National CIWEM Conference. 2003
29
Mallik A.U., Giminham C.H., Rahman A.A., Ecological effects of heather burning: I. Water
infiltration moisture retention and porosity of surface soil. Journal of Ecology, 72, pp. 767-776.
1984
30
Meyles, E.W., Hillslope and watershed scale hydrological processes and grazing
management in a Dartmoor catchment, southwest England. PhD Thesis. University of
Plymouth, Plymouth. 2002.
32
Conway V.M. & Millar A., The hydrology of some small peat-covered catchments in the
northern Pennines. Journal of the Institute of Water Engineers, 14, pp. 415 424. 1960
33
Robinson, M., The hydrological effects of moorland gripping: a reappraisal of Moor House
research. Journal of Environmental Management, 21, pp. 205-211. 1985.
34
Dunn C.E., Hydrological responses to moorland land-use change. Phd Thesis, Department
of Geography, The University of Hull. 1086.
35
Ramchunder S.J., Brown L.E. & Holden J., Environmental effects of drainage, drain
blocking and prescribed vegetation burning in UK upland peatlands. Progress in Physical
Geography, 33, 49-79. 2009.
36
Heathwaite A.L., Burt T.P. and Trudgill S.T. Runoff, sediment and solute delivery in
agricultural drainage basins: a scale-dependent approach. 182, International Association of
Hydrological Science Publications. 1989
38
Heathwaite A.L., Burt T.P. and Trudgill S.T., Land-use controls on sediment production in a
lowland catchment, south-west England. In: J. Boardmand, I.D.L. Foster and J.A. Dearing,
Soil Erosion on Agricultural Land. John Wiley and Sons Ltd. Chichester, UK. 1990.
39
Holman I.P., Hollis J.M. and Thompson T.R.E., Impact of agricultural soil conditions on
floods Autumn 2000. R&D Technical Report. Environment Agency. 2001.
-147-
40
Goodwin R.J. and Dresser M.L. Review of soil management techniques for water retention
in the Parrett catchment, Environment Agency, Almondsbury, Bristol. 2003.
41
Hollis J.M., Dresser M., Thompson T.R.E. and Newland T., Comparison of agricultural soil
conditions in the Uck and Bourne catchments during the winter periods of 2000/2001 and
2002/2003. Report for the Environment Agency. 2003.
42
Cranfield University, British Trust for Ornithology, The Open University, University of
Reading and Rothamstead Research. Scoping study to assess soil compaction affecting
upland and lowland grassland in England and Wales, Final Report. Project BD2304. 2007
43
Johns M., The Impact of Grazing and Upland Management on Erosion and Runoff:
Additional Information. R&D Project Record P2/035/9, Environment Agency, 1998.
44
Harrod, T.R. Soil suitability for grassland. p. 51-70. In M.G. Jarvis and D. Mackney (ed).
Soil Survey Applications. Technical Monograph No.13, Harpenden, UK 1979.
45
Palmer, R.C. Preliminary estimation of soil structural conditions within 10 soil landscapes in
South West England during 2002 and 2004. Final Report to Environment Agency for England
and Wales, Bristol. 2004
46
Williams A.G., Borthwick M.F., Mtika E., Ternan J.L. and Sullivan A., The influence of
changes in farming patterns on the runoff characteristics of the River Camel, Cornwall, UK.
st
Hydrology: Science & Practice for the 21 Century. Volume II. British Hydrological Society,
2004.
47
Meyles, E.W., Williams, A.G., Ternan, J.L., Anderson, J.M. and Dowd, J.F, The influence
of grazing on vegetation, soil properties and stream discharge in a small Dartmoor catchment,
southwest England, UK. Earth Surface Processes and Landforms, 31, 622-631. 2006
48
Packman J.C., Quinn P.F., Hollis J., OConnell P.E. Review of impacts of rural land use and
management on flood generation: Short-term improvement to the FEH Rainfall-Runoff model:
Technical Background. DEFRA R&D Project FD2114/PR3. 2004.
49
Boorman D.B., Hollis J.M. and Lilly A., Hydrology of Soil Types: A hydrologically-based
classification of soils in the United Kingdom. Institute of Hydrology Report 126 Wallingford,
UK. 1995.
50
Kjeldsen T.R., Flood Estimation Handbook, Supplementary Report No1. The revitalised
FSR/FEH rainfall-runoff method. Centre for Ecology and Hydrology, 2007.
51
Holtan, H.N. and Kirkpatrick, M.H., Rainfall, infiltration and hydraulics of flow in runoff
computation. Transactions of the American Geophysical Union, 31: 771-779. 1950.
52
Packman J.C., Quinn P.F., Hollis J., OConnell P.E. Review of impacts of rural land use and
management on flood generation: Short-term improvement to the FEH Rainfall-Runoff model:
Technical Background. DEFRA R&D Project FD2114/PR3. 2004.
53
JBA Consulting & Cranfield University, Land management CFMP Tool, Development of a
software tool to investigate the potential of changes in rural land use and land management
on flood generation, Environment Agency, 2008.
54
Packman J.C., Quinn P.F., Hollis J., OConnell P.E. Review of impacts of rural land use and
management on flood generation: Short-term improvement to the FEH Rainfall-Runoff model:
Technical Background. DEFRA R&D Project FD2114/PR3. 2004.
55
OConnell E., Ewen J., ODonnell G., Quinn P. Is there a link between agricultural land-use
management and flooding? Hydrology & Earth System Sciences, 11(1) 96-107, 2007.
56
Robinson M., Impact of improved land drainage on river flows. Report 113. Institute of
Hydrology. Wallingford, UK.1990.
57
Hardiman N. et al, Technical review of NFRM techniques, their effectiveness and wider
benefits, RSPB, 2009.
-148-
58
Calder I.R., Reid I., Nisbet T.T and Green J.C., Impact of lowland forests in England on
water resources: application of the Hydrological Land Use Change model. Water Resources
Research, 39, 2003.
59
Hall R.L, Hydrological aspects of broadleaved woodland: Implication for water supply.
Forests and water supply, proceedings of a discussion meeting. Heriot Watt University,
Edinburgh. Institute of Chartered Foresters. P44-60
60
Nisbet T.R., Implications of climate change: soil and water. Forestry Commission Bulletin
125, p53-67. Forestry Commission, 2002.
61
Roberts J. and Rosier P., The impact of broadleaved woodland on water resources in the
lowland UK: III The results from Black Wood and Bridges Farm compared with those from
other woodland and grassland sites. Hydrology and Earth Systems Sciences, 9(6), 614-620,
2005.
62
Harding R.J., Hall R.L., Neal C., Roberts J.M., Rosier T.T.W. and Kinniburgh D.K.
Hydrological impact of broadleaf woodlands: Implications for water use and water quality.
Institute of Hydrology, 1992.
63
Reed D., Faulkner D., Robson A., Houghton-Carr H., Bayliss A., Flood estimation
handbook: Volume 1: Overview. Institute of Hydrology, 1999.
64
Nisbet T.R., Information Note: Water use by trees. Forest Research, 2005.
65
Nisbet T.R., Silgram M., Shah N., Morrow K., and Broadmeadows S., Woodland for Water:
Woodland measures for meeting Water Framework Directive objectives. Forest Research
Monograph, 4. Forest Research, Surrey. 2011.
66
Calder I.R., Harrison J., Nisbet T.R. and Smithers R.J., Woodland actions for biodiversity
and their role in water management, Woodland Trust, 2009. (Peer reviewed).
68
Forestry Commission, A new focus for Englands woodlands strategic priorities and
progress. 1999.
69
Bird SB., Emmett BA., Sinclair PA., Reynolds B., Nicholason S., Jones T., Ponbren: The
effects of tree planting on agricultural soils and their functions. Centre for Ecology and
Hydrology Report, 2003.
70
Jackson B.M., Wheater H.S., Mcintyre N.R., Chell J., Francis O.J., Frogbrook Z., Marshall
M., Reynolds B., Solloway I., The impact of upland land management on flooding: insights
from a multiscale experimental and modelling programme, Journal of Flood Risk
Management, Volume 1 Issue 2, p71-80. 2008.
71
Alaoui A., Caduff U., Gerke H.H., Weingartner R., Preferential Flow Effects on Infiltration
and Runoff in Grassland and Forest Soils. Vadose Zone Journal, vol 10, p367 -377. 2011.
72
Nisbet T., Thomas H. and Shah N., Short Rotation Forestry and Water, In: Short Rotation
Forestry: Review of growth and environmental impacts. Ed McKay H., Forest Research,
Forestry Commission, 2011.
73
Calder I.R., Water use of eucalpt a review, In: Growth and water use of forest plantations,
Calder I.R., Hall R.L., Adlard P.G (eds). John Wiley and Sons, Chichester. 1992.
74
David, JS. And Ledger, DC. (1988) Runoff generation in a plough drained peat bog in
Scotland, Jnl of Hydrol. 99 187-199
75
Rutter A.J., Studies in the water relations of Pinus sylvestris in plantation conditions, I:
Measurements of rainfall and interception Journal of Ecology 51 191-203, 1963
76
UK forestry standard water guidelines forests and water, Forestry Commission 2011.
77
78
Robinson, M. Moore, RE. Nisbet, TR. Blackie, JR. Report no 133 From moorland
to forest: the Coalburn catchment experiment Institute of Hydrology. 1988
79
Archer, D. Newson, M. (2002) The use of indices of flow variability in assessing the
hydrological and instream habitat impacts of upland afforestation and drainage.
Journal of Hydrology. x 244-258
80
Calder, IR. Harrison, J. Nisbet, TR. Smithers, RJ. Woodland actions for biodiversity and
their role in water management Woodland Trust/Forest Research/Environment Agency. 2008.
81
McCulloch J.S.G., & Robinson M., History of forest hydrology. Journal of Hydrology, 150:
189-216. 1993.
83
OConnell P.E., Beven K.J., Carney J.N., Clements R.O., Ewen J., Fowler H., Harris G.L.,
Hollis J., Morris J., ODonnell G.M., Packman J.C., Parkin A., Quinn P.F., Rose S.C.,
Shepherd M. and Tellier S., Review of impacts of rural land use and management on flood
generation: Impact study report. DEFRA R&D Technical Report FD2114/TR. 2004
84
Robinson, M. and Dupeyrat, A., Effects of commercial forest felling on streamflow regimes
at Plynlimon. Hydrological Processes 19:1213-1226. 2005.
85
Kirby C., Newson M.D. and Gilman K, Plynlimon research: The first tow decades, Institute
of Hydrology Report 109, Wallingford. 1991.
86
Neal C., Water quality of the Plynlimon catchments. Hydrology Earth Systems Science,
Special Issue, 1, 381-763. 1997.
87
88
Bates C.G. & Henry A.J., Forest and streamflow experiment at Wagon Wheel Gap,
Colorado, Mon. Weather Rev., Soppl., 30: 1-79. 1928.
90
Hornbeck J.W., Pierce R.S. & Federer C.A., Streamflow changes after forest clearing in
New England. Water Resources. Res., 6: pp1124-1132. 1970
91
Rothacher J. Increases in water yield following clear cut logging in the Pacific Northwest.
Water Resources. Res, 6: 653 658. 1970.
92
Swank W.T. and Crossley D.A. (Editors), Forest Hydrology and Ecology at Coweeta.
Ecological Studies No 66. Springer-Verlag, New York, 469 pp. 1988
93
Naden P., Crooks S. and Broadhurst P., Impact of climate change and land-use change on
the flood response of large catchments. 31st MAFF Conference of River and Coastal
Engineers, Keele University (3-5 July 1966), 2.1.1 2.1.16. 1996.
94
Calder I.R., Evaporation in the uplands. Wiley, Chichester. 148 pp. 1990
95
Hall R.L.,. Processes of evaporation from vegetation of the uplands of Scotland. Trans. R.
Soc. Edinburgh. Earth Sci., 78, pp 327-334. 1987
96
Hall R.L. and Harding R.J., The water use of the Balquhidder catchments: a process
approach. Journal of Hydrology, 145, pp 285-314. 1993.
97
Wright IR and Harding RJ, Evaporation from natural mountain grassland. Journal of
hydrology, 145, pp 267-283. 1993.
-150-
98
Haria A. and Price D.J., Evaporation from native pine woodlands in the Cairngorm
Mountains of Scotland. Hydrology and Earth Sciences, 4(3) pp 451-461. 1999.
99
Price D.J., Kaya Y. & Tingle S., Land-use change and flood attenuation the predicted
potential benefits of semi-natural woodland. 35th MAFF Conference of River & Coastal
Engineers, Keele, July 2000.
100
Francs F., Garca-Bartual R., Ortiz E., Salazar S., Miralles J., Blschl G., Komma J.,
Habereder C., Bronstert A., Blume T., Efficiency of non-structural flood mitigation measures:
"room for the river" and "retaining water in the landscape", CRUE Research Report No I-6
2008
101
Park J.S., Cluckie I. & King P., The Parrett Catchment Project (PCP): Technical Report on
the Whole Catchment Modelling Project. 2006
102
Marshall M.R., Frogbrook Z.L., Francis O.J., Reynolds B., McIntyre N.R. and Wheater
H.S., The impact of upland land management on flooding: preliminary results from a multiscale experimental programme. In: Proceedings of the British Hydrological Society 9th
National Hydrology Symposium, pp. 85-91. 2006.
103
Wheater H.S., McIntyre N., Jackson B.M., Marshall M.R., Ballard C., Bulygina N.S.,
Reynolds B. and Frogbrook Z.,. Flood Risk Science and Management Chapter 3 Multiscale
Impacts of Land Management on Flooding. Ed Pender G, Thorne C, Cluckie I and Faulkner
H. Blackwell Publishing Ltd. 2010
104
Wheater H., Reynolds B., McIntyre N., Marshall M., Jackson B., Frogbrook Z., Solloway I,
Francis O., Chell J., Flood Risk Management Research Consortium, Impacts of Upland Land
Management on Flood Risk: Multi-scale Modelling Methodology and Results from the
Pobntbren Experiment, December 2008, FRMRC Research Report UR16. 2008
105
Kozak, J. A., Ahuja, L. R., Green, T. R., and Ma, L., Modelling crop canopy and residue
rainfall interception effects on soil hydrological components for semi-arid agriculture,
Hydrology Process., 21, 229241, 2007.
107
Law, B.E., T. Arkebauer, J.L. Campbell, J. Chen, O. Sun, M. Schwartz, C. van Ingen, S.
Verma. Terrestrial Carbon Observations: Protocols for Vegetation Sampling and Data
Submission. Report 55, Global Terrestrial Observing System. FAO, Rome. 87 pp. 2008
108
Price J.S., Heathwaite A.L. & Baird A.J. Hydrological processes in abandoned and
restored peatlands: an overview of management approaches. Wetlands ecology and
management 11, 65-83. 2003
110
Ramchunder S.J., Brown L.E. & Holden J., Environmental effects of drainage, drain
blocking and prescribed vegetation burning in UK upland peatlands. Progress in Physical
Geography, 33, 49-79. 200.
111
Soulsby C., Tetzlaff D., Rodger P., Dunn S. & Waldron S. Runoff processes, stream water
residence times and controlling landscape characteristics in a mesoscale catchment: An initial
evaluation. Journal of Hydrology, 325, 197-221. 2006
113
Rydin and Jeglum, The biology of peatland. Oxford University Press, Oxford. 2006.
-151-
114
Grayson R,, Holden J, & Rose R., Long-term change in storm hydrographs in response to
peatland vegetation change. Journal of Hydrology 389, 336-343. 2010.
115
Ballard C.E., McIntyre N., Wheater H.S., Effects of peatland drainage management on
peak flows. Hydrology Earth Systems Science, 8, 6533-6563, 2011.
116
Lewis W.K., Investigation of rainfall, runoff and yield of the Alwen and Brenig catchments.
Proc. Instn Civil Engrs 8: 17-51. 1957.
117
Robinson M, Farm drainage a cause for flooding? Agricultural and Advisory Service
Bulletin 159, MAFF, Oxford 10-12. 1985.
118
Robertson R.A., Nicholson I.A. and Hughes R., Runoff studies on a peat catchment. Proc.
2nd International Peat Congress, HMSO, London 161-166. 1968
119
Leeks G.J.L. and Roberts G., The effects of forestry on upland streams with special
reference to water quality and sediment transport. In: Good, JEG (ed), Environmental Aspects
of Plantation Forestry in Wales. Institute of Terrestrial Ecology Symposium 22: 9-24. 1987.
120
Acreman M., The effects of afforestation on the flood hydrology of the upper Ettrick valley.
Scottish Forestry, 39, 89-99. 1985
121
Holden J., Chapman P.J. and Labadz J.C., Artificial drainage of peatlands: hydrological
and hydrochemical process and wetland restoration. Progress in Physical Geography 28, 1,
95-123. 2004.
122
Holden J., Burt T.P., Evans M.G. and Horton M., Impact of land drainage on peatland
hydrology. Journal of Environmental Quality 35, 1764 1778. 2006.
123
Armstrong A., Worrall F. and Holden J., Monitoring grip-blocking techniques. (Whitendale
Trial). Draft Final Report for United Utilities. 2006.
124
Ballard C, McIntyre N and Wheater HS, Peatland drain blocking: Can it reduce peak flood
flows? BHS Third International Symposium, Managing Consequences of a Changing Global
Environment, Newcastle. 2010.
125
Geris J., Ewen J., ODonnell G. and OConnell P., Monitoring and modelling the pre-and
post-blocking hydrological response of moorland drains. BHS Third International Symposium,
Managing Consequences of a Changing Global Environment, Newcastle 2010.
126
JBA, Ripon land management project - Final Report. Department for Environment, Food
and Rural Affairs. 53pp. 2007.
127
Johnson R, Flood Planner: A manual for the natural management of river floods. Report for
WWF Scotland. 32 pp. 2007.
128
Carty A.H., Sholz M., Heal K., Keohane J., Dunne E., Gouriveau F., Mustafa A.,
Constructed Farm Wetlands (CFW) Design Manual for Scotland and Northern Ireland,
Prepared for the Northern Ireland Environment Agency and the Scottish Environment
Protection Agency, 2008.
130
131
Wilkinson M.E., Quinn P.F. and Welton P., Runoff management during the September
2008 floods in the Belford catchment, Northumberland. Journal of Flood Risk Management 3,
p285-295, 2010.
132
Environment Agency, Belford Flood Alleviation Scheme, 2011. http://www.environmentagency.gov.uk/homeandleisure/floods/129472.aspx (accessed 22 February 2012)
133
Ackers J. and Bartlett J., Flood storage Chpt 10 In Fluvial Design Guide, Environment
Agency, 2010. http://evidence.environment-
-152-
agency.gov.uk/FCERM/en/FluvialDesignGuide/Chapter_10_Background.aspx (accessed 22
Feb 2012)
134
Hall M.J., Hockin D.L. and Ellis J.B. Design of flood storage reservoirs, B014. CIRIA and
Butterworth-Heinemann. 1993
135
Bullock A. and Acreman M., The role of wetlands in the hydrological society, Hydrology
and Earth Systems Sciences, 7(3), 358-389, 2003.
136
137
Boorman D.B., Hollis J.M. and Lilly A., Hydrology of soil types: a hydrologically based
classification of the soils of the United Kingdom. Report 126, Institute of Hydrology. 1995.
138
J.A Martnez-Casasnovas, A spatial information technology approach for the mapping and
quantification of gully erosion, CATENA, Volume 50, Issues 24, 1, P293-308, 2003
141
Zhuo L., The effects of forest in controlling gully erosion. Erosion, Debris Flows and
Environmenta in Mountain Regions (Proceedings of Chegdu Symposium, IAHS Publication
209, 1992.
142
Raven E.K., Lane S.N., Bracken L.J., Understanding sediment transfer and morphological
change for managing upland gravel-bed rivers. Progress in Physical Geography, 34(1) 23-45.
2010.
143
Blackwell M. and Maltby E., Ecoflood Guidelines, How to use floodplains for flood risk
reduction. EUR 22001, European Commission, Directorate- General Research, 2006.
145
Bullock A. and Acreman M., The role of wetlands in the hydrological cycle, Hydrological
and Earth Systems Sciences, 7(3), 358-389, 2003.
147
Potter K.M., Wheres the Space for Water? How floodplain restoration projects succeed.
Masters Dissertation for the Department of Civic Design, Liverpool University. 2006.
148
Environment Agency, River Habitat Quality: the physical character of rivers and streams in
the UK and Isle of Man. Environment Agency, Bristol. 1998.
149
Broadmeadow S. and Nisbet T.R., The effects of riparian forest management on the
freshwater environment: a literature review. Hydrology and Earth System Sciences, 8(3),
286-305, 2004.
150
Forest Research, The Robinwood Robinflood report: Evaluation of Large Woody Debris in
Watercourses, Forestry Commission (Wales), INTEREG IIIC, 2008.
151
Linstead C., And Gurnell A.M. Large woody debris in British headwater rivers: Physical
habitat and management guidelines. R&D Technical Report W185, School of Geography and
Environmental Sciences, University of Birmingham, 1999.
152
Gregory K.J., Gurnell A.M. and Hill C.T., The permanence of debris dams related to river
channel processes. Hydrological Sciences Journal des Sciences Hydrologiques, 20, 3, 9,
1985.
-153-
153
Thomas H. Modelling the Hydraulic Impact of Reintroducing Large Woody Debris into
Watercourses. The Journal of Flood Risk Management, DOI: 10.1111/j.1753318X.2010.01137.x, Accepted paper, 2012.
154
Thomas H. & Nisbet T. R., An assessment of the impact of floodplain woodland on flood
flows. Water and Environment Journal, 2006.
155
Nisbet T.R. and Thomas H., Restoring Floodplain Woodland for Flood Alleviation. Forest
Research DEFRA, 2008.
156
Jones M., Farming Floodplains for the future, Staffordshire Wildlife Trust, 2010.
157
Sartor J.F., Flood water retention by riverine and terrestrial forests. Trier University of
Applied Sciences, Schneidershof.
Aatz, M. and Musong, S. Two-dimensional dynamic calculation of natural water courses
(in German), Diploma thesis, Trier University of Applied Sciences (2005)
158
159
160
162
McGahey, C. & Samuels, P.G., Methodology for conveyance estimation in twostage straight, skewed and meandering channels, XXX IAHR Congress,
Thessaloniki, Volume C1 of Proceedings published by IAHR, 2003.
163
164
Arcement G.J. and Schneider V.R., Guide for Selecting Manning's Roughness
Coefficients for Natural Channels and Flood Plains. United States Geological Survey
Water-supply Paper 2339. 1989
165
Lane S. N., Roughness time for a re-evaluation, Earth Surface Processes and
Landforms, 30, 251-2. 2005.
166
Whatmore S. and Langstrom C., Mannings n: putting roughness to work, in How well do
facts travel? Morgan M and Howlett P (eds.) Chapter 4, Cambridge University Press,
Cambridge. 2010.
167
Nisbet T.R. and Thomas H., Flooding and Flood Management Inquiry, Rural Affairs and
Environment Committee, Submission from Forestry Commission Scotland, The Role of
Woodland in Flood Control: A Landscape Perspective. Scottish Parliament Archives, 2007.
Accessible at:
http://archive.scottish.parliament.uk/s3/committees/rae/inquiries/flooding/ForestryCommission
Scotland.htm (accessed 16 January 2011)
168
Hardiman N., Cunningham R. & Johnstonova A., Technical review of NFRM techniques,
their effectiveness and wider benefits. RSPB, 2009.
170
Schultz S.D. and Leitch J.A., The feasibility of restoring previously drained wetlands to
reduce flood damage. Journal of Soil & Water Conservation, 58(1) 21-29, 2003.
171
Morris J., Hess T.M., Gowing D.J., Leeds-Harrison P.B., Bannister N., Wade M., Vivash
R.M., Integrated washland management for flood defence and biodiversity, English Nature,
2004.
172
-154-
173
Riley A.L., Restoring Streams in Cities. A Guide for Planners, Policymakers and Citizens.
Island Press, 1998.
174
175
University of Dundee & Aberdeen City Council, Aberdeen Beach Case Study, IMCORE,
2010
176
Pye K., Samantha S., Blott S., Sand dune processes and management for flood and
coastal defence, Part 2. Defra R&D Technical Report FD1392/TR, 2007.
177
Environment Agency & Royal Haskoning, Saltmarsh Management Manual, DEFRA, 2005.
178
Mller, I., Spencer, T., French, J.R., Leggett, D.J. & Dixon, M. Wave transformation over
salt marshes: a field and numerical modelling study from North Norfolk, England. Estuarine,
Coastal and Shelf Science 49: 411-426. 1999.
179
Mller, I. & Spencer, T. Wave dissipation over macro-tidal saltmarshes: effects of marsh
edge typology and vegetation change. Journal of Coastal Research Special Issue 36, 506521. 2002.
180
Cooper, N. Coastal Data Analysis: The Wash Study 2: Wave attenuation over inter-tidal
surfaces. Report to the Environment Agency. 2001.
181
Mller, I., Spencer, T., French, J.R., Leggett, D.J. and Dixon, M. The sea-defence value of
salt marshes: field evidence from north Norfolk. Journal of Chartered Institution of Water and
Environmental Management (CIWEM) 15: 109-116. 2001.
182
Reed D.J., Davidson-Arnott R. and Perillo G.M.E., Estuaries to coastal dunes: Chapter 5
Estuaries, coastal marshes, tidal flats and coastal dunes . 2009.
183
Rupp S. & Nicholls R.J., Managed Realignment of Coastal Flood Defences: A Comparison
between England and Germany. Flood Hazard Research Centre, 2002.
184
186
Environment Agency, The Alkborough tidal defence scheme, Fact Sheet, 2006.
187
Leggett D.J., Cooper N., Harvey R., Coastal and estuarine managed realignment design
issues. C628 CIRIA, 2004.
188
Scyphers S.B., Powers S.P., Heck K.L., Byron D., Oyster Reefs as Natural Breakwaters
Mitigate Shoreline Loss and Facilitate Fisheries. PLoS ONE 6(8): e22396.
doi10.1371/journal.pone.0022396. 2011.
190
OConnell P.E., Beven K.J., Carney J.N., Clements R.O., Ewen J., Fowler H., Harris G.L.,
Hollis J., Morris J., ODonnell G.M., Packman J.C., Parkin A., Quinn P.F., Rose S.C.,
Shepherd M. and Tellier S., Review of impacts of rural land use and management on flood
generation: Impact study report. DEFRA R&D Technical Report FD2114/TR. 2004.
191
JBA Consulting, Delivery of Making Space for Water, HA6 & Ha7, Identification of
catchments sensitive to land use change. Environment Agency, 2008
192
Broadmeadow & Nisbet T., Opportunity Mapping for Woodland to Reduce Flooding in
Yorkshire and the Humber Region. Forest Research, 2009.
194
Halcrow CRESS Allan Water Natural Flood Management Scoping Study. SEPA, 2011.
-155-
195
Pagella T. et al, POLYSCAPE: Multiple criteria GIS toolbox for negotiating landscape scale
ecosystem service provision. Presentation, 2009.
http://www.futureforest.eu/uploads/polyscape_future_forests_09.pdf
196
Watts K. & Eycott A. Biological and Environmental Evaluation Tools for Landscape
Ecology. Forest Research.
197
Hewett et al. Farm and Agriculture Risk Matrix (FARM) technical manual (Newcastle
University, 2006) http://www.ceg.ncl.ac.uk/theFARM/index.html
198
Odoni N. & Lane S., Assessment of the impact of upstream land management measures
on flood flows in Pickering Beck using OVERFLOW. Durham University, Forest Research
and Defra, 2010.
199
Lane, S.N., Reaney, S.M. and Heathwaite, A.L., 2009. Representation of landscape
hydrological connectivity using a topographically driven surface flow index. Water Resources
Research, 45, W08423, doi:10.1029/2008WR007336.
200
Porter R.A.I., An Evaluation of a Hydrological Model Used to Predict the Impact of Flow
Attenuation on Downstream Flood Flows. Masters thesis, Durham University. 2011 Available
at Durham E-Theses Online: http://etheses.dur.ac.uk/813/ (accessed 21 February 2012)
201
Leopold L.B. and Maddock T., The Hydraulic Geometry of Stream Channels and Some
Physiographic Implications. US Department of the Interior. 1953.
202
ADAS UK & Macaulay Land Use Research Institute, Provision of a Screening Tool to
Identify and Characterise Diffuse Pollution Pressures: Phase II. SNIFFER, 2006.
203
Beven K.J., Rainfall-Runoff Modelling: The Primer. Second Edition, Wiley-Black, 9780470714591, 2012.
204
Beven K.J., Changing ideas in hydrology the case of physically-based models. Journal of
Hydrology. 105: 157-172.
205
Bulygina N., McIntyre N., Wheater H., Conditioning rainfall-runoff model parameters for
ungauged catchments and land management impacts analysis. Hydrological Earth Systems
Science, 2009.
206
Moore, R.J.; Bell, V.A.; Cole, S.J.; Jones, D.A.. Rainfall-runoff and other modelling for
ungauged/low-benefit locations. Research Contractor: CEH Wallingford, Environment
Agency, Bristol, UK, 249pp. 2007
207
Kjeldsen T., Jones D., Bayliss A., Spencer P., Surendran S., Laeger S., Webster P.,
McDonald D., Improving the FEH Statistical Method. In: Flood & Coastal Management
Conference 2008. Environment Agency & Defra, 2008.
209
Kjeldsen T., Jones D., Bayliss A., Spencer P., Surendran S., Laeger S., Webster P.,
McDonald D., Improving the FEH Statistical Method. In: Flood & Coastal Management
Conference 2008. Environment Agency & Defra, 2008.
210
Parker, C., Thorne, C.R., and Clifford, N.J. Broad. A broad-scale assessment of sediment
dynamics in British rivers: Developing and assessing ST:REAM a reach based sediment
balance model. Earth Surface Processes and Landforms - In review.
211
ABPmer, The Met Office & Proudman Oceanographic Laboratory, Atlas of UK Marine
Renewable Energy Resources, Commissioned by DTI and produced by y (2008)
http://www.renewables-atlas.info
212
-156-
213
Werritty A., Ball T., Spray C., Bonell M., Rouillard J., Archer N., Bowles C. and Moir H.,
Restoration Strategy: Eddleston Water Scoping Study. Dundee University and Cbec. 2010
214
Hough M.N. and Jones R.J.A., The United Kingdom Meteorological Office rainfall and
evaporation calculation system: MORECS version 2.0 an overview. Hydrology and Earth
Systems Sciences, 1(2), 227-239, 1997.
217
Hough M.N. and Jones R.J.A., The United Kingdom Meteorological Office rainfall and
evaporation calculation system: MORECS version 2.0 an overview. Hydrology and Earth
Systems Sciences, 1(2), 227-239, 1997.
218
Nisbet T.R., and Thomas H., The role of woodland in flood control: a landscape
perspective Forest Research, 2008
219
Moore R.J., The probability-distributed principle and runoff production at point and basin
scales. Hydrological Sciences Journal, 30, 273 to 297. 1985.
220
Moore R.J., The PDM rainfall-runoff model. Hydrology and Earth System Sciences, 11(1),
483 to 499, 2007.
221
Bell V.A., and Moore R.J., A grid-based distributed flood forecasting model for use with
weather radar data: Part 2. Case studies Hydrology and Earth Systems Sciences, 2(3), 283298, 1998.
222
Bell V.A., Kay A.L., Jones R.G., Moore R.J., Development of a high resolution grid-based
river flow model for use with regional climate model output. Hydrology Earth Systems
Science, 11(1), 532-549, 2007.
223
Hough M., An historical comparison between the Met Office Surface Exchange SchemeProbability Distributed Model (MOSES-PDM) and the Met Office Rainfall and Evaporation
Calculation System (MORECS), Met Office 2003.
225
Robson A.J. and Reed D.W., Statistical procedures for flood frequency estimation. Flood
Estimation Handbook Vol 3., Institute of Hydrology, Wallingford, 1999.
226
Reed, Duncan, Faulkner, Duncan, Robson, Alice, Houghton-Carr, Helen, Bayliss, Adrian, ,
Flood estimation handbook: Vol. 4: Restatement and application of the flood studies report
rainfall-runoff method, Institute of Hydrology, 1999.
228
Beven K.J. and Binley A.M., The future of distribute models: Model calibration and
uncertainty prediction. Hydrological Processes, 6(3): 279-298, 1992.
-157-
Appendix A
Review of Jacobs Section 20 Review
(CREW, 2012)
Page | 0
Final Report
Project CD2011_07
20/01/2012
Page | 0
CREW 2012
The views expressed in this document are not necessarily those of CREW. Its members or agents accept
no liability for any loss or damage arising from the interpretation or use of the information, or reliance
upon views contained herein.
Whilst this document is considered to represent the best available scientific information and expert
opinion available at the stage of completion, it does not necessarily represent the final or policy
positions of the project funders.
Dissemination status
Unrestricted
Research contractors
This document was produced by:
1) Prof A Werritty and Dr T. Ball
School of the Environment, and
UNESCO Centre for Water Law, Policy and Science,
University of Dundee,
Dundee DD1 4HN,
Scotland, UK
2) Dr Scott Arthur
CREW Management
All queries related to this document should be directed to the CREW Facilitation Team
Executive Summary
Project reference: Review of Jacobs Section 20 review and GIS tool (January, 2012)
Background to research
Section 20 of the Flood Risk Management (Scotland) Act 2009 requires SEPA, by 22 December 2013, to
undertake national assessment of natural flood management (NFM), whether alteration (including
enhancement) or restoration of natural features and characteristics of any river basin or coastal area in
a flood risk management district could contribute to the management of flood risk for that district.
Objectives of research
To review the Jacobs section 20 review and GIS tool based on a set of pre defined questions.
Key words
Flooding, Natural Flood Management, GIS, Land Use
1. INTRODUCTION
Guide questions to guide expert reviews of the Section 20 methodology were supplied by
SEPA.
1.1 Questions
1.1.1 Section 2 Literature review:
Q1) The literature review was limited to those studies that provided quantitative evidence of the
effects of NFM techniques on catchment scale flooding. In your view, is this the right approach
or have any important studies been missed?
Q2) In your view, are the studies identified in the literature review representative of Scottish
catchments?
Q3) Do you agree with conclusions and recommendations in sections 2.2.4, 2.3.4 and 2.4.4?
1.1.2 Section 3 GIS tool:
Q4) Do you agree with the outlined approach described in section 3.1?
Q5) Do you have any observations in relation to the testing of the model described in section
3.3?
Q6) In your opinion, is the preliminary rule set described in section 3.4 acceptable and are there
sufficient grounds for the rule set? Suggest, if possible, alternative ways to improve the rule
set.
Q7) In your opinion is the outline method for coastal floodplains correct?
Q8) Do you agree with the limitations specified in section 3.7 and recommendations in section
8.8?
Q9) Do you have any other observations from Section 3 that you would like to highlight?
Q10) Overall, would you recommend the use of the proposed GIS tool and the rule for use as
the screening tool to provide an indication of the effect of NFM techniques?
Q11) Do you have any recommendations for future work that would be beneficial to further
develop the rule set and the modelling tool?
2. REVIEW OF JACOBS SECTION 20 REVIEW AND GIS TOOL BY ALAN WERRITTY AND
TOM BALL, UNIVERSITY OF DUNDEE.
2.1 Overview
The task of providing a GIS tool to deliver the Section 20 requirements in the Flood Risk
Management (Scotland) Act 2009 was a very challenging one. As noted in Werrittys earlier
CREW report on the SAIFF Task and Finish group Land Management and Flooding paper, the
evidence base for effective and efficient NFM measures is still very fragile with many key issues
yet to be resolved. The most notable of these are:
Quantifying the effects of individual NFM measures in reducing flood risk across
catchments with diverse characteristics (effectiveness)
Determining how individual measures at a local scale contribute to flood risk reduction at
the catchment scale (scaling)
Reducing uncertainty by developing better physically-based rainfall-runoff models
(modelling)
Deploying measures within a catchment in an optimal way so that those installed
upstream do not adversely impact on communities downstream and vice-versa
(location)
Future-proofing the selection of measures to include potential impacts of climate
(changing risk)
Developing measures which, in terms of ecosystem services, provide other catchmentbased benefits in terms of enhanced water quality, protecting biodiversity and storing
carbon (multiple benefits).
The comments below should be read with these issues in mind by way of context.
2.2 Review:
Q1) The literature review was limited to those studies that provided quantitative evidence
of the effects of NFM techniques on catchment scale flooding. In your view, is this the
right approach or have any important studies been missed?
The literature review is exhaustive, authoritative and up to date. it is also extremely well
structured with a clearly specified context, identification of uncertainties (taken from
Environment Agency, 2007) and prefaces to each major group of measures (upland forest
cover, riparian woodland, floodplain woodland, benefits of upland drain blocking, benefits of
wetlands and floodplains) with an informed commentary. Each of the three major groups of
measures (upland forests and woodlands, upland drain blocking and wetlands and floodplains)
is reported on in a series of summaries. The quality of the reflective commentary in the
prefaces to each of the three major groups of measures is very impressive. The review is
written in very clear English with technical terms largely avoided or, where they are used,
suitably explained in footnotes.
Table 2.1 which itemises potential NFM measures and where they are likely to generate the
optimal benefit includes the following measures which are not addressed in the rest of the
literature review nor via the GIS tool:
This is acknowledged in the statement (section 2.1.1) that other techniques will be considered
as the Section 20 methodology and GIS tool continues to be further enhanced. But for now the
Section 20 GIS tool has excluded any assessment of changes in agricultural practice most
notably reduced stocking densities and land-use change both of which could offer significant
NFM benefits. In future development of the Section 20 GIS tool, these omissions will need to be
addressed.
The rest of section 2 focuses on three groups of NFM measures relating to upland forests and
woodland, upland drain blocking and wetlands and floodplains. Each of these groups of
measures is commented upon in turn:
risk. The precise nature of this impact is contested (Robinson, 1990), but it should be included
in any further developments of the Section 20 GIS tool.
1.3 Wetlands and floodplain
This is another thorough and detailed review with all the major sources incorporated. The
preface to the wetlands review (section 2.4.1) provides a valuable summary of the key findings
from the SNIFFER wetland typology. The associated Table 2.9 summarising the character of
each type of wetland (plus linked photographs) is useful, and needs to be cross-related to Table
3.1 which identifies potential NFM measures and associated GIS tool rule sets. It is good to
note the major reservations in terms of providing unqualified estimates of downstream benefits
of many of the measures reviewed in section 2.4.1. In particular, we endorse the statement that
there is less certainty about the potential benefits of different wetland types on flood risk
reduction and accept that this section provides a more speculative approach than other
sections.
The most important issue when using wetlands and floodplain to promote NFM is the
channel/floodplain interaction and the ratio of water stored (on the floodplain) and conveyed
down the channel. The cautionary advice (Section 2.4.3 c) that the Section 20 GIS tool should
only be used to provide a broad indication of possible benefits (and sometimes disbenefits) is
well judged as it the recommendation that should any measures be considered further they
must be more fully appraised by hydraulic modelling for the river and floodplain in question.
This points up the difficulty of devising a generic GIS tool to assess the NFM potential
floodplains with diverse properties in terms of areal extent, slope, connectivity with the channel,
and location and nature of any flood embankments. Especially important is the claim that
agricultural embankments can help reduce downstream flood risk by, once locally breached,
impounding significant volumes of water. Thus the often favoured measure of setting existing
flood embankments further back on the floodplain is not universally beneficial.
As a result of these qualifications, it is much more difficult to identify the quantifiable benefits of
wetlands and floodplains in terms of NFM. The particularity of each floodplain site and its local
connection with the channel can result in both enhanced and reduced downstream flood risk
depending on its areal extent, slope, and location and nature of existing flood embankments.
As the review makes clear, this is the most challenging group of NFM measures to incorporate
into a GIS-based assessment.
Q2) In your view, are the studies identified in the literature review representative of
Scottish catchments?
Each major group of measures is commented on separately:
1.4 Upland forests and woodlands
The section on upland forest reports many studies from around the world including classic
paired catchment projects in Colorado (Bates and Henry, 1928), New England (Hornbeck et al.,
1970), Oregon (Rochacher, 1970) and North Carolina (Swank and Crossley, 1988). Whilst
valuable in summarising the impacts of forests on the water cycle, these studies rarely provide
detailed information on flood runoff generation and are drawn from settings in which the tree
species, soils, topography and climate are markedly different from Scotland. As already noted,
the results from spatially-distributed process-based models under semi-arid, sub-alpine and
alpine climates in Poyo (Spain,) Kamp (Austria) and Germany (Iller) are used to compile key
findings in Table 2.2. The transferability of these findings (with no differential weightings) is
questionable. As was discovered by the Institute of Hydrology over 20 years ago, even the
water yields from coniferous forests and grasslands in Plynlimon (Wales) could not be replicated
in the Balquhidder (Scotland) due to the higher incidence of snow and contrasting plant and
scrub species in the latter (Hall, 1987).
25% should be reduced to 20%. The proposed 10% most likely attenuation for 0.5% annual
probability flows is credible.
2.3.4 Riparian woodland
We agree that the Laver and Pickering studies are the most robust in providing estimates for
flood risk reduction. But the 60 minute delay for upper and middle catchment locations strongly
reflects the local contexts. For example, the Pickering study involved both planting riparian
woodland and installing a large number of large woody debris dams. In modelling the impact of
these measures Odoni and Lane (2010) observe that:
the sizes of the peak flow and excess flood volume reductions are lower for
smaller events. This is an important property of the kinds of interventions being
discussed: as the size of the event increases, so the contribution that the
interventions make to reducing flood risk increases. It arises from the fact that
bigger events are more able to transfer water into the interventions .... and reflects
the general property of the Pickering Beck system viz. that the river is relatively
incised into floodplain material, reducing the natural attenuating effect of
floodplains on peak flows.
This illustrates the importance of local topography. They also add the caveat that location vis-vis the urban site to be protected is an important factor with interventions in the upper part of the
catchment generally preferred.
Nisbet et al. (2011) reporting on the Laver study also report that riparian woodland can be used
to desynchronise flood flows from specified tributaries (in their case the Laver and the Skell)
with reduction in downstream flood risk (in this case Ripon). This illustrates the need to take
into account downstream conveyance and contributions from other tributaries,
Given no other relevant studies available the proposed rule set is sound, but should be used
with due regard for location within the catchment.
2.4.4 Floodplain woodland
Again given the absence of other studies, the recommended rule set is endorsed, but again with
many caveats. As the Laver and Pickering studies demonstrate, local context can markedly
influence the impact of floodplain woodland and, as the photos in this section of the report
reveal, actual values for Mannings n are strongly influenced by the extent and nature of the
woody species present. The 300 mm increase in water level is thus likely to include much local
variation.
[Note: In seeking to evaluate the GIS tool, we sought to obtain access to the software
developed by Jacobs. Although the code was released by SEPA in late December, this was not
supported by access to the underlying data due to licensing issues. The answers to questions
4-11 are thus restricted to what can be gleaned from the report and are inevitably more partial
than we would have wished.]
Q4) Do you agree with the outlined approach described in section 3.1?
We commend the overall approach which builds on the findings reported in the literature review.
The limitations are clearly identified and the absence of measures involving channel
enhancement and floodplain reconnection noted. There is, however, one fundamental
weakness in the design of the tool the division of the catchment into equal sections based on
travel times down the main stem (upper, middle and lower) irrespective of where the
downstream outlet point Tp is located. If the tool is always used for a large catchment in which
there are clearly defined upland reaches (small tributaries, flowing through moorland and/or
forests into main stems with minimal floodplain development upper) followed downstream by
an emerging floodplain and larger tributaries (middle) which eventually gives rise to an
extensive floodplain in which the bulk of the flow is along the main stem (lower), then the tool is
appropriate. But if the downstream outlet point Tp is much further upstream (eg in which upland
catchment characteristics dominate) thus precluding this balanced mix of reaches, the results
will reflect the imposition of a tri-partite division which fails to reflect the actual characteristics of
the catchment. In addition, the arbitrary allocation of equal weighting (33% of average travel
time from source to outlet down the main stem) to each upper middle and lower reach will not
be appropriate for all catchments. The next iteration of the tool should allow users to define the
mix of upper, middle and lower segments of the catchment.
Q5) Do you have any observations in relation to the testing of the model described in
section 3.3?
Inevitably testing a model of this kind poses many challenges and given prior information on the
NFM opportunities afforded by the Enrick, this catchment has much to offer. But there are gaps
in the information provided which make it difficult to assess the robustness of the derived rule
sets in section 3.4. These are:
Only one flow gauge is identified and this is in the middle reaches so any flow
modelling results downstream cannot be compared with actual data.
The removal of the floodplain reach at Corrimony and Loch Mieklie (which is immediately
upstream of the only flow gauge) provides a valuable simplification of the catchment
characteristics but severely weakens the robustness of the results which are used to
derive the rule.
The quality of the DTM is not fully specified. How dense was the network of crosssections across the floodplain and channel? What was the extent of the LiDAR
coverage?
This points to a more fundamental problem the extent that the Enrick (or any single
catchment) can provide a robust calibration of the GIS tool set. The values reported for the
Enrick inevitably reflect a highly elongate catchment with a channel network with tributaries
steadily decreasing in size as they flow into the main stem in that sense it cannot be seen as
generic. Standard hydrologic theory tells us that the changing hydrograph along the main stem
will strongly reflect the shape of the catchment and drainage network. Were the catchment to
be roughly circular with most of the tributaries flowing into the main stem very close to the outlet,
the shape of the hydrograph along the main stem would be very different to that on the Enrick.
Similarly, if all the tributaries were to be of approximately the same size and equally spaced, the
main stem hydrograph would differ again. Clearly, the tool cannot be tested against all these
possibilities although we do recommend later that it be re-calibrated against the Eddleston
Water for which there is now a dense network of flow gauges and extensive 2D modelling of
floodplain inundation at the 0.05 and 10% flood probabilities But it would also be worth
undertaking further modelling with idealised channel networks (in quasi-circular and quasi-fishbone networks) to determine how far catchment shape and drainage network properties
condition the findings.
The development of rule sets based on key quantitative findings from the literature is clearly
developed and, with the reservations already noted on some of the values imported in terms of
forest cover see question 3 above generally works well. Clearly the weakest part of the
calibration of the GIS tool relates to the generic model of floodplain enhancement. Whilst the
methodology cannot be faulted, its heavy reliance on the Enrick site (with its lack of information
on the quality of the DTM, and the removal of part of the floodplain at Corrimony and Lock
Meiklie) raises serious questions as to the transferability of the findings. By the authors own
admission this is the least robust component of the GIS tool and should be subject to much
more scrutiny and testing before it is recommended for operational use.
Q6) In your opinion, is the preliminary rule set described in section 3.4 acceptable and
are there sufficient grounds for the rule set? Suggest, if possible, alternative ways to
improve the rule set.
Table 3.1 which cross references potential NFM measures (for which rule sets have been
developed) against a typology of wetlands provides an appropriate framework. But the forest
cover and riparian woodland rule sets are strongly conditioned by the arbitrary demarcation of
every catchment into 33% upper, 33% middle and 33% lower. As noted above (question 4)
this straight jacket needs to be loosened with consequential changes in how the rule sets are
used operationally. The current rule sets in Tables 3.2, 3.3 and 3.4 are contingent on the 33%
division of the catchment into three reaches.
As already noted (question 5) the derivation of the floodplain enhancement rule set strongly
reflects the topography and hydrology of the Enrick test catchment. Given the qualifications
already noted, top priority should be given to testing this rule further ideally against
catchments for which better flow data, and 2D modelling of flood inundation exist. Given the
limitations on the application of this particular component of the GIS tool noted by the authors, it
may not yet be possible to model this floodplain enhancement NFM measures without recourse
to site specific 1D/2D models. The use of an Efficiency Factor whilst conceptually valid, is
weakened by the lack of evidence and supporting case for the values reported in Table 3.8.
There is no reference to the literature review, nor any model results, to justify the values
proposed.
Q7) In your opinion is the outline method for coastal floodplains correct?
The outline method quantifies the potential for flood storage by an inundated area of coastline
subject to managed realignment. The difficulty in applying this approach to quantification of
flood risk reduction is the inherent nonlinearities that must be encompassed in an assessment
of flood peak reduction when applied to tidal surge conditions. This is especially true where the
realignment is provided by a breach in flood embankments rather than a wholesale removal,
usually the most cost-effective option. Thus, although the outline methods is not correct in a
scientific sense (to be so, it would need to incorporate surge characteristics that are highly
specific to the site in question, precluding generic assessment), it does represent an
assessment of theoretical maximum storage potential under optimal scenarios.
Q8) Do you agree with the limitations specified in section 3.7 and recommendations in
section 3.8?
The limitations identified in section 3.8 are generally well-specified. But the second limitation,
relating to the floodplain enhancement tool, fails to take into account undue reliance on results
specific to the topography and hydrology of the Enrick (addressed in question 6 above). We
strongly endorse the final limitation which suggests that lower reaches in large river systems
may require further consideration and some adjustment. As already noted, the rule sets for
lower reaches in Tables 3.2, 3.3 and 3.4 are contingent on the arbitrary demarcation of every
catchment into 33% upper, 33% middle and 33% lower reaches. Revisiting this demarcation
is likely to result in major adjustments to the values for lower reaches. In addition, issues
relating to the robustness of the rule set for floodplain enhancement will be most acute in the
lower reaches of large river systems. We also underline the statement that channel/floodplain
interactions and processes are invariably complex and outcomes are not always as expected.
Floodplain enhancement measures may well produce disbenefits as well as benefits, and these
will need to be thoroughly assessed, ideally via dynamic modelling.
Q9) Do you have any other observations from section 3 that you would like to highlight?
Our main concerns arising from Section 3 have already been itemised above. Here we flag up
the most significant ones:
undue reliance on calibration of rule sets from the Enrick catchment which inevitably is
not generic for other Scottish catchments
the straight jacket that all basins are assigned 33% proportions to upper, middle, and
lower reaches raises major issues in terms of the operational use of the GIS tool.
Q10) Overall, would you recommend the use of the proposed GIS tool and the rule for
use as the screening tool to provide an indication of the effect of NFM techniques?
As noted at the beginning of this commentary, the task to develop a GIS tool to deliver the
Section 20 requirements in the Flood Risk Management (Scotland) Act 2009 was a very
challenging one. Jacobs are commended for the thoroughness of their literature review and the
innovative way they have sought to translate a very partial evidence basis into an operational
GIS tool. They have skilfully circumvented the twin perils of over-interpreting a fragile evidence
base on the impacts of NFM measures, whilst at the same time developing a well-calibrated GIS
tool applicable to catchments the length and breadth of Scotland.
At this stage in the
development of flood risk science, there are few alternative approaches available. However, it
would be of great interest to compare the findings from the Jacobs GIS tool on the Allan Water
with those produced by the Centre for River EcoSystems Science/Halcrows Allan Water Natural
Flood Management Techniques and Scoping study report for SEPA (2011). This would begin to
address many of the issues raised above re undue reliance on the Enrick results. We much
regret that although the GIS tool code was released to us by SEPA, data licensing constraints
prevented us from running it on the Allan Water.
Q11) Do you have any recommendations for future work that would be beneficial to
further develop the rule set and the modelling tool?
There are many recommendations implicit in the earlier comments. Here we gather together
the key ones:
The forest cover and riparian woodland rule sets are strongly conditioned by the arbitrary
demarcation of every catchment into 33% upper, 33% middle and 33% lower. This
straight jacket needs to be loosened with consequential changes in how the rule sets are
used operationally. The next iteration of the tool should allow users to define the mix of
upper, middle and lower segments of the catchment.
The outcomes of the existing rule set for the GIS tool should be tested on the Eddleston
Water for which there is now a dense network of flow gauges and extensive 2D
modelling of floodplain inundation at the 0.05 and 10% flood probabilities.
Further modelling be undertaken with idealised drainage networks (in quasi-circular and
quasi-fish-bone networks) to determine how far catchment shape and drainage network
properties condition the outcomes of the existing rule set
The rule set relating to floodplain enhancement is very dependent on its calibration on
the Enrick Water. Top priority should be given to testing this rule further ideally against
catchments for which better flow data, and 2D modelling of flood inundation exist.
Initially this could be undertaken using the extensive database now available for the
Eddleston Water.
Given the limitations on the application of this particular component of the GIS tool, it
may not yet be possible to model floodplain enhancement NFM measures without
recourse to site-specific 1D/2D models.
The coastal inundation component of the tool be also subjected to site-specific testing in
relation to tidal dynamics and surges, looking particularly at the nature of the breach in
the site defences.
Table 2.1 itemises the full range of NFM measures. As explained in section 2.1.1 some
of these have yet to be included in the GIS tool. In due course the following measures
will need to be incorporated:
- Reducing the impact of sheep grazing
- Blocking field drains
- Land and soil management activities to delay surface water runoff
- Restoring rivers and river channels
- Re-meandering
- Land-use change arable reversion
References
Acrement G J and Schneider V R (1989) Guide for Selecting Mannings Roughness Coefficients
for Natural Channels and Floodplains. US Geological Survey Water Supply Paper 2339.
Bates C G and Henry A J (1928) Forest and streamflow experiment at Wagon Wheel Gap,
Colorado, Monthly Weather Review, Suppl. 30, 1-79.
Environment Agency (2007) Delivery of Making Space for Water: HA6 Catchment Scale LandUse Management: HA7 Land Management Practices, Final Report.
Forestry Commission (2003) Forest and Water Guidelines, Forestry Commission.
Hall R L (1987) Processes of evaporation from vegetation of the uplands of Scotland,
Transactions of the Royal Society of Edinburgh, 78, 327-334.
Hornbeck J W, Pierce R S and Federer C A (1970) Streamflow changes after forest clearing in
New England, Water Resources Research, 6, 1124-1132.
Johnson R, Watson M and McQuat E (2008) The Way Forward for Natural Flood
Management in Scotland, Report for Scottish Environment LINK
Jones M (2010) Farming floodplains for the future, Final Report, Staffordshire Wildlife Trust.
Lane S N (2005) Roughness time for a re-evaluation, Earth Surface Processes and
Landforms, 30, 251-2.
Nisbet N, Silgram M, Shah N, Morrow K and Boardmeadow S (2011) Woodland for Water:
Woodland Measures for meeting Water Framework Objectives, Environment Agency.
Odoni N A and Lane S N (2010) Assessment of the impact of upstream land management
measures on flood flows in Pickering Beck using Overflow. Appendix 12.2 to Final Report of
Slowing the flow at Pickering, Defra FCERM Multiple objective flood management
demonstration project RMP5455, Nisbet, T R, Marrington S, Thomas, H, Broadmeadow S and
Valantin G.
Raven E K, Lane S N, Ferguson R I and Bracken L J (2009) The spatial and temporal patterns
of aggradation in a temperate, upland, gravel-bed river, Earth Surface Processes and
Landforms, 34, 1181-1197
Rochacher J (1970) Increases in water yield following clear cut logging in the Pacific Northwest,
Water Resources Research, 6, 653-658.
Robinson M (1990) Impact of improved land drainage on river flows, Institute of Hydrology
Report 12, Institute of Hydrology.
Scottish Environment Protection Agency (2011) Allan Water Natural Flood Management
Techniques and Scoping study, report by Centre for River EcoSystems Science/Halcrows.
Swank W T and Crossley D A eds. (1988) Forest Hydrology and Ecology at Coweeta,
Ecological Studies No. 66, Springer- Verlag, New York.
Whatmore S and Langstrom C (2010) Mannings n: putting roughness to work, in How well do
facts travel? Morgan M and Howlett P (eds.) Chapter 4, Cambridge University Press,
Cambridge.
3) Review of Jacobs Section 20 review and GIS Tool by Scott Arthur & Nick Wallerstein,
Heriot-Watt University.
3.1) Review
Q1) The literature review was limited to those studies that provided quantitative evidence
of the effects of NFM techniques on catchment scale flooding. In your view, is this the
right approach or have any important studies been missed?
The approach undertaken is obviously driven by the questions that need to be addressed. In
this case it is assumed that the focus upon quantitative studies was initiated in order that a
practical tool (i.e. the Section 20 GIS Tool) could be developed for assessing the impact of NFM
scenarios upon flood attenuation through the catchment system. As such the approach adopted
was very effective.
With regard to the broader scope of this subject there are a significant number of other
management practices associated with NFM, which are not discussed in the literature review.
These are listed in the report however, (see Table 2.1) and comprise: blocking field drains;
reducing the impact of sheep grazing; restoring gully woodlands; land and soil management
activities, and; landuse change arable reversion. These are acknowledged in the report as
techniques that need to be addressed in future but which have not been considered as yet. This
might perhaps be due to the perceived lack of hard data associated with these measures.
However, several of these topic do have a significant body of literature associated with them,
especially relating to the impact of sheep grazing (reversion to arable land), and upland
agricultural landuse management. A selected list of relevant studies relating to these factors and
practices is supplied in a reference list at the end of this document.
Q2) In your view, are the studies identified in the literature review representative of
Scottish catchments?
The review is comprehensive with respect to the body of the literature cited. The main texts from
which quantitative data have been extracted are either directly associated with studies made in
Scottish catchments, or associated with other research conducted elsewhere in the UK in
catchments which have similar hydro-climatic conditions, landuse types and topography. The
review also discussed findings from a number of overseas studies, which were deemed to be
relevant to the UK.
A selection of other useful papers and reports, not cited in the literature review, which are
relevant to some of the NFM measure examined in this study are supplied in the list of
references at the end of this document. Three studies which are directly related to Scottish
catchments are highlighted in bold-italic in the list.
Q3) Do you agree with conclusions and recommendations in sections 2.2.4, 2.3.4 and
2.4.4?
Conclusions in section 2.2.4
The conclusions relating to Forest Woodland studies are consistent with the literature reviewed,
and the rule set values have been extracted from a reasonably broad range of literature which is
summarized in Table 2.5. Data obtained from a number of studies conducted overseas, which
were deemed relevant to the review, have been included in this Table but in reality there is
probably some disparity between those studies and the ones conducted in the UK (in terms of
hydro-climatic regime, etc.), which is likely to have introduce some uncertainty into the rule set.
The conclusions regarding Riparian Woodland are consistent with the findings obtained from the
two main sources of literature; the Pickering Beck and River Laver studies (refer to Table 2.6).
However, as the rule set values are only based upon the data extracted from these two studies,
the degree of uncertainty in the values is likely to be significant.
The conclusions relating to Floodplain Woodland are generally consisted with the findings
obtained from two simulation studies based upon the River Carey and River Laver (refer to
Table 2.8). It is not made clear in the summary that these studies were based upon model
simulations with extent and type of floodplain woodland simulated by manipulating mannings n
values rather than from field studies. Consequently, there is likely to be uncertainty associated
with transferring these outcomes to field conditions. From these studies a value of 300mm is
suggested for the increase in flood water level. However, actual values quoted in Table 2.8 are
50-270mm (River Cary study) and 610mm (River Laver Study).
Conclusions in section 2.3.4
The conclusions regarding Upland Drain Blocking are based upon data obtained from studies
associated with two catchments; the Coalburn, and Blacklaw Moss. In these two studies the
experimental procedure was to introduce ditching to observe the reduction in time-to-peak. The
assumption is made, in the absence of data relating to drain blocking, that the potential increase
in time-to-peak associated with blocking is likely to be of the same order as that caused by
ditching. It is through this rationale that the rule set values have been generated. It is considered
that these assumptions are very weak, introducing a serious degree of uncertainty. This issue,
along with the fact that there is a high level of uncertainty and inconclusive evidence inherent in
the available literature anyway, is made explicit in the text however.
Conclusions in section 2.4.4
These conclusions relate to wetlands and floodplains. The main conclusion is that, while there is
a general consensus that floodplain restoration and enhancement techniques have a positive
influence with respect to flood attenuation, it is very hard to draw general conclusions on the
magnitude of such effects. This conclusion is in agreement with the findings obtained from the
body of literature examined, (in which the results from more than ten studies are cited) because
the range of techniques and magnitude of physical adjustments made amongst these studies
are widely varied thus making it difficult to draw information from one study that is corroborated
by that of another.
Q4) Do you agree with the outlined approach described in Section 3.1?
The outlined approach, in general, appears to be clear in its objectives and the rationale for the
method adopted is sound. It acknowledges the difficulties which arise when devising a generic
NFM scenario-based Tool given the relative paucity of hard data which can be incorporated into
such a model.
There are two major components of the mechanics of the methodology which are of concern
however. The first relates to catchment areas upstream of 'Target Sites' (Tp points) along the
river network; the locations at which the user chooses to examine how upstream management
effects might impact the local hydrograph. The issue is associated with the fact that, given any
Tp, the upstream contributing catchment area is divided into thirds, representing Upper
(headwater), Middle and Lower catchment components, an operation which is undertaken to
enable the differences in influence had by some NFM measures between these three subcatchment area, to be accounted for in the model calculations. Incorporating this idea of subcatchment division into the Tool is sound in principal with respect to observations made in the
pertinent literature. However, division into thirds of, for example, the catchment area above a Tp
point which is located in the true headwater area of the complete catchment under study does
not make physical sense as all sub-catchments above that Tp point are, in effect, located in the
wider river headwaters. The second issue is that the effects had by each management option
selected upstream of a chosen Tp point are simply summed, a condition which is highly unlikely
in practice.
Q5) Do you have any observations in relation to the testing of the model described in
Section 3.3?
The approach adopted for testing the Tool, based upon the use of the River Enrick model, is
considered sound, as this catchment has a rich source of verified modelling data that was
directly applicable to this task. However, there is a question of transferability of the test
outcomes to other catchments/regions in Scotland (this is acknowledged in the text). The use of
the generic form of this model as the basis for generating the floodplain rule sets is reasonable,
given the lack of actual data, but it is still concerning that there are no ground-truth values
which can be applied. Given that this was a simplified model, it is of concern that so few
scenarios were tested to generate the rule-set (refer to Figure 3.5 in that text), there being only
Small (S) and Large (L) storage basin tests, and only one storage depth.
Q6) In your opinion, is the preliminary rule set described in Section 3.4 acceptable and
are there sufficient grounds for the rule sets? Suggest, if possible, alternative ways to
improve the rule set?
Generic conditions:
a) The fact that the Tool uses the SEPA 0.5% and 20% AEP national annual probability
mapping to generate design hydrographs is credible and demonstrates that consideration
has been given as to the ability for uptake of the Tool on a nationwide basis using the same,
currently available, baseline data.
b) The use of scenario 'banding' for hydrograph reduction (optimistic, most likely, and
pessimistic) associated with each NFM technique is a good idea, offering as it does
'sensitivity' bounds for the degree of attenuation associated with each measure. However,
given the limited numerical values available in the pertinent literature these bandings must
be considered with care, and must not be interpreted as a true measure of uncertainty
associated with the results.
c) There is a rule set, which applies to all NFM measures, whereby, if the area covered by a
particular measure is less than the total area available in any given catchment, the
hydrograph reduction potential for that measure is reduced via linear scaling between the
two areas. It is considered that this is a reasonable manipulation given that there appears to
be no data applicable to this issue in the pertinent literature.
Comments on specific rule-sets
Forest Cover
a) The rule set of 10% (most likely), 20% (optimistic) and 0% (pessimistic) attenuation values
for the 0.5% AEP, and 25% (most likely), 40% (optimistic) and 10% (pessimistic) values for
the 20% AEP (refer to Table 3.2 in report) appear reasonable given the data available in the
relevant literature (summarised in Section 2.4) and the model testing results.
b) The breakdown of hydrograph reduction by sub-catchment areas (i.e. upper, middle and
lower catchments), based upon the outputs from the generic model scenario testing (Section
3.3) is considered reasonable. However, this sub-catchment breakdown of hydrograph
reduction strictly only applies to the generic model so these values should be treated with
great care.
c) There is a rule within the model, which is considered reasonable, whereby, if the user
selects an area on the GIS map for forestation (via creation of a polygon) which intersects
with existing forest cover or indeed any other NFM cover (drain blocking etc.), the new area
identified has the existing cover removed before the rule set is applied.
Riparian Woodland
a) The rule set of 0.5% (most likely), 1% (optimistic) and 0% (pessimistic) hydrograph
attenuation values, which apply to both the 0.5% AEP and 20% AEP events (refer to Table
3.3) are in line with the pertinent literature associated with the Pickering and Laver studies
reviewed in Section 2.2.2.
b) The breakdown of hydrograph reduction values by sub-catchment, which have been based
upon the results from the generic Enrick model, are, as discussed in point b under Forest
Cover, susceptible to considerable uncertainty, and should be viewed with caution.
c) The same rule, defined in point c under Forest Cover, applies to Riparian Woodland, which
is again, considered to be a reasonable approach.
Upland Drain Blocking
The basis for the attenuation rule set associated with this management type (Figure 3.4)
appears to be weak, both in terms of the overall attenuation values and the reductions by
catchment sub-units. This issue is, to a degree, associated with conflicting findings in the
literature and also a considerably 'liberal' interpretation of the generic model results. The rule set
associated with Upland Drain Blocking should therefore be highlighted as a component of the
Tool, in its current form, which has significant uncertainty associated with it.
Floodplain Enhancement
For the range of techniques associated with this management option the Tool incorporates a
GIS layer that defines the floodplain extents, based upon the SEPA indicative mapping for the
estimated 0.5% AEP flood inundation extent, from which the user can select floodplain areas for
enhancement. There being no field data, a baseline rule set was derived by combining results
from a range of generic model runs. This baseline rule set is presented in Table 3.5, from which
the user selects, along with its sub-catchment location, a hydrograph reduction value based
upon the channel slope approximately within the range of that where the technique is to be
applied. This value is then adjusted according to the floodplain area selected, an 'enhancement
level' (the depth of floodplain inundation which is considered to bring about a significant
reduction in the design hydrograph), and an 'efficiency factor', which is designed to moderate
the output value according to its potential uncertainty. The variables required for input by the
user are therefore: the area of floodplain selected for application of the measure; the
enhancement level, and; a efficiency factor value. This methodology is utilised by all four
floodplain based management options these being: physically increasing floodplain storage;
increasing floodplain roughness using floodplain woodland; re-connecting the floodplain, and;
setting-back of embankments. All these techniques require the estimation of an enhancement
level. In the case of physically increasing floodplain storage capacity it is suggested that the
level of increase can be broadly estimated by dividing the increase in storage capacity created
by the relevant floodplain area, an assumption that seems reasonable. In the case of increasing
roughness using floodplain woodland it is acknowledged in the text that there is very limited
data, with only one value having been obtained from the literature; a potential increase in level
of 300mm. Table 3.6 presents rule set values of 100mm (pessimistic), 200mm (most likely) and
300mm (optimistic) for this technique, the only value having any physical basis being that of
300mm. These values are therefore considered to be highly unreliable. In the case of reconnecting the floodplain no data appears to have been available to make any judgment upon
enhancement levels such that recourse is taken to suggesting that 'equivalent' increases can be
applied based upon the (already weak, rule set values applied to the increasing roughness
option. As such therefore Table 3.7 presents values for the 20% AEP as per those in Table 3.6,
while values for the 0.5% AEP of -150mm (pessimistic), -50mm (most likely) and 100mm
(optimistic) appear to have no rational basis at all. The rule set values in Table 3.7 are also
applied to the setting-back of embankments option, again with no substantial justification.
Values computed for all of the floodplain enhancement options are then moderated by the
'efficiency factor', a percentage value that is designed as a means of modifying the hydrograph
reduction outputs according to their uncertainty. These values are presented in Table 3.8 where
pessimistic, most likely, and optimistic percentage rule set values are presented for the three
floodplain enhancement measures. This rule set appears to have no rational basis at all and
none is justified.
Consequently, it is considered that the rule set values offered for all floodplain techniques are
highly unreliable, there being three levels of degradation in the knowledge base. This is
because; first, the baseline rule set is derived from values obtained from a generic model;
second, the set of equivalent storage depth values, while having a reasonably sound basis for
the physically increasing floodplain storage capacity option, have a very limited basis in the
case of increasing floodplain roughness, and appear to have no physical basis at all in the case
of re-connecting the floodplain and setting-back of embankments, and; third, the efficiency
values appear to have been set with no justification other than simple guess work. Any results
obtained when considering these techniques in the Tool are therefore considered likely to be
extremely uncertain especially in the case of the 're-connecting' and 'setting-back' options and it
is quite possible that inclusion of these measures in any model run will greatly compromise the
overall predicted results.
It is strongly recommended that, while inclusion of this family of floodplain techniques is
important as management options, with the exception of the physically increasing the floodplain
storage capacity technique, they be excluded from the Tool until such time as any meaningful
data can be brought to bear on the calculations.
Wetlands
Multiple types of wetland (marshy, fen, etc.) are listed as potential independent candidates for
NFM options, and Table 2.10 in the literature survey indicates the NFM role that is likely to be
relevant for each. Within the Tool there is not a set of independent rules associated with each
type rather, suggestions are made as to which of the rule sets associated with the other options
(Forest Cover, etc.) might be used to simulate that type. This scheme is presented in Table 3.1.
This idea is sound in principal given the paucity of appropriate numerical data available to
characterise this parameter. However, given the range of NFM measures suggested for some of
the wetland types, the user is likely to be unsure as to which rule set is truly appropriate for the
catchment site under consideration. It is obviously desirable to make rule-sets as specific as
possible for different NFM measures and, although there may be many different wetland types
to consider, it might perhaps be better to collapse some of these into a smaller number of
broader categories, within which the range of sub-types might potentially benefit from the same
range of NFM options.
Q7) In your opinion is the outline method for coastal floodplains presented in Section 3.6
correct?
The approach outlined for assessing potential locations for managed realignment along the
open coast, if it has been interpreted correctly, is basic very basis. It is simple to implement,
needing just the SEPA national indicative flood mapping dataset loaded into say, ArcGIS, and
the Section 20 Tool does not necessarily have to be used at all. The approach does not actually
help the user to decide where defence breaching might be applicable, and is only correct in as
much as it would simply identify the landward area that would be inundated if breaching were to
be undertaken at a given location.
In simple terms this prospective component of the Tool does not offer any inbuilt, processbased, component that the user might use for testing breach location scenarios and as such it is
not in line with the Tools main functional utility.
The method outlined for estuarine situations has a more quantitative component, which offers a
way to make a broad-brush estimate of the potential change in water level that might occur due
to the introduction of user defined areas of managed realignment. The approach is considered
to be valid, although very crude and would need to be backed up by more intense modelling
studies. Once again, however, it does not necessarily have to be placed within the structure of
the Tool, and, does not have any inbuilt component that can be applied to help the user choose
as where suitable, locations of realignment might lie.
Q8) Do you agree with the limitations specified in Section 3.7 and are there any further
issues not identified?
The key limitations presented in Section 3.7 are all valid. However, a considerable number of
other limitations are not documented (although some are presented elsewhere in the report),
these being:
a) The effect had by multiple measures within a target catchment are purely additive in nature,
and this is not likely to be the case in practice as there will undoubtedly be interactions
between the various measures.
b) The issue associated with catchment partition upstream of any selected Tp point (refer to
Question 1).
c) The rule sets associated with the Floodplain restoration/enhancement/alteration techniques
are extremely tenuous and therefore, as tempting as it is to make the model as 'complete' as
possible by incorporating all possible factors, only the option associated with physically
increasing the floodplain storage capacity ought to be retain (given that it has the strongest
physical basis) and the others removed until field-based corroborative evidence becomes
available.
Q9) Do you agree with the recommendations made in Section 3.8 and do you have any
further recommendations, not identified in Section 3.8 that would be beneficial for
development of the model?
The recommendations are all valid with respect to improving the scientific validity and
functionality of the model.
However, a general observation that may be made is that a number of the recommendations
suggest the incorporation of new rule sets associated with other NFM techniques such as
channel restoration and enhancement. The idea of incorporating such factors is valid but it must
be stressed that they ought not be included until the uncertainty in rule set values associated
with other approaches, especially the Floodplain factors, is reduced significantly.
Referring to the concern raised under Question 1, regarding the validity of splitting the
catchment upstream of any target point into thirds, if this process has been interpreted correctly,
it is strongly urged that it be rectified in some form. A suggested approach would be to have the
GIS divide the entire study catchment into thirds (based upon travel time) and define the
hydrological reduction rule sets above any selected Tp point according to which overall
catchment sub-types are encompassed upstream; i.e. a Tp point located in the overall
catchment headwaters would take on the hydrological reduction conditions of only this sub-type,
while a Tp point located in the true middle reach would contain components of the true middle
and true upper catchment components.
Q10) Do you have any other observations from Section 3 that you would like to
highlight?
a) There is repeated mention of the use of practical experience in generating the rule sets, etc.
but no specifics are offered. If certain values/variables are currently present in the model
which are derived from such sources they should be made as explicit as possible, as
otherwise, there may be attributes which are effectively a 'black box' or which might be
simply 'fudge-factors'. It is quite acceptable to use practical/personal experience in the
absence of hard numbers, but knowledge of where, and how, such information is applied in
the Tool is crucial when it comes to others manipulating and improving on the model code.
b) It is stressed throughout the document that application of the Tool to other catchments is a
relatively simple task, a characteristic that is essential if it is to be applied to catchments
across Scotland. The manner in which this can be implemented is presented in Section 3.2
of the report). However, while this would appear a reasonably simple task, made all the
more compelling by the fact that a pilot catchment is already incorporated into the Tool for
the purposed of demonstration, its transferability is not entirely trivial. In practice, when
applying the Tool to any given target catchment, the necessary DEM data requires reconditioned in Arc, followed by manipulation using ArcHydro, along with updating parameter
values specific to that catchment within the Access database tables. In a practical sense
therefore the prospective user must have a skill-set which encompasses the use of these
various pieces of software.
c) The modified FEH approach utilised in the Tool should be fully documented in the report.
d) In the case of the example catchment currently available in the Tool, the DTM is based upon
NextMAP data but, of course, LiDAR might just as well been used if available. However, it is
strongly urged that a protocol be established as to the resolution of the topography data
which should be applied, if the approach were to be applied nationwide. This is because
ArcHydro is quite sensitive to terrain resolution when it defines the flow paths through a
DEM; information which will influences the calculation of slopes between Tp point and
consequently the apportioning of catchment areas (refer to 'Automation Scripts' in Section
3.2). Therefore, if LiDAR were to be utilised in one catchment and a different resolution of
data in another (or indeed two different resolutions of data in replications of the same
catchment) it is possible that inconsistencies might be introduced.
e) One point to considered, which is purely associated with the analysis presented in the
Section 20 report, is that the figures which display the results output from the River Enrick
test study are poor being both hard to read and interpret. The discussion of these graphs is
also very weak. This is not be a problem with the model per se but these outputs do
represents information that the user may well make reference to when interpreting their own
results.
Q11) Overall, would you recommend the use of the proposed GIS tool and the rule for
use as the screening tool to provide an indication of the effect of NFM techniques?
The approach is conceptually sound in general, and recognises the fact that only simple
processes can be simulated at present using, in general, rule sets based upon results obtained
from the pertinent literature. Also, the calculation processes are not top-heavy with respect to
any one individual sub-process. The use of an ArcGIS platform and associated Access
databases for data manipulation and calculation processes is considered to be appropriate as it
allows access to the Tools internal structure, enabling the user to update rule sets and internal
GIS layers relatively easily. The use of this platform is also appropriate as the software is likely
to be relatively familiar to in-house computer modelling staff.
It is therefore cautiously considered that the Tool could be applied as a broad-brush, high end
NFM scenario assessment tool, but the results must be interpreted intelligently, and be back up
by ground-truth data when targeting specific catchments. However, the Tool should not be
applied as of yet, until the three issues defined below are addressed.
The first, and most important is the concern over the sub-division of catchments above any
given Tp point into upper, middle and lower regardless of where the Tp point exists in the
wider catchment (refer to Question 1). As possible solution for dealing with this is issue offered
under Question 6.
Second, the rule sets associated with the 'increasing floodplain roughness by floodplain
woodland', 're-connecting the floodplain' and 'setting-back of embankments' Floodplain
enhancement techniques (Section 3.4) are considered to be too tenuous at present, and it is
strongly suggested that they be removed until such time as field based, or other data, becomes
available to substantiate these values.
Third, the fact that hydrograph reduction at a given Tp point is based upon the simple
summation of effects had by different measures in the upstream catchment is considered
extremely simplistic, and unlikely to be the case in reality. If any information might be gleaned
from the literature as to the impact had by combined measures this should be considered to
enable incorporation for potential adjustment factors. However, it is acknowledged that if only
tenuous information is available, the uncertainty associated with any adjustment factors might
outweigh that of the possible errors incurred by the simple addition process. Therefore, in the
light of this, it is considered that this issue should be flagged up as a serious limitation in the
current model structure, but does not render it completely unacceptable in its current form.
References relating to NFM techniques (by topic)
Upland Forest Cover
1. Calder, I.R. and Aylward, B. (2006) Forest and floods: Moving to an evidence-based
approach to watershed and integrated flood management. Water International, 31, 1-13.
2. Carroll, Z.L., Bird, S.B., Emmett, B.A., Reynolds, B. and Sinclair, F.L. (2004) Can tree
shelterbelts on agricultural land reduce flood risk? Soil Use and Management, 20, 357-359.
3. Johnson, R.C. (1983) Effects of forestry on suspended solids and bed load yields in the
Balquidder catchments. Journal of Hydrology, 145, 403-417.
4. Mount, N.J., Sambrook Smith, G.A. and Stott, T.A. (2004) An assessment of the impact of
upland afforestation on lowland river reaches: the Afon Trannon, mid-Wales.
Geomorphology, 64, 255-269.
5. Robinson, M. and Newson, M.D. (1986) Comparison of forest and moorland hydrology in an
upland area with peaty soils. International Peat Journal, 1, 49-68.
6. Stott, T.A., Ferguson, R.I., Johnson, R.C. and Newson, M.D. (1986) Sediment budgets in
forested and unforested basins in upland Scotland. In R.F. Hadley (ed.) Drainage Basin
Sediment Delivery
Upland Drain Blocking
1. Birnie, R.V. and Hulme, P.D. (1990) Overgrazing of peatland vegetation in Shetland.
Scottish Geographical Magazine, 106, 28-36.
2. Dunn, S.M. and Mackay, R. (1996) Modelling the hydrological impacts of open ditch
drainage. Journal of Hydrology, 179, 37-66.
3. Robinson M. (1980) The effect of pre-afforestation drainage on the streamflow and water
quality of a small upland catchment. Institute of Hydrology Report No. 73. Wallingford:
Institute of Hydrology.
Grazing (removal) and other upland landuse issues (not including gripping)
1. Bailey-Denton, J. (1861). On the discharge from underdrainage and its effects on the arterial
channels and outfalls of the country. Proceedings of the Institution of Civil Engineers, 21, 48
130.
2. Department for Environment, Food and Rural Affairs (2008) Catchment sensitive farming.
Available at: http://www.defra.gov.uk/farm/environment/water/csf/index.htm (Accessed 8
April 2008).
3. Gifford, G.F. and Hawkins, R.A. (1978) Hydrologic impact of grazing on infiltration: a critical
review. Water Resources Research, 14, 305-314.
4. Jackson, B.M., Wheater, H.S., Mcintyre, N.R., Chell, J., Francis, O.J., Frogbrook, Z.,
Marshall, M., Reynolds, B. and Solloway, I. (2009) The impact of upland land management
on flooding: insights from a multiscale experimental and modelling programme. Journal of
Flood Risk Management, 1, 71-80.
5. Jones, M. (1998) The impact of grazing and upland management on erosion and runoff.
R&D Technical Report P123. Swindon: Environment Agency.
6. Marshall, M.R., Francis, O.J., Frogbrook, Z.L., Jackson, B.M., McIntyre, N., Reynolds, B.,
Solloway, I., Wheater, H.S. and Chell, J. (2009) The impact of upland land management on
flooding: results from an improved pasture hillslope. Hydrological Processes, 23, 464-475.
7. Meyles, E.W., Williams, A.G., Ternan, J.L., Anderson, J.M. and Dowd, J.F. (2006) The
influence of grazing on vegetation, soil properties and stream discharge in a small Dartmoor
catchment, southwest England, UK. Earth Surface Processes and Landforms, 31, 622-631.
8. OConnell, P.E., Ewen, J., ODonnell, G.M. and Quinn, P.F. (2007) Is there a link between
agricultural land-use management and flooding? Hydrology and Earth System Sciences, 11,
96-107.
Appendix B
Summary Table of
Natural Flood Management
Measures
NFM
technique
Type
Conifer
woodland
Fluvial and
pluvial
Location
Principal action
Literature
Model requirements
Upper and
middle
catchment
Canopy interception
reduces net rainfall and
evapotranspiration
reduces antecedent
conditions
With moderate
uncertainty, complete
conifer forest cover
reduces runoff and may
offer 10-20% reductions in
flood peak when compared
to grazing land uses. The
pre-forest land use which
the forest is compared
against is significant.
Sub-type
Runoff
reduction
Research or data
need
Inclusion
Include within
screening for high
runoff areas and
priority for assessment
tool
Fluvial and
pluvial
Runoff
reduction
Upper and
middle
catchment
Canopy interception
reduces net rainfall and
evapotranspiration
reduces antecedent
conditions
Appendix B - 1
Include within
screening for high
runoff areas and
priority for assessment
tool
NFM
technique
Type
Good
muirburn
practice
(compliance
with muirburn
code)
Reducing
grazing
pressure on
pasture
Location
Principal action
Literature
Model requirements
Research or data
need
Inclusion
Upper
catchment
Improving ground
cover through the
provision of a mosaic
of heather cover. The
temporal and spatial
spread of burnt bare
ground ensures a
relatively consistent
ground cover. Good
ground cover increases
interception and slows
overland runoff rates.
May improve
antecedent moisture
condition.
High uncertainty of
runoff reduction.
Absence of literature to
quantify effects, Dunn
(1986) reports a doubling
of flood peaks for whole
catchment burns.
Include within
screening for high
runoff areas and
priority for assessment
tool
Improving the
hydrological condition
of soils (increasing
infiltration)
Moderate uncertainty of
runoff reduction,
magnitude quantifiable.
Include within
screening for high
runoff areas and
priority for assessment
tool
Include within
screening for high
runoff areas and lower
priority for assessment
tool. In the absence of
inclusion within the
assessment tool
provide outline of
potential method to
facilitate local
assessment.
Sub-type
Runoff
reduction
Creation of
cross slope
tree shelter
belts
Catchment wide
(Mainly sloping
intensive
upland areas)
In addition to benefits
of forest (increased
interception and
reduced antecedent
moisture) runoff from
upslope can be
infiltrated at the tree
shelter belt
Appendix B - 2
NFM
technique
Type
Creation /
restoration of
nonfloodplain
wetlands
Location
Principal action
Literature
Model requirements
Research or data
need
Inclusion
Principally
catchment
headwaters
It is commonly
misstated that wetlands
act as slow release
sponges. They
frequently act as
sources of flashy runoff
due to the high
antecedent moisture
condition but in some
cases can serve to
attenuate flows.
Detailed continuous
simulation hydrological
model which includes an
accurate groundwater
modelling component.
Would require a large
amount of data and
technical input to
accurately represent a
single wetland.
Include within
screening of high runoff
areas and lower priority
for assessment tool. In
the absence of
inclusion within the
assessment tool
provide outline of
potential method to
facilitate local
assessment.
Moderate uncertainty of
runoff reduction,
magnitude quantifiable.
Requires data of
existing land
management practices
Include within
screening of high runoff
areas and priority for
assessment tool
Requires detailed
mapping of existing
drains
Include within
screening of high runoff
areas and lower priority
for assessment tool
Sub-type
Runoff
reduction*/delay
, attenuate
Reducing soil
compaction
in arable
areas,
improving
soil texture
and reducing
bare earth in
wetter
seasons
Mid-catchment
Runoff
reduction
Fluvial (&
pluvial)
Runoff
reduction
(Rapid shallow
groundwater)
Appendix B - 3
Complex distributed
hydrological model with
shallow groundwater
component allowing the
representation of
modified antecedent
conditions and the
accelerated shallow
water flow where pipes
are present.
Arterial ditches would
require a flexible flow
routing model that allows
the presence of known
arterial ditches to be
represented.
NFM
technique
Type
Upland drain
blocking
Fluvial
Floodplain
reconnection
Fluvial
Location
Principal action
Literature
Upper
catchment
Drains provide an
improved route for
water to reach
watercourses (but can
conversely lower
antecedent moisture
levels). Upland drain
blocking is a complex
issue which requires
consideration of soil
type, topography and
proximity to
watercourse.
In floodplains,
mostly lower
and middle
parts of
catchments
where
floodplains are
sufficiently wide
to merit flood
protection
embankments
Extends catchment
time to peak by slowing
travel time through the
catchment. Can allow
the desynchronisation
of sub-catchment
peaks. Often
associated with river
restoration which can
also slow down the
flow of water.
Model requirements
Research or data
need
Inclusion
Requires detailed
mapping of existing
upland drains
Include within
screening of high runoff
areas and lower priority
for assessment tool
Requires detailed
topographic data to
enable modelling.
Include within
screening of
channel/floodplain
areas and lower priority
for assessment tool. In
the absence of
inclusion within the
assessment tool
provide outline of
potential method to
facilitate local
assessment.
Sub-type
Runoff
reduction,
delay/desync
Attenuation ,
delay/desync
Appendix B - 4
Preferable to use a
distributed or semidistributed hydrological
model which
incorporates a detailed
hydraulically based flow
routing engine
Requires development
of an appropriate
hydrological model
which includes a
detailed hydraulic flow
routing engine (1D-2D,
diffuse wave)
NFM
technique
Type
Creation of
washlands
Fluvial
Creation of
constructed
farm
wetlands or
ponds
Location
Principal action
Literature
Model requirements
Research or data
need
Inclusion
Mostly in lower
and middle
parts of
catchments
where
floodplains are
wide
Increases the
availability of storage of
flood water within the
catchment, thus
attenuating or delaying
the flood peak.
Engineer arrangements
can serve to reserve
volume for flood peak.
Preferable to use a
distributed or semidistributed hydrological
model which
incorporates a detailed
hydraulically based flow
routing engine
Requires detailed
topographic data to
enable modelling.
Include within
screening of
channel/floodplain
areas and lower priority
for assessment tool. In
the absence of
inclusion within the
assessment tool
provide outline of
potential method to
facilitate local
assessment.
On minor
watercourses
and ephemeral
flow routes
Preferable to use a
distributed or semidistributed hydrological
model which can allow
the hydraulic
representation of the flow
controls.
Requires detailed
topographic data
Sub-type
Attenuation,
delay/desync
Fluvial (and
pluvial)
Attenuation,
delay/desync
Appendix B - 5
Can be represented
using existing models
Requires development
of an appropriate
hydrological model
which includes a
detailed hydraulic flow
routing engine (1D-2D,
diffuse wave)
NFM
technique
Type
Afforestation
of floodplains
Fluvial
Location
Principal action
Literature
Model requirements
Research or data
need
Inclusion
In floodplains,
mostly lower
and middle
parts of
catchments with
existing
floodplains
Extends catchment
time to peak by slowing
travel time through the
catchment. Can allow
the desynchronisation
of sub-catchment
peaks. Often
associated with river
restoration which can
also slow down the
flow of water.
Preferable to use a
distributed or semidistributed hydrological
model which
incorporates a detailed
hydraulically based flow
routing engine
Requires detailed
topographic data to
enable modelling.
Include within
screening of
channel/floodplain
areas and lower priority
for assessment tool. In
the absence of
inclusion within the
assessment tool
provide outline of
potential assessment
method to facilitate
local assessment.
Sub-type
Attenuation,
delay/desync
Requires development
of an appropriate
hydrological model
which includes a
detailed hydraulic flow
routing engine (1D-2D,
diffuse wave)
Fluvial
Attenuation or
delay/desync
Catchment
wide, typically
lower reaches
where naturally
sinuous rivers
have been
straightened
Appendix B - 6
Preferable to use a
distributed or semidistributed hydrological
model which
incorporates a detailed
hydraulically based flow
routing engine
Include within
screening of
channel/floodplain
areas and low priority
for assessment tool. In
the absence of
inclusion within the
assessment tool
provide outline of
potential assessment
method to facilitate
local assessment.
NFM
technique
Type
Creation of
riparian
woodland
Fluvial
Location
Principal action
Literature
Model requirements
Research or data
need
Inclusion
Catchment wide
Assessment requires a
distributed hydrological
model that has an
integrated detailed
hydraulic flow routing
engine that is capable of
representing the
increased channel
roughness and should
allow for channelfloodplain flow linkage
A high degree of
uncertainty particularly for
cases relying on improving
floodplain connection.
Effects are marginal and
there are indications that
incorrect woodland
placement can increase
risk.
Guidance on the
natural spacing of large
woody dams and how
the Mannings
roughness values vary
with slope, channel
size and vegetation
type.
Include within
screening and low
priority for assessment
tool. In the absence of
inclusion within the
assessment tool
provide outline of
potential assessment
method to facilitate
local assessment.
Geomorphological aspects
are very uncertain
depending on local
circumstances.
Sub-type
Delay/desync or
attenuation
Fluvial
Delay/desync or
attenuation
Upper
catchment
gullies or high
energy
watercourses
(potential
gullies)
Assessment requires a
distributed hydrological
model that has an
integrated detailed
hydraulic flow routing
engine that is capable of
representing the
increased channel
roughness and should
allow for channelfloodplain flow linkage
Ability to modify runoff
generation in a
proportion manner with
flow routing element to
estimate accumulated
effects
Appendix B - 7
Include within
assessment via
approaches for riparian
woodland and the
management of
instabilities.
NFM
technique
Type
Location
Principal action
Literature
Model requirements
Research or data
need
Inclusion
Placed large
woody debris
and boulders
Fluvial
Mid catchment
Attenuation or
delay/desync
In channel,
upstream of
flood receptors
Slowing in channel
flows and also
increasing floodplain
connectivity
Assessment requires a
distributed hydrological
model that has an
integrated detailed
hydraulic flow routing
engine that is capable of
representing the
increased channel
roughness and should
allow for channelfloodplain flow linkage
Requires guidance on
the selection of
Mannings roughness or
other modelling
technique
Include within
screening and low
priority for assessment
tool. In the absence of
inclusion within the
assessment tool
provide outline of
potential assessment
method to facilitate
local assessment.
Improve quality of
artificial structures
database
Include within
screening of
channel/floodplain
areas and not included
within assessment tool.
Provide outline of
potential assessment
method to facilitate
local assessment.
Sub-type
Fluvial
Convey/protect
At or
immediately
downstream of
receptors where
the channel is
artificially
constricted
Can assist in
conveying water past
receptors
Appendix B - 8
Large number of
interventions required to
generate significant
impact.
NFM
technique
Type
Managing
channel
instabilities
Fluvial
Location
Principal action
Literature
Model requirements
Research or data
need
Inclusion
Catchment wide
within the river
channel and at
flood receptors
Completion of the
sediment budget
modelling
Include within
screening via sediment
budget modelling being
developed by the
hydromorphology
team. Do not include
within assessment tool
but provide outline of a
potential methodology
for local assessment.
Sub-type
Protection
(When flooding
can be
attributed to
geomorphologic
al activity such
as channel
aggradation)
Instabilities can
be natural or
caused by
humans.
Use of SUDS
Urban drainage
Reduce, delay
and attenuate
Entire urban
drainage
system
By increasing
attenuation, infiltration
and mimicking natural
processes the impact
of urban development
on the water
environment can be
reduced. Reduction of
discharges to
greenfield rates
Development of new
models, or validation of
existing dynamic
models (eg CAESAR)
that are able to predict
rates of aggradation,
rates of channel
movement etc.
Hydraulic impact of
estimated or observed
aggradation can be
assessed using existing
models
Predicting rate of
aggradation or incision,
or locations of avulsion,
requires complex and
little-tested 2D dynamic
modelling and carries
high uncertainty.
Appendix B - 9
Existing assessment
techniques are
appropriate
NFM
technique
Type
Using high
water
demand
vegetation to
reduce
groundwater
levels
Groundwater
Beach
management
(beach
recharge
schemes and
shingle
management
)
Coastal
Artificial/
biogenic
reefs and
detached
breakwaters
Coastal
Location
Principal action
Literature
Model requirements
Research or data
need
Inclusion
Entire aquifer
(benefits for
locations where
groundwater is
close to the
surface)
Existing groundwater
modelling techniques are
appropriate for making
assessments
National mapping of
groundwater flood risk
would be beneficial
Typically
exposed
coastal areas
where beaches
are eroding. If
the source of
material is
related by
coastal
processes to
the eroding
area then this
approach is
termed
recycling.
Artificial addition of
sand or gravel material
to maintain beach
levels. Increased width
and reduced slope of
beach dissipates wave
energy. Generally not
possible in many areas
as depends on the
availability of large
quantities of suitable
material from local
sources e.g. port and
harbour dredging.
Often needs hard
engineering structures
to keep material in
place and often needs
to be replenished as
material erodes away
over time.
Include within
screening. Dont
include within
assessment tool.
Provide outline of
methodology which can
be used for local
assessment.
Exposed
coastal areas
Establishment of
biogenic reefs e.g.
oyster and mussel
beds to dissipate wave
energy.
Include within
screening. Dont
include within
assessment tool.
Provide outline of
methodology which can
be used for local
assessment.
Sub-type
Reduce
recharge
Protect
Protect
Detached breakwaters
reduce wave energy
reaching the shoreline
but do not completely
isolate natural beach
processes.
Appendix B - 10
NFM
technique
Type
Sand dune
restoration
(e.g. dune
fencing and
thatching,
marram
grass
planting).
Coastal
Standline
and beach
management
techniques
Coastal
Creation/
restoration of
intertidal
area
including
mudflats and
saltmarsh
Coastal
Location
Principal action
Literature
Model requirements
Exposed
coastal areas
behind sandy
beaches were
sand dunes
would naturally
form
Creates a defence
between the beach and
land to dissipate
largest waves and a
direct defence to high
water levels
Existing assessment
techniques are
appropriate
High amenity
beaches.
No known assessment
techniques. The validity
of the approach would
need to be assessed
using the judgement of
an experienced
geomorphologist
Due to absence of
literature there is a high
degree of uncertainty
about the magnitude of
effects.
Areas where
flats would
naturally form,
typically
estuarine
Creates an improved
buffer to dissipate
wave energy.
Sub-type
Protect
Attenuate
Protect
Research or data
need
Include within
screening. Dont
include within
assessment tool.
Provide outline of
methodology which can
be used for local
assessment.
Areas of
reclaimed land
and
uneconomic
flood defences.
Areas of
degraded
mudflats and
saltmarsh
Appendix B - 11
Up-to-doubling of wave
energy dissipation.
Inclusion
There is an absence of
literature detailing the
effectiveness of the
altered maintenance
approach
Include within
screening. Dont
include within
assessment tool.
Provide outline of
methodology which can
be used for local
assessment.
NFM
technique
Type
Managed
realignement
, creation/
restoration of
intertidal
area
including
mudflats and
saltmarsh
Coastal
Location
Principal action
Literature
Model requirements
Research or data
need
Inclusion
Typically mid
and upper
estuaries.
Provides increased
volume for the storage
of estuarine surges.
The increased flooded
area also increases the
resistance to flow
further helping to slow
propagation up
estuary.
Moderate uncertainty
Baseline hydraulic
models of all estuaries
linked to potentially
vulnerable zones would
be beneficial.
Include within
screening. Dont
include within
assessment tool.
Provide outline of
methodology which can
be used for local
assessment.
Moderate uncertainty
Sub-type
Delay/attenuate
Areas of
reclaimed land
and
uneconomic
flood defences.
Allows natural
processes to continue
with no active
intervention. Ensures
natural morphological
processes are
sustained.
Regulated
tidal
exchange
Coastal
Attenuate
Typically
estuarine
Areas of land
claim and
uneconomic
flood defences.
Provides increased
volume for the storage
of estuarine surges.
Can form buffer to
dissipate wave energy.
Include within
screening. Dont
include within
assessment tool.
Provide outline of
methodology which can
be used for local
assessment.
Coastal
Delay/attenuate
Typically
estuarine
Areas where
artificial
channels have
been
created/maintai
ned for shipping
or drainage
purposes.
Dredging of channels
can be detrimental to
surrounding flats
Appendix B - 12
Effects unknown.
A dataset of artificially
sustained channels
would be beneficial
Appendix C
Other considerations
Draft
Neil Nutt
Blockage risk
It was identified that culverts and bridges with openings greater than approximately 6m are less
prone to blockage during extreme flood events (Rigby et al., 2002)1. The study was based on
four forested catchments within Australia following a large flood event. Despite the blockage
condition of 92 culverts being observed no correlations between any of the investigated factors
could be identified.
A more recent study of blockage risk at 140 screened culverts in Belfast (Wallerstein et al.,
2010)2 found that organic material such as leaves, twigs and branches form the bulk of material
found to be captured at culvert screens. The study identified that there was a strong seasonal
trend for blockage caused by small organic material such as leaves, with the highest risk being
in November and correlating with the autumn leaf drop from deciduous trees.
It is proposed in Johnson (2007)Error! Bookmark not defined. that the use of floodplain
woodland can serve to trap floating debris helping to reduce the risk of blockage at downstream
culverts. Nisbet et al (2011)Error! Bookmark not defined. that short rotation forestry, and in
particular short rotation coppice, may be particularly effective at entrapping floating debris
before it forms blockages at culverts.
In summary, while there is little scientific information on the subject, it would seem prudent that
due consideration is given to the impact of increasing the supply of organic material might have
to the blockage of constricted downstream structures or those which are fitted with a screen.
Rigby E.H., Boyd M.J., Roso S., Silveri P., Davis A., Causes and effects of culvert blockage
during large storms, Forbes Rigby Consultants & Wollongong City Council, 2002.
2
Wallerstein N., Arthur S., Sisinnggih D, Towards predicting flood risk associated with debris at
structures, (unpublished), 2010.
3
ARUP Water, Water of Leith Flood Prevention Scheme, Hydrological and hydraulic modelling
Design Report, Volume 1. City of Edinburgh Council, 2003.
Appendix C - 1
within their spillways so that reservoir storage could be reserved for the flood peak. The use of
the three reservoirs is understood to reduce the 1 in 200yr plus climate change flow from
approximately 108cumecs to approximately 76cumecs significantly reducing the extent and high
of traditional direct defences downstream.
It is understood that Scottish Borders Council proposes to modify the existing outlet control
structure at St Marys Loch to both reduce rural flooding along the Yarrow Water and to provide
compensatory storage for the areas of active floodplain that will be lost through the construction
of direct defences as part of the Selkirk Flood Protection Scheme (Halcrow, 2011)4. The
reduction of the lochs normal operating level by approximately 250mm and the raising of the
existing flood spillway by 250mm will significantly reduce rural flood damages along the Yarrow
Water and will more than compensate for the loss in floodplain volume as a consequence of
direct defences within the town of Selkirk. The estimated reduction in flood flows in the Yarrow
Valley range between 10-30% with the effect reducing by event magnitude. The reduction in
water storage available for release as freshets or compensation flow is overcome via the
replacement of a manually controlled release with an automated system.
Backwater effects
A number of natural flood management measures have been highlighted as having a potential
impact on upstream flood risk through the creation of backwater effects. A Forest Research
paper on restoring floodplains for flood alleviationError! Bookmark not defined. found for one
model based case study the backwater could extent upstream between 120-330m upstream of
introduced forest floodplain but that the effect would be primarily dependent on channel gradient
in addition to the increase in water depth. The potential to increase water levels upstream
highlights the need to give due consideration to the placement of in flow obstructions (such as
floodplain woodland, in channel interventions or riparian vegetation). Unfortunately there is an
absence of appropriate literature discussing what the key variables controlling the extents are.
Therefore a look at the issue from first principles is presented below to provide an initial
overview.
The rate of change of flow depth within a gradually varied flow is defined by Equation 1. Where
the Froude Number ( ) is defined in Equation 2 and the hydraulic gradient (
approximated using Mannings Equation (Equation 3).
) can be
Equation 1: The general differential equation of the water surface profile for gradually varied flow
Halcrow Group Limited, Selkirk Flood Protection Scheme; St Marys Loch Flood Storage
Option; Feasibility Study. Scottish Borders Council, 2011.
Appendix C - 2
), flow (
), wetted perimeter (
) in terms of
).
By rearrangement and integration of Equation 1 the change in distance between two depths can
be expressed as shown in Equation 4.
By considering the channel geometry, channel slope and flow to be constant in space and time
it can be concluded that the backwater effect diminishes asymptotically with the extent being
dependent on channel slope, roughness, flow and deviation above the normal depth as follows:
The backwater extent is inversely related to channel slope, that is shallower slopes will
generate longer back water extents
Further discussion and analysis are presented in the following draft paper.
Appendix C - 3
Abstract
There is currently great interest in the use of
natural processes to manage flood risk.
One such process is the use of floodplain
woodland to increase floodplain roughness
and thus help attenuate floods. Concern
has been raised that the elevated water
levels (backwater) generated by floodplain
roughening could propagate a significant
distance upstream thus increasing upstream
flood risk. In an absence of published
material on the extent of backwaters this
paper will investigate the nature of these
backwater profiles via first principles.
Introduction
There is increasing interest in the use of
natural processes to manage flood risk
(Scottish Government, 2009; Pitt, 2008).
Woodland for Water (Nisbet et al, 2011)
discusses the practice of using vegetation
roughen floodplains thus serving to slow the
flow of flood water and increasing the depth
of flood water that can be held on the
floodplain. The elevated water profile
(backwater) backwater created by this
mechanism has been highlighted as having
the potential to extend a significant distance
upstream leading to an increase in
upstream flood risk (Thomas & Nisbet,
2006; Nisbet & Thomas, 2008). However,
there is an absence of literature on the
subject on the factors which influence the
magnitude and extent of these backwaters.
It is proposed to investigate the factors
which influence the extent and magnitude of
backwater effects from first principles.
Draft
) in terms
), flow (
),
).
Neil Nutt
Source
(Leopold and Maddock,
1953)
0.015<n<0.2
Conveyance
Estimation System
(Samuels et al., 2002)
0.01<Q<5cumecs Broad range in
absence of guidance
from literature
0.05<dy<0.5m
(Thomas & Nisbet,
2006; Nisbet &
Thomas, 2008)
Table 1: Variable ranges and sources used
in the analysis
Appendix C - 2
Discussion
The analysis shows that floodplain
roughening on shallow gradient rivers has
the greatest potential to generate an
extensive backwater. Increasing roughness
and flow also serve to increase backwater
length but their effects are less marked.
The magnitude by which the water depth is
increased over the normal depth does serve
to increase the backwater length, however
the extent is largely insensitive to small
changes.
Conclusion
Through the consideration of first principles,
this study has found that the backwater
Appendix C - 3
References
Blanger J.B., (1828) "Essai sur la Solution
Numrique de quelques Problmes Relatifs
au Mouvement Permanent des Eaux
Courantes." ('Essay on the Numerical
Solution of Some Problems relative to
Steady Flow of Water.') Carilian-Goeury,
Paris, France (in French).
Chanson, H., (2008) "Jean-Baptiste Charles
Joseph Blanger (1790-1874), the
Backwater Equation and the Blanger
Equation." Hydraulic Model Report No.
CH69/08, Div. of Civil Engineering, The
University of Queensland, Brisbane,
Australia, 40 pages
Leopold L.B. and Maddock T., (1953)The
Hydraulic Geometery of Stream Channels
and Some Physiographic Implications.
United States Department of the Interior.
Nisbet T.R., Silgram M., Shah N., Morrow
K., Broadmeadow S., (2011) Woodland for
water: Woodland measures for meeting
Water Framework Directive objectives.
Forest Research.
Nisbet T.R., Thomas H., Restoring
Floodplain Woodland for Flood Alleviation,
DEFRA, 2008.
Appendix C - 4
Appendix D
Requirements on SEPA in relation to Natural Flood
Management under the FRM Act
Amendments 9 to 11, 17 to 24, 35, 37, 38, 40 to 42, 53, 57, 58, 61, 63 and 64 relate directly to
the assessments of natural features that will be made under section 16. I will lodge separate
amendments to later sections that deal with wider issues concerning the role of natural flood risk
management and its relationship to sustainable flood risk management. I will refer to those
amendments when I address the amendments that committee members have lodged.
We all appreciate that the evidence base supporting the use of more natural approaches to
managing flooding is still limited and evolving. However, available information suggests that a
number of techniques show significant promise. Furthermore, we should aim to take advantage
of the added benefits that can be gained from adopting more natural approaches to managing
flooding, which include environmental and social benefits.
The Government is in the process of developing a long-term research strategy to co-ordinate
investment in improving our understanding of natural flood management techniques. We have
also set up a stakeholder group to advise the Government on natural flood management. Those
important steps will contribute to the cultural shift that, as the committee has rightly highlighted,
must occur.
The Scottish Government has worked closely with stakeholders on considering the role of
natural flood management and the committee's recommendations on section 16. The
amendments to which I will speak have been drafted in close consultation with stakeholders,
including SEPA, Scottish Natural Heritage and Scottish Environment LINK.
Amendments 18, 23 and 24 mean that the assessments that will be prepared under section 16 will
consider not only natural features but how alterations or restoration of natural characteristics of
river basins and coastal areas could contribute to managing flood risk. The reference to river
basins indicates that the emphasis is clearly on taking a catchment-based approach.
Amendments 19 and 20 replace the examples of natural features that are set out in the bill with a
new set of examples that covers natural features and characteristics of river basins and coastal
areas. In setting out those examples, we have given particular consideration to ensuring that they
express some of the key concepts and aims of natural flood management, including using natural
features and characteristics to assist in the retention or slowing of flood water. I stress that the
examples that are set out in the amendments are not intended to form an exhaustive list. In
compiling them, the Government liaised closely with key stakeholders, including Scottish
Environment LINK and SEPA.
are recognised as flood protection measures when they contribute to managing flood risk. I am
confident that those amendments, combined with the definition of flood protection work, which
covers
"any work of construction, alteration, improvement, repair, maintenance, demolition or
removal",
will be wide enough to encompass
"the alteration ... or restoration of natural features and characteristics of any river basin or coastal
area".
For those reasons, I do not support Rhoda Grant's amendment 88 and I ask her not to move it.
Amendment 69 was lodged by Elaine Murray. I agree fully that the assessments that are prepared
under section 16 should be subject to consultation with local authorities, other responsible
authorities and other key organisations, including SNH. That is why the bill requires SEPA to
have regard to the advice of advisory groups when preparing assessments. I assure the member
that the bodies that her amendment names will need to be involved in the preparation of
assessments and that SEPA will be required to have regard to their advice. To initiate a separate
consultation exercise would risk undermining the important role that is envisaged for advisory
groups. For those reasons, I do not accept amendment 69. The bill already makes adequate
provision in that respect, therefore I ask the member not to move the amendment.
Amendment 85, which was lodged by Peter Peacock, does not relate to the content of
assessments that are prepared under section 16. Instead, it relates in part to how the information
that is garnered from assessments is used when SEPA sets objectives and measures to manage
flood risk for inclusion in flood risk management plans. I would first like to clarify that the bill
already requires SEPA to have regard to section 16, along with various other important factors,
when it identifies objectives and measures.
The other aspects of Peter Peacock's amendment would require SEPA to select the most
sustainable measures and to give reasons for selecting measures that are inconsistent with things
that are identified through assessments that are made under section 16. On the first point, I will
later discuss a proposed Government amendment that will require SEPA to select measures that
will achieve objectives in the most sustainable way. On the second point, I fully support the
principle that SEPA should be required to set out reasons for selecting particular measures and I
believe that that principle should apply to all measures, not only to those that relate to section 16.
I ask Peter Peacock not to move amendment 85, on the understanding that I will bring forward
an amendment at stage 3 to require SEPA to set out the reasons behind the selection of measures.
I move amendment 9.
Peter Peacock: I welcome all
where natural approaches can be used to remove the peaks of floods and reduce the need for hard
engineering solutions downstream, they should be consideredI stress the word "considered".
The stage 1 report is clear that natural flood management techniques will not work in every
circumstance and are therefore not a panacea. The committee was equally clear that, as others
have mentioned, a change of culture is required in the consideration of future flood management
approaches. If we have a system for developing flood risk management plans that is dominated
by engineersestimable people though they are, and I include John Scott in thatwe should not
be surprised if the subsequent debate is dominated by engineering solutions or arrangements for
flood management.
Recommendation 15 in the committee's stage 1 report calls for measures that are likely to cause
the shift in culture that the committee feels to be necessary, but which are not prescriptive about
the use of natural flood management techniques. The committee wants to
"require responsible authorities to consider what contribution natural flood management
approaches could make",
and recommends that
"Such an amendment should stipulate that, where natural flood management approaches are
assessed as being able to make such a contribution but are not proceeded with, authorities must
set out the reasons for that decision."
Amendment 27, which the minister mentioned and which we will debate later, is the
Government's way of addressing the issue. However, it does not go far enough and it is not
explicit enough. Amendment 85 would deliver the committee's unanimous recommendation, and
I hope that it embraces the purpose of the Government's amendment 27.
Amendment 85 is both reasoned and reasonable and I urge the committee to support it. It seeks
to ensure that natural flood management techniques are considered, but it would not require
action using those techniques to be taken above any other action. Under amendment 85, when
the contribution that natural techniques could make was considered and it was decided not to use
them, the reasons for the decision would have to be set out. The amendment would not require
natural flood management techniques always to prevail, which is an understandable concern of
the Government. It would provide for the necessary considerations to bring about the change in
culture to which the committee has referred.
I noted the minister's comments about the intention to return at stage 3 with an amendment to
take care of the reporting aspects. I welcome that, but it does not deal with the central point of
amendment 85of which reporting is a partwhich is the requirement to consider natural
approaches, although not necessarily to implement them. I stress that point, which would give
discretion to SEPA and local bodies.
Liam McArthur: The
hard engineering is not the approach with which local authorities and SEPA naturally kick off
the debate about flood risk management.
I am inclined to support amendment 85. Natural flood management was central to
the committee's report and, if amendment 85 would help to deliver that, it is important.
Notwithstanding the minister's intention to introduce amendments at stage 3, amendment 85 is
worthy of support.
John Scott:
Appendix E
Examples Effect of land use on runoff generation
Initial estimation of routing model coefficients
1.
The proposed runoff generation model has been tested for a range of land use types
over a range of rainfall depths. The testing has been undertaken using HOST class
19 as this is a relatively common soil class that is susceptible to degradation by land
use. It should be noted that this analysis represents the effect of land use on the
runoff generated by a single cell within the proposed hydrological model and does
not provide a complete statement on the flood flow generated by a catchment.
The variation in initial soil moisture with land use has been represented via the
modification of PROPWET using the following equation:
LAI for
PROPWET
LAI for
r
interception
BFI
(HOST)
PE
SAAR
Baseline
PROPWET
from donor
catchment
6
6
0.47 (19)
6
6
70
80
0.47 (19)
0.47 (19)
500mm/yr
500mm/yr
1200mm/yr
1200mm/yr
80
0.47 (19)
500mm/yr
1200mm/yr
60
0.47 (19)
500mm/yr
1200mm/yr
60
0.47 (19)
500mm/yr
1200mm/yr
40
0.32 (22)
500mm/yr
1200mm/yr
40
0.32 (22)
500mm/yr
1200mm/yr
70
0.47 (19)
500mm/yr
1200mm/yr
Conifer forest
Broad leaf forest
(summer)
Broad leaf forest
(winter)
Unimproved
pasture
Semi improved
pasture
Improved
pasture
Potatoes (winter
bare earth)
Conifer forest
(with
HoyningenHunene canopy)
The testing is presented for a range of rainfall depths (15mm to 150mm) and it is
intentional that no guidance has been provided on what event rarity these depths
represent. As discussed in the FEH vol 2 the rarity of a given rainfall depth is related
to the locality, storm area and storm duration. The range of rainfall depths has been
selected as the lower end is broadly representative of a frequent flood event in a
small catchment through to a relatively rare flood event in a large catchment at the
upper end.
The results of the testing are presented in Figures 1 to 6 for the selected range of
rainfall depths. These results are then summarized in Figure 7 and 8 to provide an
indication of the impact land use has on runoff generation. In summary the analysis
indicates that the impact of soil degradation increases with event magnitude, while
the impact of forestation decreases. The analysis introduces an interesting question
of that to compare the results to. While a baseline scenario has been presented
using PROPWET from a selected donor catchment it is cannot be stated that this
donor PROPWET is representative without further work.
It should be stressed that these results are provided as simply a guide to the general
trends which the proposed method might produce. There is still a large requirement
for calibration and verification (and in the case of PROPWET a method selected)
therefore the magnitude of effects reported by the final tool could be significantly
different.
1.8
Broad leaf forest summer
Runoff depth per time interval (mm)
1.6
Broad leaf forest winter
1.4
Semi improved pasture
1.2
Improved pasture
(degraded soil)
1.0
Unimproved pasture
0.8
Potatoes (winter bare
earth) (degraded soil)
Forest (Hoyningen-Huene
interception)
Baseline
0.6
0.4
0.2
0.0
0
10
15
Figure 1: Runoff depth generated for each time interval (13 interval storm) for a range
of land use types for a 15mm rain storm
4.0
Runoff depth per time interval (mm)
3.5
Broad leaf forest winter
3.0
Semi improved pasture
2.5
Improved pasture
(degraded soil)
Unimproved pasture
2.0
1.5
1.0
0.5
0.0
0
10
15
Figure 2: Runoff depth generated for each time interval (13 interval storm) for a range
of land use types for a 30mm rain storm
9.0
8.0
7.0
6.0
Improved pasture
(degraded soil)
Unimproved pasture
5.0
4.0
3.0
2.0
1.0
0.0
0
10
15
Figure 3: Runoff depth generated for each time interval (13 interval storm) for a range
of land use types for a 60mm rain storm
16.0
Runoff depth per time interval (mm)
14.0
Broad leaf forest winter
12.0
10.0
Improved pasture
(degraded soil)
Unimproved pasture
8.0
6.0
4.0
2.0
0.0
0
10
15
Figure 4: Runoff depth generated for each time interval (13 interval storm) for a range
of land use types for a 90mm rain storm
25
Runoff depth per time interval (mm)
20
15
10
0
0
10
15
Figure 5: Runoff depth generated for each time interval (13 interval storm) for a range
of land use types for a 120mm rain storm
30
Runoff depth per time interval (mm)
25
20
15
10
5
0
0
10
15
Figure 6: Runoff depth generated for each time interval (13 interval storm) for a range
of land use types for a 150mm rain storm
Baseline
Conifer forest
25mm
20mm
15mm
Semi improved pasture
10mm
5mm
mm
mm
20mm
40mm
60mm
80mm
Figure 7: Peak runoff depth for a range of land use types over a range of rain storm
depths
Baseline
130%
Conifer forest
120%
110%
100%
Unimproved pasture
baseline
90%
80%
70%
60%
mm
20mm
40mm
60mm
80mm
Figure 8: Percentage change in the peak runoff depth for a range of land use types
over a range of rain storm depths
2.
Initial testing of the flow routing model has been undertaken for a collection of
Scottish gauged catchment with reported observed catchment lag times in the Flood
Estimation Handbook vol 4 (1999)i as presented in Table 1. The aim of this initial
testing phase was to demonstrate that the flow routing model can demonstrate
reasonable potential in matching the observed data.
Catchment ID
Name
Observed lag
(hrs)
Number of
observed events
19001
Almond
8.2
5
20001
Tyne
10.8
10
21030
Megget Water
6.9
4
77002
Esk
7.3
46
79006
Nith
7.7
25
84002
Calder
3
4
84008
Rotten Calder Wtr
5.1
7
84012
White Cart Water
7.6
6
85002
Endrick Water
5.7
4
Table 1: Observed catchment lag times published in the FEH vol 4 for a selection of
Scottish catchments
The first step was to identify an appropriate initial estimate of the sinuosity factor,
which would be taken to be a universal constant for both the channel and land flows.
The initial estimate was made by investigating the impact of DTM grid resolution on
the drainage path length. In the absence of whole catchment coverage by LiDAR it
was necessary to resort to the 5m NextMap DTM. The selected catchment area was
the River Teviot at Hawick being a moderate sized catchment with no open
waterbodies. The effects of changing grid resolution on the maximum drainage path
length (LDP) and the average (DPLBAR) are presented in Table 2. The analysis
assumes that the extrapolated drainage path length using a 1m grid resolution is
representative of the actual drainage path length. It is acknowledged that the
analysis is sensitive to both the function used for the extrapolation and the selection
of 1m being representative of the true answer, which could be argued to be only
reached with a resolution of 0m however this could indicate an unpractical answer of
infinity.
Grid
resolution
(m)
LDP
DPLBAR
Fitted
LDP
Calculated
Fitted
DPLBAR
Calculated function sinuosity DPLBAR
function sinuosity
LDP (m)
(m)
(unitless)
(m)
(m)
(unitless)
(below
(below
1
resolution)
36979
1
resolution)
16456.49
5
34539.5
34727.4
1.064837
16363.2
16211.12
10
33872.4
33757.68
1.095425
16110.9
16027.26
25
32900.9
32475.79
1.138664
15823.1
15662.45
50
31755.5
31506.08
1.17371
15371.3
15251.32
100
30112
30536.37
1.210982
14532.6
14669.9
150
29643.3
29969.12
1.233903
14019.3
14223.76
200
29350.4
29566.65
1.2507
13585.2
13847.64
250
29281.2
29254.48
1.264046
13214.8
13516.28
350
28814.1
28783.75
1.284718
12789.2
12941.22
400
27770.6
28596.94
1.29311
12310.9
12684.8
500
29420.3
28284.76
1.307382
13127
12216.18
Table 2: The effect of DTM resolution on the estimated drainage path lengths
1
1.015136
1.026781
1.050697
1.079021
1.121786
1.156972
1.188396
1.217531
1.271634
1.297339
1.347106
1.4
1.2
1
LDP
0.8
DPLBAR
0.6
0.4
0.2
0
1
10
100
1000
The catchments presented in Table 1 were tested within the time to outlet model on a
25m grid derived from NextMap using a representative 6hr RMed event which was
locally weighted by SAAR. A broad range of permutations of universal land and
channel roughnesses were run through the model with the objective of identifying a
combination of universal roughness values which maximised the Rsq value while
maintaining a modelled lag time divided by observed lag of unity. All model runs were
run with a universal sinuosity factor of 1.1. It should be noted that the mean of the
time to outlet rather than median was used to estimate the lag time for each
catchment, this was on the basis of computational efficiency within ArcGIS and it is
recognised that future analysis should be based on the median to discount outliers. A
100% runoff was assumed universally, this has since been superseded by a
recommendation of basing runoff on HOST. The threshold of channel formation was
set at 0.1cumecs based on a visual comparison of the RMed flows and the OS 25K
river network, this assumption will need to be revisited if the percentage runoff is
changed.
The outcome of this stage of the flow routing demonstration process was that for a
25m grid using a sinuosity of 1.1 the optimum universal channel and land
roughnesses are 0.045 and 0.1. The average channel roughness (0.045) correlates
well with the channel roughnesses contained within the literature and the land
roughness (0.1) is comparable to the shallow surface flow roughness values
presented by USDA. The results of this analysis are presented in Figure 10 and
Figure 11.
Average modelled lag divided by observed for a range of land and channel
roughnesses using a 25m grid and a sinuosity factor of 1.1
0.2
1.4-1.5
0.1
1.3-1.4
1.2-1.3
1.1-1.2
1-1.1
0.9-1
0.8-0.9
0.7-0.8
0.6-0.7
0.5-0.6
0.02
0.03
0.05
0.05
0.04
Figure 10: Contour plot of the mean modelled catchment lag divided by observed for
the nine test catchments for a range of channel and land roughnesses
10
8
y = 0.9787x
R2 = 0.7093
6
0
0
10
12
Figure 11: Plot comparing the modelled and observed catchment lag with the optimal
universal land and channel roughnesses using a 25m grid and a sinuosity of 1.1
The next stage of the analysis was to test the impact of modifying individual land
classes on the models ability to predict lag times. In this analysis the channel
roughness and sinuosity were kept constant along with the land roughness for all
land use classes except the class which was being investigated in that run. After
investigating the effect of changing the roughness of a land class it was returned to
the default value (0.1) prior to testing the influence of another class. The results of
the analysis are presented in Figure 12 and Figure 13.
Methodical analysis of the effect of changing the roughness for
each LCM2000 class(n,channel = 0.045, S=1.1, n,land = 0.1 with
only 1 class being varied)
11
0.78
0.77
0.76
0.75
0.74
0.73
0.72
0.71
0.7
0.69
0.68
0
0.05
0.1
0.15
0.2
0.25
21
41
42
43
51
52
61
71
81
101
102
121
131
161
171 & 172
201
211
Figure 12: Plot showing the effect of varying individual land classes on the models
ability to predict the lag for the selected nine test catchments
Best fit catchment lag
12
10
y = 1.0309x
2
R = 0.884
8
S1.1,nL0.1,nC0.045
S1.1,nC0.045,Lv1
S1.15,nC0.045,Lv1
Linear (S1.15,nC0.045,Lv1)
0
0
10
12
Figure 13: Plot showing the modelled against observed catchment lag for the nine test
catchments using the best performing calibration coefficients
The provision roughness values for each land class are presented in Table 3.
LCM class
Location of
maxima
Adopted land
roughness
(provisional)
11
-0.627
0.1
21
0.323
0.3
41
0.147
0.15
42
0.157
0.15
43
0.256
0.1
51
-0.427
0.05
61
0.218
0.15
71
(minima, diverging)
0.05
81
0.221
0.15
101
(minima, diverging)
0.1
102
(minima, diverging)
0.05
121
(minima, diverging)
0.05
131
(minima, diverging)
0.05
161
(minima, diverging)
0.05
171 & 172
(minima, diverging)
0.025
201
(minima, diverging)
0.1
211
(minima, diverging)
0.1
Table 3: The provisional land roughness for each land class based on the analysis of
nine Scottish catchments
It should be noted that not all LCM classes were available within the nine test
catchments. In some cases some LCM classes only appear in one test catchment,
hence the optimisation process is biased. Indeed the small number of catchments
compared to the large number of variables does suggest that there is considerable
risk that the calibration is just bending the model to match the tested cases and that
the selected coefficients would perform poorly for other catchments.
Recommendations:
The analysis should be repeated using locally derived RMed depths for
storm durations which are representative of the catchment. At the same
time the percentage runoff should be based on HOST data not a 100%
runoff.
The impact of varying land roughness with HOST class should also be
investigated to determine whether HOST is a better predictor of land
roughness. It may be the case that both HOST and LCM class should be
used to define land roughness
Reed, Duncan, Faulkner, Duncan, Robson, Alice, Houghton-Carr, Helen, Bayliss, Adrian,
Flood estimation handbook: Vol. 4: Restatement and application of the flood studies report
rainfall-runoff method, Institute of Hydrology, 1999.