You are on page 1of 34

Portland and

Seattle Cycle
Analysis
Adam Bejan Parast
AG 1323 - Advanced GIS
Table of Contents
1 Executive Summary................................................................................................................ 6

2 Introduction ........................................................................................................................... 8

3 Background ............................................................................................................................ 8

3.1 Cycle Zone Analysis ......................................................................................................... 8

3.2 Project Goals.................................................................................................................... 9

3.3 Factors Contributing to Bikeability...................................................................................10

3.3.1 Street Connectivity ....................................................................................................10

3.3.2 Bicycle Faculties .......................................................................................................10

3.3.3 Land Use ..................................................................................................................11

3.3.4 Slope ........................................................................................................................11

3.3.5 Barriers .....................................................................................................................11

4 Data ......................................................................................................................................12

4.1 Address Points ................................................................................................................12

4.2 Digital Elevation Model ..................................................................................................12

4.3 Land Use ........................................................................................................................13

4.4 Road Network.................................................................................................................13

4.5 Bicycle Facilities..............................................................................................................13

5 Methodology .........................................................................................................................13

5.1 Factors ............................................................................................................................14

5.1.1 Street Connectivity ....................................................................................................14

5.1.2 Bicycle Facilities .......................................................................................................14

5.1.3 Land Use ..................................................................................................................15

5.1.4 Slope ........................................................................................................................17

5.1.5 Barriers .....................................................................................................................18

5.2 Constraints......................................................................................................................18
5.3 Factor Weights................................................................................................................18

6 Results ..................................................................................................................................20

6.1 Street Connectivity ..........................................................................................................21

6.2 Bicycle Facilities..............................................................................................................22

6.3 Land Use ........................................................................................................................23

6.4 Slope...............................................................................................................................25

6.5 Barriers ...........................................................................................................................27

6.6 Current and Potential Bikeability .....................................................................................29

7 Analysis and Discussion ........................................................................................................30

7.1 Comparison of Portland and Seattle .................................................................................30

7.1.1 Current .....................................................................................................................30

7.1.2 Potential ...................................................................................................................31

7.2 Discussion.......................................................................................................................32

7.2.1 Do the results agree with intuition? ............................................................................32

7.2.2 Density vs. Nearest distance for line feature conversion ..............................................32

7.2.3 Appropriateness of input data ....................................................................................33

7.2.4 Appropriateness of reclassed values. ..........................................................................33

7.2.5 AHP not used ...........................................................................................................33

7.2.6 Barriers in initial analysis may have meant something else..........................................32

7.2.7 Is the detail level too high to be useful ........................................................................34

7.2.8 How important are hills really ...................................................................................34

7.2.9 Hills create low scores even if most bicycle travel is along contour lines of the hill. ......34
Table of Figures
Figure 1 - Current Bikeability ........................................................................................................ 6

Figure 2 - Potential Bikeability ...................................................................................................... 7

Figure 3 - Portland Cycle Zone Analyis Zones............................................................................... 9

Figure 4 - Comparison of Connectivity Score................................................................................20

Figure 5 - Connectivity Scores......................................................................................................21

Figure 6 - Comparison of Facility Score........................................................................................22

Figure 7 - Facility Score ...............................................................................................................23

Figure 8 - Comparison of Land Use Score ....................................................................................24

Figure 9 - Land Use Score............................................................................................................25

Figure 10 - Comparison of Slope Score.........................................................................................26

Figure 11 - Slope Score ................................................................................................................27

Figure 12 - Comparison of Barrier Score.......................................................................................28

Figure 13 - Barrier Score ..............................................................................................................29

Figure 14 - Comparison of Current Bikeability, Figure 15 - Comparison of Potential Bikeability ....30

Figure 16 - Current Bikeability .....................................................................................................31

Figure 17 - Comparison of Potential Bikeabilty .............................................................................31


Table of Tables
Table 1 - Facility Scores ...............................................................................................................15

Table 2 - Portland Zoning Code and Assigned Scores ...................................................................16

Table 3 - Seattle Land Use Code and Assigned Scores...................................................................16

Table 4 - Barrier Type and Assigned Score....................................................................................18

Table 5 - Factor Weights for Current and Potential Maps..............................................................18


1 Executive Summary
This project extends previous bike analysis done for Portland, to Seattle by first recreating the
previous analysis of Portland and then by reproducing the analysis for Seattle. This analysis uses
Geographic Information Systems (GIS) to perform a Multi-Criteria Evaluation (MCE) at a very high
resolution. Essentially, mathematical operations are applied to images in which each pixel represents
a numerical value. MCE is an established tool that can be used to make spatially related decisions
when competing factors of varying importance are present.

This analysis uses five factors; street connectivity, bicycle facilities, land use, slope and barriers to
analyze how bikeable Portland and Seattle currently are, as well as what their potential
bikeability is. These maps are the result of many intuition-based subjective scores and weights,
and should not be taken as fact, but rather rough approximations. In addition data quality and time
limitations were major constrains that should be kept in mind.

Seattle
Cycle
Analysis
Current
Bikeability
Portland Cycle Analysis
Current Bikeability

Legend
< 45
45 - 50
50 - 55
55 - 60
60 - 65
65 - 70
70 - 75
Legend
75 - 80
< 45 80 - 85
45 - 50 85 - 90
50 - 55 90 >
55 - 60
60 - 65


65 - 70
70 - 75
75 - 80
80 - 85
> 90 0 0,5 1 Miles

0

0,5 1 Miles

Figure 1 - Current Bikeability

As Figure 1 shows this analysis indicates that Portland is currently far more bikeable than Seattle,
especially in the city center. Portland has a very bikeable city center, with surrounding
neighborhoods also displaying good bikeabilty. In these areas all five of the factors contribute to a
bikeable environment. Seattle has a less than bikeable city center, while surround neighborhoods are
somewhat better. For the rest of the city, Seattle is found lacking compared to Portland. The large
difference between Portland and Seattle is a manifestation of both inherent factors as well as policy
dependant factors.

Figure 2 below shows another map that attempts to model potential bikeability of Portland and
Seattle. Downtown Portland and inner Portland have the potential to become a very bikeable area,
while north and eastern Portland can be fairly bikable as well. The city has a fairly uniform potential
with large continuous bikable areas. Seattle also has the potential to become very bikeable in some
parts of the city, and fairly bikeable within many of the urban villages. However, Seattle has unique
challenges mostly due to topography. In some ways Seattle is characterized by islands of
bikeability. Most urban villages are located somewhere within these islands, as they are typically
located at the top of a hill or bottom of a valley. These differences bring up many interesting
questions and implications.

For example while Portland might wisely choose to blanket the city with bicycle facilities in a grid
like fashion, it might be wiser for Seattle to focus on fewer but higher quality facilities on routes that
connect the most bikeable islands, with a second objective of providing facilities that improve
internal circulation of individual islands. For areas where islands are separated by large areas of low
bikeability high capacity transit that is integrated with the bicycle network could be used as a way to
bridging the gap between these islands.

Seattle
Cycle
Analysis
Potential
Bikeability
Portland Cycle Analysis
Potential Bikeability

Legend
< 45
45 - 50
50 - 55
55 - 60
60 - 65
65 - 70
Legend
70 - 75
< 45
75 - 80
45 - 50
80 - 85
50 - 55 85 - 90
55 - 60 90 >
60 - 65
65 - 70


70 - 75
75 - 80
80 - 85
85 - 90
90 > 0 0,5 1 Miles

0

0,5 1 Miles

Figure 2 - Potential Bikeability

Again this analysis should not be taken as fact, but should rather be viewed as the starting point of a
discussion about how best to model bikeability on a detailed level as well as the strengths and
weaknesses of each city in relation to biking. The report that follows goes into more detail about the
background, methodology and results of this analysis. The document concludes with a discussion of
issues and questions raised while working on the project. Three appendixes contain full size images
of the maps included throughout the report.
2 Introduction
This project builds off of a previous bikeability analysis done for the City of Portland. This project
attempts to first recreate the previous bikeability analyze of Portland and then use the same
methodology to analyze Seattle. With this, a comparison of how bikeable Portland and Seattle are
relative to themselves and to each other is possible. This analysis was done as a final project for an
advanced geographic information system (GIS) class by a single student and thus has significant
limitations, such as but not limited to input data quality, available time, and appropriateness and
rigor of assigned measures, factors or weights. These results should not be taken as fact and are
simply a good faith effort to use this type of GIS analysis to address an interesting real world
problem. A healthy discussion at the end of the paper touches on some of these limitations and
difficulties, as well as questions raised.

What makes a city bikeable is generally well understood, at least in a qualitative manner. A very
bikeable city is where biking is enjoyable and easy, and as a results a large and diverse cross section
of society uses bikes as a means of every day transportation. A less bikeable city will still have those
who choose to bike, mostly consisting of younger and predominantly male bicyclist. In this way the
design user of this analysis is more similar to those used in Europe than those typically used in the
United States.

The factors that contribute to a bikeable city are dense urban environments with a large number of
people within biking distance of destination they want to go to, a connected and high quality non-
motorized transportation network, comfortable and safe bicycle facilities separated in some way
from vehicles when vehicles are present in high-volumes or at high-speeds, and more or less level
terrain. The relative importance of each are hard to quality and is currently being studied.

Many of the most bikeable cities, especially those in Europe, display these features in abundance and
in many ways are good with respect to all of these factors. In this way these cities have an advantage
because all factors are more or less working to their advantage. Many cities in United States are not
the same. While they might have one of two features that contribute to a more bikeable city, they
most likely dont have all of them. So while some factors contribute, others detract. The conflict
between of these factors makes analysis difficult, requiring a more formalized and structured analysis
technique.

This paper will outline the factors used to measure bikeability and why they were used. It will then
discuss previous work done in the area of GIS based bike analysis. Specific data inputs and the
general analysis methodology will then be discussed. Results will be presented followed by an
analysis and discussion. Full size maps are included in the Appendix A, B and C.

3 Background

3.1 Cycle Zone Analysis


As stated before this project builds off of previous work done by the City of Portland and Alta
Planning with the Cycle Zone Analysis. That project analyzed the quality of bicycling within 32
different zones of the city. These zones were created from areas where bicycling felt similar and vary
in size and shape as can be seen in Figure 3.

Figure 3 - Portland Cycle Zone Analyis Zones

The above mentioned bikeability factors were rated for each zone, and then combined into a
current and potential score for each zone. The current score takes all factors into account and
uses weighting factors described later in this paper. The current map attempts to mirror the on-the-
ground reality of how biking in different zones of the city currently feels. The potential score is
meant to show the potential of bicycling in the different zones, ignoring bicycle facilities and barriers.
While land use patterns, street network, and slope are more or less permanent, bicycle facilities and
barriers are not. In this way the potential analysis looks at how the underlying and permanent
factors contribute or dont contribute to a bikeable zone.

3.2 Project Goals


The primary project goal is to apply knowledge, methods, and tools learned the course. This
includes using the same basic methodology of the cycle zone analysis but increases the granularity of
the analysis. Rather than using manually defined zone in which to conduct analysis, this project uses
raster surfaces with a pixel size of 25 ft by 25ft. This means that for most practical purposes a
continuous surface is analyzed rather that discrete zones. A new technique is used for measuring the
connectivity and quality of the street network based. The new techniques uses the route directness
index measure.
The secondary goal of this project is to then compare the new analysis for Portland with an analysis
of Seattle, using the same methodology, scoring and weighting. Inter-city or intra-city comparison of
the quality of the bicycle environment is currently extremely problematic. By creating a transferable
analysis tool a more rigorous, yet still somewhat subjective comparison, becomes possible between
the two cities. This analysis could be extended to many cities creating a consistent way of measuring
the bikeability of cities across the US or around the world.

Again please note that while the final results are a good step in the right direction they should not be
taken as fact. The analysis shows that this methodology has promise but it also shows that high
quality data that represents the truth on the ground is obviously necessary for good results.
Additionally, future analysis should re-evaluate the best way to convert localized vector features into
raster features as well as what are the most appropriate factor weights and ranges. This should be
kept in mind when results are viewed.

Below is an explanation of the different factors used and why it was used in the evaluation. Some
background might also be provided. The exact data files and how they were processed is discussed in
the Section 4 and 5.

3.3 Factors Contributing to Bikeability


3.3.1 Street Connectivity
Both Seattle and Portland are older west coast cities, with gridiron street networks. Portland has an
extremely dense network downtown while Seattles is larger and more rectangular in most areas.
Both cities have to some degree dendritic type street network in newer parts of the city or in areas
with step grades.

Street networks that exhibit good connectivity and thus shorter and more direct travel are good for
non-motorized transportation. When the connectivity of a street network decrease the directness of
travel also decreases, ultimately leading to increased travel distances. This is acutely noticeable in
suburban developments where a trip to another house just across the fence could result in a walk so
long some would choose to drive. Less direct travel decreases the number of locations that are
quickly reachable by non-motorized modes of transportation. While this isnt extremely important
for motorized transportation, since speeds are high and effort it low regardless of speed, this is very
important for non-motorized transportation since speeds are so low and increased speed increases
effort.

Existing best practices typically use 4-way intersections and road network density as a measure of
connectivity, however these are only corollaries, not direct measures. This analysis uses a method
called route directness index instead and is a direct measure of the network connectivity.

3.3.2 Bicycle Faculties


Bicycle facilities are a critical part of the bicycling environment. Studies have shown that bicyclist
travel out of their way to use them. Additionally, bicycle facilities attract bicyclist that might
otherwise not ride since bike lanes and other facilities are viewed as more comfort or safer than
riding with vehicles.
Portland has spent significant resources over the last few decades on bicycle facilities. This first stage
of investment has focused on the center city mostly with bicycle lanes, bicycle boulevards, and multi-
use paths. Portland recently embarked on a second stage of investment to increase the facility
density in the center city and expand the coverage of facilities over more of the city.

Seattle in contrast has only recently started investing in bicycle infrastructure, except a few popular
multi-use paths. The city is making its first stage of investments, which predominantly focuses on the
most needy and obvious projects, mostly using bike lanes and sharrows, as well as connecting gaps
in multi-use paths.

3.3.3 Land Use


Portland has been a national leader in land use planning, with the passage of the Urban Growth
Boundary in 1973. This has lead to relatively dense and focused growth in the Portland region.
While Seattle has only had a similar state law, the Growth Management Act since 1991, it has a
denser and large downtown core than Portland. Portlands dense neighborhoods are mostly focused
in the center city, while Seattles density is more spread out and multi-nodal.

Density was used as a corollary for travel demand that can reasonably be meet using a bicycle.
Higher density areas have a large number of trips. Additionally areas that are dense have a larger
number of destinations that are within an easy bike ride. These areas also often have limited or
expensive parking, further encouraging other modes of transportation.

3.3.4 Slope
Any bicyclist knows that climbing hills is not enjoyable. This is common sense however the impacts
that hills have on bicycling are hard model. Attempts to qualify the importance of this factor will be
somewhat subjective. Regardless, it is important that this input are used because it obviously has an
important role on the bikeablity of a city. Portland is generally a flat city while Seattle is hillier with
most hill ridges and valleys running north/south across the city.

3.3.5 Barriers
In this analysis barriers refers to characteristics of streets that are not accounted elsewhere in the
analysis, i.e. high-traffic, high-speed streets. These roads create barriers for bicyclist to cross and are
also undesirable for bicyclist to ride along unless there is some type of separate bicycle facility along
them. This factor is an attempt to take into account how vehicles reduce the comfort and safety of
biking since no other factor does. Other parts of this analysis take into account the quality of the
network or bicycle facilities, but the quality of the street with relation to vehicle speed or volumes is
not taken into account elsewhere.

There are several ways barriers can be factored into the analysis. The simplest classification possible,
and the one used in this analysis, is based on the functional class of a road, i.e. primary or secondary
arterials as an indication of poor comfort. More complex and probably more appropriate
classifications can use speed limit, traffic volumes, 80th percentile speed, accident record, etc.

Initially this input was not included in the analysis because it was believed that the street
connectivity score would take barriers into account. However after reviewing initial results it
became evident that heavily trafficked areas with high traffic volumes or high speed limits, and thus
uncomfortable areas for bicyclist, were not being accurately represented. Some feedback from Kim
Voros at Alta Planning who is working on further development of the cycle zone analysis reinforced
this conclusion. However, inclusion of this factor has probably over penalized Downtown Seattle as
discussed elsewhere in this paper. A possible solution is to still use this factor, but change the
weighting factors used in the MCE.

4 Data
The data used in this analysis is fairly common if not universally available for most large and
medium sized cities. This is a major benefit because it allows for the methods used in the analysis to
be applied to a variety of cities with different levels of GIS sophistication and data precision. Data
for the Portland analysis was obtained through the City of Portlands Bureau of Technology Services.
Data for the city of Seattle was obtained through the Washington State Geospatial Database Archive
(WAGDA), which is maintained by the University of Washington Libraries. Data for Seattle was
obtained from WAGDA unless otherwise stated.

The normal cautions and warning associated with using GIS data that is not your own must be kept
in mind. In general most data was only briefly reviewed for accuracy, ensuring that there werent
any systematic problems with the data. Several known problems with the street network files were
corrected although other known issues were not corrected because of the large effort it would have
taken to correct. Also note that a large amount of data manipulation and reclassing was conducted.

Another point that should be made is that this analysis is an effort to apply knowledge, methods and
tools learned during the course to a real world problem. Therefore data quality wasnt the
paramount concern of this project. If future analysis uses this methodology more stringent quality
control of the data is important.

4.1 Address Points


Address points were used in both analyses. Randomly selected points were used in the analysis as
the complete database was too large to practically use.

Portland (portland_address.shp) Confirmed address points within the city of Portland. This
is a shape file with point data. Points are located at the centroid of tax lot. There can be
multiple address points per location since different addresses associated with the same parcel
will have the same location.
Seattle (dap.shp) Discrete address point locations.

4.2 Digital Elevation Model


Contour files for both cities were imported and converted to create a TIN digital elevation model.
Both data files had increments of 5ft, which created processing and storage problems so only
increments of 10ft were used for TIN creation.

Portland (portlandcontours5ft.shp) Contour shape file within the City of Portland.


Seattle (contour.shp) Contour shape file for the City of Seattle
4.3 Land Use
Polygon shape files of zoning throughout both cities. Zoning was aggregated over continuous areas
with the same zoning.

Portland (zoning_PDX.shp) Shapefile of all zoning within the city, including un-buildable
areas.
Seattle (zoning.shp) Shapefile of all zoning within the city including un-buildable areas.

4.4 Road Network


Line shape file with street functional class, name and other attributes. This file was used in the
network analysis. Some freeways were includes in the data files, others were not. Data problems like
this were not rectified

Portland (portlandstreets.shp)
Seattle (snd.shp) Several state highways owned by the city are included.

4.5 Bicycle Facilities


Line shape file including planned and existing bicycle facilities by type as well as other details were
included.

Portland (portlandbicyclenetwork.shp) Facility types are multi-use paths, bike boulevards,


bike lanes, and signed bike routes.
Seattle (Bike_facility.shp) Facility types are multi-use paths, bike boulevards, bike lanes in
both directions, bike lane on one-way street, climb bike lane in uphill direction and sharrow
in the downhill direction, sharrows in both directions, sharrows on one-way street, and
signed bike routes. Data file was obtained directly from the City of Seattles Department of
Transportation because WAGDA data was either out of date or the result of some
processing that did not clearly differentiate between types of bicycle facility.

5 Methodology
The section below explains the methodology of the analysis and in some cases compares this
methodology to that used in the original cycle zone analysis. The factors and constrains used in the
MCE are discussed. An MCE was used because there are multiple competing factors that impact the
quality of the bicycling environment. No single factor has an overriding impact, yet none of them
can be overlooked.

The raster math for the MCE was done within ArcMap. Factor maps were first created from the
input data. These maps were then re-classed on a scale from 0 to 100. Zero being the worst possible
score and 100 being the best possible score. These re-classes were always linear in nature. Finally
these maps were multiplied by weighting factors and added together to create composite factors
maps. This process can also be called a weighted sum raster analysis. The constraint maps were then
combined with the weighted sum factor map to create the final MCE maps. An MCE was done to
analyze both the current bikeability as well as potential bikeability.
Although Analytical Hierarchy Process (AHP) was not used to determine factor weights it was
included in the discussion of the factor weights.

5.1 Factors
Five factors were used in the MCE. They are bikeway quality, comfort of street, quality of street
network, slope, and land use.

Below is a brief description of how each factor map was created.

5.1.1 Street Connectivity


This measure diverges most from the previous cycle zone analysis. The previous analysis used 4-way
intersection density and road network density as a corollary for a connected and direct street
network. This analysis uses route directness index instead. This measure is based on the ratio
between network distance and the straight-line distance. A trip that is less direct because of a poorly
connected street network will have a lower route directness value than a trip of equal straight-line
distance which can be made in a more direct manner because of a more interconnected street
network.

The maximum value possible is 1 and a practical minimum of roughly 0.3. A value of .75 is typical
of gridiron networks. It is important to note that this measure is less sensitive with relation to grid
density than other measures. It also important to note that the surfaced created for this factor was
created using randomly selected sample points, so values between points were created using
interpolation.

Route directness values between 0.50 and 0.90 were re-classed as 0 and 100 respectively, with values
between converted linearly.

5.1.2 Bicycle Facilities


The type of facility was used as an indicator of the quality of the facility. This is a simplification of
reality but more rigorous measures would be too time consuming to create, or require data not
available. Multi-use paths had the highest quality followed by bike boulevards, bike lanes, sharrows
and signed bike routes. Consistency of score between Portland and Seattle facility types was
attempted.

The vector data was converted into raster data using the line density method. This conversion from
vector to raster data was necessary for the MCE. The densities method was used for multiple reasons.
A discussion of why distance measures were not used is located within the discussion section.
Table 1 - Facility Scores

Facility Score
Multi-use Path 10
Bike Boulevard 9
Bike Lane (two-way) 6
Climb Lane (up
lane, downhill sharrow) 6
Bike Lane (one-way) 3
Sharrow (two-way) 3
Signed bike route 3
Sharrow (one-way) 1.5

Table 1 above shows the values assigned to each facility type. These values are subjective estimates
of the quality of that type of facility. Only roads with facility get points, thus low volume streets that
can informally be used like bicycle boulevards are not counted. Additionally, for example there are
no distinctions between high quality bike lanes and low quality bike lanes.

Bike lanes or sharrows on a one-way street are assigned half the score of a two-way street. This
means that two one-way streets will have the same score as a two-way street with bike lanes in both
directions. The highest value was determined by the highest score of Portlands facility factor map.
These values were re-classed to cover a range of 0 to 100 using a linear conversion.

5.1.3 Land Use


Land use data was re-classed using two criteria, destination and originating quantity of trips.
Commercial, industrial, and retail land uses are destination land uses while residential land uses are
originating land uses. Scores were given on a 0 to 10 scale for each type of trip and then added
together for a total score. One exception was made. Downtown Portland is less dense than
downtown Seattle. Since values were first created for Portland and a value of 10 was given to
downtown Portland a value of 13 was assigned to Seattles Downtown Core 1.

A maximum total score of 13 resulted for central business distracts as well as multiuse midrise zones.
Detailed scores for Portland can be seen in Table 2, detailed scores for Seattle are in Table 3.
Although the zoning codes for Portland and Seattle are not the same an attempt was made to ensure
that zone that are roughly equivalent have equal scores. Although central business districts certainly
have the largest concentrations of activity, work related trips make up a small percent of total trips.
Trips for other purposes such a shopping, recreation and leisure were also emphasized rather than
just focusing on commuting trips.

More robust scores could be based off travel demand generation handbooks or metropolitan
planning organization travel surveys. Housing density and employment density could also be used.
Table 2 - Portland Zoning Code and Assigned Scores

Originating Destination
Zone Name Demand Demand Total
CG General Commercial 5 5
Mixed
CM commercial/residential 7 2 9
CN1 Neighborhood Commercial 1 4 4
CN2 Neighborhood Commercial 2 3 3
CO1 Office Commercial 1 3 3
CO2 Office Commercial 2 4 4
CS Storefront Commercial 4 4
CX Central Commercial 10 10
EG1 General Employment 1 5 5
EG2 General Employment 2 5 5
IG1 General Industrail 1 3 3
IG2 General Industrail 2 3 3
IH Heavy Industrial 3 3
IR Institutional Residential 3 6 9
OS Open Space 2 2
R1 Residential 1,000 6 6
R2 Residential 2,000 3 3
R2.5 Residential 2,500 3 3
R3 Residential 3,000 2 2
R4 Residential 4,000 2 2
R5 Residential 5,000 2 2
R7 Residential 7,000 1 1
R10 Residential 10,000 1 1
R20 Residential 20,000 1 1
RF Residential Farm 0 0
RH High Density Residential 10 3 13
RX Central Residential 10 3 13
EX 5 6 11

Table 3 - Seattle Land Use Code and Assigned Scores

Originating Destination
Zone Name Demand Demand Total
C1 Commercial 1 5 5
C2 Commercial 2 5 5
DH1 Downtown Harborfront 1 5 5
DH2 Downtown Harborfront 2 5 1 6
DMC Downtown Mixed Commercial 9 9
DMR Downtown Mixed Residential 10 3 13
DOC1 Downtown Office Core 1 13 13
DOC2 Downtown Office Core 2 10 10
DRC Downtown Retail Core 10 10
HR Highrise 10 3 13
IB Industrial Buffer 3 3
IC Industrial Commercial 7 7
IDM International District Mixed 5 5
IDR International District Residential 7 7
IG1 General Industrial 1 3 3
IG2 General Industrial 2 3 3
L1 Lowrise 1 3 3
L1/RC Lowrise 1 Residential/Commercial 3 1 4
L2 Lowrise 2 4 4
L2/RC Lowrise 2 Residential/Commercial 4 1 5
L3 Lowrise 3 5 5
L3/RC Lowrise 3 Residential/Commercial 5 1 6
L4 Lowrise 4 6 6
L4/RC Lowrise 4 Residential/Commercial 6 1 7
LDT Lowrise/Duplex/Triplex 4 4
MIO Major Institution Overlay 3 6 9
MR Midrise 7 7
MR/RC Midrise 1 9 9
NC1 Neighborhood Commercial 1 1 4 5
NC2 Neighborhood Commercial 2 4 4 8
NC3 Neighborhood Commercial 3 7 3 10
PMM Pike Market Mixed 6 6
PSM Pioneer Square Mixed 2 6 8
RSL Residential Small Lot 3 3
SF 5000 Residential Single Family 5000 2 2
SF 7200 Residential Single Family 7200 1 1
SF 9600 Residential Single Family 9600 1 1
SM Seattle Mixed 6 6 12
SMR Seattle Mixed Residential 8 2 10

5.1.4 Slope
Ten-foot interval contour lines were converted into a TIN surface. This was converted into a raster
elevation surface. That raster surface was then processed into a percent slope surface. Finally this
surface was then re-classed with 15% slope as the lowest value, zero, and 0% slope as the highest
value, 100. In this way all slopes greater the 15% are assumed as completely unattractive from a
slope perspective. Areas that are flat are most attractive. The same re-class scale was used for both
Seattle and Portland. 15% slope was used as the maximum cutoff because this probably the
maximum slope that a casual bicyclist would ride up.
5.1.5 Barriers
The functional class of a road was used as an indicator of the barrier impact a road has. Vehicle
volumes or speed limit would probably be a more representative measures. Primary arterials were
deemed least appropriate and secondary arterials half as appropriate. These vector line features were
then converted into raster maps using a feature density method.

This measure indicates areas with a high concentration of heavily trafficked streets, and thus less
comfortable for bicyclist. A single arterial does not have significant impact but areas close to freeway
interchanges or areas in the central business district with a large number and dense arterial network
will be impacted.

Table 4 - Barrier Type and Assigned Score

Functional Class Score


Primary Arterial 0
Secondary Arterial 5

Table 4 above shows the values assigned to different types of streets. Primary arterials are viewed as
largest barrier with secondary arterials half as onerous as primary arterials. These values are
subjective estimates and have no rigorous foundation. Seattles barrier factor map had the worst
barrier scores and thus the ranges of re-classed values were determined by it. The same range was
applied to Portland. Note that other streets were not included.

5.2 Constraints
Only two constrains were used in the MCE, all other inputs were factors. A constraint map was
create to cancel out areas were water is located. A second constraint map created from zoning data
canceled out areas that were located outside of the city boundaries, or not controlled by the city.
Values of one were assigned to areas that should be included in the final map and values of zero
were assigned to areas that should not be included in the final map.

5.3 Factor Weights


One of the major objectives of this analysis was to compare it to the original analysis. For this reason
the same factor weights were used. Additionally, the extra time needed to verify new weights was
restrictive. How the values used in the original analysis were created is unknown. Table 5 below
shows the weighted values used for the current and potential bikeability maps.

Table 5 - Factor Weights for Current and Potential Maps

Current Potential
Factor Map Map
Land Use 12% 25%
Barriers 25% -
Street
Connectivity 26% 50%
Bicycle Facilities 31% -
Slope 6% 25%

Analytical Hierarchy Process (AHP) is a decision making tool that helps determine weight of
competing factors. Multiple one-to-one relative importance scores are used to determine the overall
relative weights to be used in the MCE. This method helps to ensure that weights are created in a
reproducible manner. Additionally, AHP calculates the consistency of input ratings, showing
whether or not the relative importance input values is consistent. Although using AHP is best
practice for MCEs it was not used because it was desired to use the same weights used in the
previous analysis. Additionally, using independently generated weighting factors would add one
more variable that hasnt been either previously used or rigorously verified.
6 Results
This section contains factor maps as well as current and potential bikeability maps. The size of the
maps have been reduced so they can be easily displayed inline with text. Full size maps are
contained in Appendix A, full size comparisons of the original cycle zone analysis and results from
this analysis for Portland are in Appendix B and full size comparisons of Portland and Seattle are
contained in Appendix C. Maps comparing the original cycle zone analysis to this analysis are first
shown, followed by maps that compare Portland and Seattle.

Portland Cycle Analysis


Connectivity Score


Legend
High : 100

Low : 0 0 0,5 1 Miles

Figure 4 - Comparison of Connectivity Score

All Portland and Seattle maps created in this analysis can be directly compared to their complement
map since color-coding is standardized between the two cities. All factor maps use the same color
scheme with the lowest scores represented by red and the highest score represented by blue. This
color scheme is similar to the color scheme used in the cycle zone analysis.

For example the current map for Portland and Seattle are directly comparable and the potential map
for Portland and Seattle are also directly comparable. However, note that the while the current and
potential maps for each city are related they are not directly comparable to each other because they
use different MCE weights as described in Section 5.3. The most notable example of this is the
weight for slope. While the current maps uses a value of 6%, the potential map uses a value of 25%,
thus the potential maps values slope almost 5 times as much as the current map.

Also note that the while the same scoring range, 0 to 100, was used for both analyses, the values that
a color represents in the original analysis may not be the same. The original analysis stretched the
color range rather than using a fixed, liner color range, which was used for this analysis. For
example a score of 50 in one of the original maps could be blue while the same score could be yellow
in another original map. All factor maps created for this analysis use the same color range, while the
two result maps, current and potential, use a different color range. Thus it is important to observe
what value a color represents.

6.1 Street Connectivity


The connectivity measure used in this analysis was different the one used in the original cycle zone
analysis. The connectivity map created for this analysis is the equivalent of two factors used in the
original cycle zone analysis. A side-by-side comparison of these maps to the map created for this
analysis are in Figure 4.The variation in score is more dramatic for the original cycle zone analysis
but still shows the expected trend. Inner Portland as well as the older neighborhoods with gridiron
street networks have high scores. The map created for this analysis has less variation but certainly
displays the same trends. The extremely high scores on the south and northeast of the maps are
irregularities caused by bad data.

Seattle
Cycle
Analysis
Connectivity
Score
Portland Cycle Analysis
Connectivity Score

Legend
High : 100

Low : 0


Legend
High : 100

Low : 0 0 0,5 1 Miles

0

0,5 1 Miles

Figure 5 - Connectivity Scores

Figure 5, shows the connectivity scores for Portland and Seattle. Both cities generally have good
connectivity. While both cities have consistently high scores in the center city or pre-WWII
neighborhoods Portland has much more variation than Seattle. The lowest score for both Portland
and Seattle occur in areas with steep slopes where there is less intensive development, most of which
is only single-family housing. It is interesting to note that Portland has three distinct areas. Portland
proper, the area located to the southwest of Downtown Portland, and the area east of I-205. The
lowest scores in Seattle are mostly located on in hilly, waterfront areas.

6.2 Bicycle Facilities


Figure 6 is a comparison between the facility scores for the original cycle zone analysis and this
analysis. Areas that are white are not in close proximity to any designated bicycle facility and thus
have a score of zero. Areas of dark red also have low scores, starting at zero. This makes it possible
to see areas that have low scores because there are no bicycle facilities located in close proximity, or
areas that do have low scores because facilities located close by are not high density or high quality.

Portland Cycle Analysis


Facility Score


Legend
High : 100

Low : 0 0 0,5 1 Miles

Figure 6 - Comparison of Facility Score

As discussed below the largest discrepancies between the two analyses are related to multi-use paths.
Also, the zone running from the South Waterfront down along the Willamette River has a very high
score in the original cycle zone analysis. This is not the case with this analysis. Overall looking
carefully zone by zone and comparing it to the map created for this analysis the relative rankings are
generally the same. In other words, relative to other parts of the city, scores in both analyses are
generally the same.
Figure 7, shows the bicycle facility scores for Portland and Seattle. Areas in central Portland both
east and west of the Willamette River have the highest scores. High scores are also found in areas
with multi-use paths such at Forest Park, near Portland International Airport, and I-205. This partly
represents the more winding nature of these paths, which increases the apparent density of facilities
without representing much of an increase in the quality of the bicycle network. Several larger gaps
exist in both the Northeast and Southeast part of the city.

Seattle
Cycle
Analysis
Facility
Score
Portland Cycle Analysis
Facility Score

Legend
High : 100

Low : 0


Legend
High : 100

Low : 0 0 0,5 1 Miles

0

0,5 1 Miles

Figure 7 - Facility Score

Seattle has much lowers scores. The highest scores are located in areas where multi-use paths are
close to or cross other types of facilities. Seward Park road, which loops around Seward Park, is
classified as a multi-use path and thus the density created by this facility is high. The high values in
Fremont and University District are the intersection of the Burke Gilman Trail and north/south bike
lanes as well as sharrows and signed bike routes located closet to bridges in the area. Likewise, the
high score in Interbay is the result of the Elliot Bay trail and the high score near I-90 and Rainier is
the result of the I-90 multi-use paths.

Seattles lack of bicycle infrastructure downtown is evident, with areas around the downtown core
scoring better. All neighborhoods except central Fremont score poorly. Also there are large areas
without any nearby facilities especially in the northeast and southeast part of the city.

6.3 Land Use


Figure 8 is a comparison of land use scores for both analyses. As described in Section 5.1.1 scores
were assigned to the different land use types.

Although it initially looks that the original cycle zone analysis has higher scores than this analysis,
this is a bit mislead. While all maps in this analysis used a linear color scheme, it appears that the
originally analysis stretched their distribution, to create maps that have a good number of zones
for each color. This creates a distorted map. For example all zones between red and blue have
scores lower than 50 points out of 100 while light purple and purple represent the remaining 50
points. This means that zones that appear blue on the original analysis would appear yellow or light
green in this analysis. Thus while the maps look very different because of the color-coding used, they
actually represent similar scores.

Portland Cycle Analysis


Landuse Score


Legend
High : 100

Low : 0 0 0,5 1 Miles

Figure 8 - Comparison of Land Use Score

There are however several areas of difference. The area north of Downtown has very high scores,
even higher than some portions of the downtown core in this analysis, while the original analysis has
higher scores in the core area. Several other areas with differences are along MAX lines. These areas
score high, but are fairly small and surrounded by lower scoring single-family zoning. For the
original analysis an average over the zone must be used, while this analysis is much more localized.
Thus it can be hard to compare land use scores in zone with a large diversity of zoning types.
Seattle
Cycle
Analysis
Landuse
Score
Portland Cycle Analysis
Landuse Score

Legend
High : 100

Low : 0


Legend
High : 100

Low : 0 0 0,5 1 Miles

0

0,5 1 Miles

Figure 9 - Land Use Score

Figure 9 shows land use scores. Both Portland and Seattle have high scores in the city center and
along major arterials or transit lines. The highest scores in Portland are downtown, the Llyod
District, along N Interstate Ave/I-5 and Gateway. Areas to the southwest of Downtown Portland as
well as parts of the Far Northeast and Far Southeast have the lowest scores. Seattle has three high
scoring areas, the Center City, University District and Northgate. Lower scores are in the north and
south of the city. Both Portland and Seattle are predominantly covered in low scoring single-family
housing.

6.4 Slope
Figure 10 compares the slope score maps of both analyses. The maps are extremely similar with a
few exceptions, Mt. Tabor City Park and Rocky Butte State Park. The reason for their high slope
scores in the original analysis is unknown.

These maps show the distinct topography of each city. Portland mostly has high scores, except in the
area to the southwest of Downtown. The rest of Portland is located in a relatively flat area in
between the Columbia River and Willamette River. Downtown Portland is located on a narrow
wedge of land between the hills and Willamette River.
Portland Cycle Analysis
Slope Score


Legend
High : 100

Low : 0 0 0,5 1 Miles

Figure 10 - Comparison of Slope Score

Seattle in comparison is much hillier with the exception of the Duwamish River valley. Seattle
neighborhoods are generally bounded by steep hillsides with most valleys and ridgelines running in
roughly a north/south orientation. Within individual neighborhoods slopes scores are generally
mediocre. The Ballard/Crown Hill neighborhoods are the largest contiguous area with relatively
good scores.
Seattle
Cycle
Analysis
Slope
Score
Portland Cycle Analysis
Slope Score

Legend
High : 100

Low : 0


Legend
High : 100

Low : 0 0 0,5 1 Miles

0

0,5 1 Miles

Figure 11 - Slope Score

6.5 Barriers
Figure 12 is a comparison of barrier score for both analyses. It is very evident that these maps are not
measuring the same thing. The original analysis map looks fairly similar to the connectivity and road
network density maps. What the map is measuring is not known and questions directed to Kim
Voros at Alta Planning were not responded to. Generally zones in the inner city had high scores,
while areas to the southwest of downtown had the lowest scores, and scores in the East decrease the
further east you go.

The barrier map for this analysis is the opposite of the original analysis, with the city core having the
lowest scores and areas outside of the inner city scoring highest. This is a very intuitive result of the
input data but it is not the same as what was measured in the original cycle zone analysis.
Portland Cycle Analysis
Barrier Score


Legend
High : 100

Low : 0 0 0,5 1 Miles

Figure 12 - Comparison of Barrier Score

Error! Reference source not found. shows that the barrier score trend of Portland and Seattle are
fairly similar, although Portland has higher scores in Downtown. Portlands lowest barrier score is
only medium low on the range. Seattle had a much higher density of streets classified as arterials in
the downtown core than Portland did. This density map does not include interstates, although SR-
99/SR-509 and the West Seattle Freeway are included in the Seattle analysis. Such a low score in
Downtown Seattle reflects that fact that almost all streets in the downtown core are classified as
primary or secondary arterials.
Seattle
Cycle
Analysis
Barrier
Score
Portland Cycle Analysis
Barrier Score

Legend
High : 100

Low : 0


Legend
High : 100

Low : 0 0 0,5 1 Miles

0

0,5 1 Miles

Figure 13 - Barrier Score

6.6 Current and Potential Bikeability


Comparison of the Portland and Seattle MCE for both the current bikeability and potential
bikeability are saved for the next sections.

Figure 14 compares the current bikeability of the original analysis and this analysis. A detailed
discussion of all the similarities and differences is difficult but generally the analyses have similar
results. For both analyses the downtown core scored lower than the northern part of downtown.
Additionally the Inner East between the Llyod District and OMSI has the highest scores outside of
downtown, with scores decreasing the farther east you go. Surrounding this area is a ring of lower
scoring areas. Areas to the east I-205 score poorly and areas to the southeast of downtown have the
worst scores.

Figure 15 compares the potential bikeability of the original analysis and this analysis. This map and
the factor maps show the same trends. Downtown Portland has the highest score while inner
Portland has good to medium scores, the area east of I-205 has low scores and the area southwest of
Downtown Portland has very low scores.

Although there are small differences between these maps its believe that overall these two analyses
present similar conclusions. While some questions remain to be answered, especially with respect to
the barrier factor map, its believe that for the intents of this project the analysis successfully
recreated the original cycle zone analysis.
Portland Cycle Analysis Portland Cycle Analysis
Current Bikeability Potential Bikeability

Legend
Legend < 45
45 - 50
< 45
50 - 55
45 - 50
55 - 60
50 - 55
60 - 65
55 - 60
65 - 70
60 - 65


70 - 75


65 - 70
75 - 80
70 - 75
80 - 85
75 - 80
85 - 90
80 - 85
90 > 0 0,5 1 Miles
> 90 0 0,5 1 Miles

Figure 14 - Comparison of Current Bikeability Figure 15 - Comparison of Potential Bikeability

7 Analysis and Discussion

7.1 Comparison of Portland and Seattle

7.1.1 Current
Figure 16 is a comparison of the current bikeability maps for Portland and Seattle. Downtown
Portland has the highest scores, with large portions of inner city neighborhoods also having medium-
high scores. In comparison Seattle has very low scores in the downtown core. Center city
neighborhoods ringing downtown have medium scores. The rest of the city generally has low scores.

The lower scores for Seattle are generally the result of low facility scores, low slope scores and in
downtown low barrier scores. Besides some multi-use paths Seattle bicycle infrastructure is very
limited. Additionally, a majority of new bicycle facilities build are sharrows or other facility types
that were assigned low scores in the analysis. This is especially notable in the downtown core where
very few bicycle facilities exist. Most of the facilities that do exist are located at the fringe of the core
area. This lack of facilities combined with low barrier scores and low slope scores results in a very
low score for Downtown Seattle.
Discussion of the barrier score is below in Section 7.2.1.

Seattle
Cycle
Analysis
Current
Bikeability
Portland Cycle Analysis
Current Bikeability

Legend
< 45
45 - 50
50 - 55
55 - 60
60 - 65
65 - 70
70 - 75
Legend
75 - 80
< 45 80 - 85
45 - 50 85 - 90
50 - 55 90 >
55 - 60
60 - 65


65 - 70
70 - 75
75 - 80
80 - 85
> 90 0 0,5 1 Miles

0

0,5 1 Miles

Figure 16 - Current Bikeability

7.1.2 Potential
The potential bikeability maps are measures of how bikeable an area can possibly become. It
measures inherent factors that cannot be changed.

Seattle
Cycle
Analysis
Potential
Bikeability
Portland Cycle Analysis
Potential Bikeability

Legend
< 45
45 - 50
50 - 55
55 - 60
60 - 65
65 - 70
Legend
70 - 75
< 45
75 - 80
45 - 50
80 - 85
50 - 55 85 - 90
55 - 60 90 >
60 - 65
65 - 70


70 - 75
75 - 80
80 - 85
85 - 90
90 > 0 0,5 1 Miles

0

0,5 1 Miles

Figure 17 - Comparison of Potential Bikeabilty


These maps show that while Portland has large contiguous areas that are very bikeable Seattle has
islands of bikeability. Both cities could have bikeable downtowns, especially on the northern edge
were land use and terrain are very favorable. It also shows that center city neighborhoods of
Portland are very bikeable, with center city neighborhoods of Seattle as bikeable but not to the same
degree. Outside of the central areas Portland has a much large potential than Seattle. Both Portland
and Seattle have neighborhoods centers or districts that are roughly as bikeable. The difference
comes in areas outside of these centers, mostly in single-family neighborhoods. While these
neighborhoods in Portland are still fairly bikeable, they are often much less bikeable in Seattle.

This means that while Portland in the long run will be a more bikeable city, Seattle also has the
potential to become a highly bikeable city in specific neighborhoods. While Portland can take a
more general and less target approach Seattle must be me deliberate about its actions. Additionally
transit and walking must play an integral role in Seattles transportation system, especially in those
area which are less bikeable.

7.2 Discussion
Below is an assortment of questions and problems that came to mind at some point during this
project.

7.2.1 Barrier Factor


All of the factors used in this analysis were well understood, except for the barrier factor. It appears
that the original analysis used some undisclosed measure like dead ends. In this case the more
advanced measure of connectivity used in this analysis would automatically take dead ends into
account.

This difference is extremely important and certainly leads to a lower than appropriate score for
Downtown Seattle. Downtown Seattle by far had the lowest score for this factor. In the current
bikeability analysis MCE this measure contributed 25% of the score. So not only is this measure
different than what was envisioned in the initial analysis, but its only major impact is on Downtown
Seattle, where it significantly reduced the score. This means that the score of Downtown Seattle
should be higher by a maximum of 25%. This is the most significant problem with this analysis.

7.2.2 Agreement with Intuition


This is probably the most important question of all. Any sound analysis must be able to recreate
what people experience and feel. A bicyclist should be able to say yes Im more likely to bike in the
area with a higher score than an area with a lower score. There are a few caveat to this but in general
this should be the result.

7.2.3 Density vs. Distance


It was hard to choose which type of measure to use when converting line features (barrier and
facility score) into surfaces. Most GIS analyses typically use distance to nearest feature to do this
conversion. However, distance to nearest measure was not used for several reasons.
First both the coverage and quality of the bicycle network were important aspects of this analysis.
While distance to nearest facility does a good job of measuring coverage it does a poor job of
measuring the quality of the network. Furthermore, it will in some respects underreport the score
of a dense, high quality network, while over report the score of low quality, but more distributed
bicycle facilities like sharrows and signed bike routes. This is because most analyses that use distance
to nearest feature do not distinguish between the different types of bicycle facilities.

Another reason density was used is that it localizes the affect of a facility. While the distance to
facility measure continues regardless of absolute distance, the density measure allows for an absolute
distance limit, beyond which the facility will not attribute any value.

The final reason is a bit complex to explain. If distance to feature was used with the same
methodology it would lead to an over reporting of scores for areas which have many different types
of facilities in close proximity. For example, if a street has a bike lane, areas around it will be within
a short distance of one type of facility (bike lane) which has a weighting factor of 6 as seen in Section
5.1.2. However, if for some reason a one block portion of the street uses sharrows instead of bike
lanes, that area is now within a short distance of two facility types, one with a weighting factor of 6
and the other with a weighting factor of 3. This results in a higher score where the transition occurs,
although in reality it should have a lower score. In this way areas with disjointed bicycle facilities
score much high than they should.

7.2.4 Input Data Quality


As mentioned throughout the report good input data is critical to any analysis. While the quality of
data used for this analysis is acceptable for this project it is not good enough for a more detailed
analysis. For example the scoring of barriers was overly simplified and should be done in a different
way in the future. Additionally several data points used in for the connectivity score were obviously
wrong but were not corrected.

7.2.5 Appropriateness of re-classed Values


All of the factor scores were re-classed. This means that a subjective decision had to be made as to
what constitutes a high value or low value for each factor. Sometimes this decision is easy. It is
obvious that flat terrain should have the highest values. However, what should be the lowest value?
Should it be the largest value in either city? Should it be the maximum slope that a competitive
cyclist will climb, or should it be the maximum that a novice would climb? Decisions like this had to
be made throughout the analysis, and often they were based simply on intuition. A more formalized
and verifiable approach should be used in the future. For land use, barrier and facility scores the
highest and lowest values were taken from the highest or lowest values found in either city. For
connectivity and slope cutoffs were used.

7.2.6 AHP Not Used


Despite initially intending to use AHP, it was not used to determine weighting factors as discussed
in Section 5.3. Values from an AHP could have been created but the acceptance of them by other
could be problematic. Rather asking a bodies like the Seattle Bicycle Advisory Board to use AHP to
set weighting factors would be best.
7.2.7 Level of Detail
The level of detail for this analysis was very high, possibly too high. For example land use scores are
very precise. Previously they were average over some area, lowering scores for some areas and
raising scores for other areas. In this analysis the scores did not changed like that. Slope scores have
similar problems. In the original analysis one large hill or crevasse in a zone would not have affect
its score, however with this analysis the impact of the hill is very localized on one specific area of the
zone. So while the entire neighborhood might be very bikeable, that one location will not be.

7.2.8 Important of Hills


How important are hills? This question is critical for Seattle because it is hilly, especially compared
to many very bikeable cities. A good understanding of how important hills are to riders is thus very
important to understand what the long term potential of biking in Seattle is. Perhaps some areas of
the city should be heavily invested in while other areas simply dont have the potential to warrant
large investments.

Another possible ramification could be that a different focus of bicycling in the city is warranted.
While Portland has consistent and high scores partly because of the flat terrain, Seattles bikeable
and flat areas are much more localized. Thus, it might be a more productive to focus on bicycle
facilities that connect the most bikeable areas rather than blanketing the city with bicycle facilities
like Portland has done. In this way investments in bicycle facilities would focus on trips that are in
areas that inherently support biking.

7.2.9 Riding Perpendicular to Hills


This is an important to note. While riding up and down hills isnt very attractive for bicyclist, riding
along the top or side of a hill at a constant elevation is not nearly as bad. Two good examples of this
are Eastlake and Westlake when traveling north or south. While streets in these neighborhoods are
not flat, they certainly arent as bad as would be expected from the slope map. The slope map always
presents a worst-case scenario while bicyclist are likely to use routing are the best-case scenario.

You might also like