You are on page 1of 13

Geophysical Prospecting, 2001, 49, 1325

Detecting small-scale targets by the 2D inversion of two-sided three-electrode data: application to an archaeological survey
M. Emin Candansayar and Ahmet T. Bas okur
niversitesi, Fen Fak., Jeofizik Mu Ankara U h. Besevler, 06100 Ankara, Turkey Received April 1999, revision accepted June 2000

ABSTRACT The detecting capabilities of some electrical arrays for the estimation of position, size and depth of small-scale targets were examined in view of the results obtained from 2D inversions of apparent-resistivity data. The two-sided three-electrode apparentresistivity data are obtained by the application of left- and right-hand poledipole arrays that also permit the computation of four-electrode and dipoledipole apparent-resistivity values without actually measuring them. Synthetic apparentresistivity data sets of the dipoledipole, four-electrode and two-sided three-electrode arrays are calculated for models that simulate buried tombs. The results of twodimensional inversions are compared with regard to the resolution in detecting the exact location, size and depth of the target, showing some advantage for the twosided three-electrode array. A field application was carried out in the archaeological site known as Alaca Hoyuk, a religious temple area of the Hittite period. The twodimensional inversion of the two-sided three-electrode apparent-resistivity data has led to locating a part of the city wall and a buried small room. The validity of the interpretation has been checked against the results of subsequent archaeological excavations.

INTRODUCTION The two-dimensional (2D) inversion of direct current (DC) sounding data has become a standard tool for the investigation of targets buried at shallow and intermediate depths. The applications include archaeology, environmental and hydrogeological works, mining exploration and engineering studies. When using the traditional electrode arrays the data acquisition is usually carried out by expanding electrodes along a measurement profile. The apparent-resistivity data obtained from all soundings are then used as the input of a 2D inversion algorithm. Pelton, Rijo and Swift (1978) and Sasaki (1981) first introduced inversion algorithms applied to the dipoledipole and Schlumberger arrays, respectively. Later, Loke and Barker (1996) introduced a fast 2D inversion method for the Wenner and dipoledipole electrode arrays. This algorithm is based on a quasi-Newton method which calculates an approximate value of partial derivatives of the

apparent-resistivity data with respect to model parameters. The finite-difference technique is used for the computation of the model response. One of the remarkable developments in DC methods is the use of modern measuring systems with multicore cables which allows automatic switching between electrodes. For practical field operations, the electrodes are located at equally spaced measurement stations. For example, Morris, Rnning and Lile (1997) described a data acquisition system that simultaneously collects Schlumberger and two dipoledipole data. However, they did not indicate a quantitative interpretation tool. Acworth and Griffiths (1985) and Griffiths and Turnbull (1985) described Wenner tripotential apparentresistivity measurements to reduce the number of ambiguities arising in the qualitative interpretation of apparent-resistivity pseudosections. The tripotential method produces three apparent-resistivity measurements for each position of a collinear equispaced quadripole of electrodes. They used the

q 2001 European Association of Geoscientists & Engineers

13

14

M.E. Candansayar and A.T. Bas okur

ratio of beta and gamma apparent-resistivity values to obtain a pseudosection presentation that is close to the trueresistivity distribution of the subsurface. Van Overmeeren and Ritsema (1988) presented a similar measuring system. Dahlin (1996) used the advantages of both automated measuring systems and 2D inversions. He illustrated some 2D inversion examples performed by multi-electrode Wenner systems, known as `continuous vertical electrical sounding'. Most of the above-mentioned automated measuring systems and 2D inversion algorithms use the traditional collinear four-electrode arrays, but it is also possible to use three-electrode arrays with one current electrode and two potential electrodes to measure one apparent-resistivity value. The second current electrode is located at a far distance that, for practical purposes, is assumed to be at infinity. In the field procedure, two apparent-resistivity values are measured by switching the current between the two current electrodes that are located on the left- and right-hand sides of the array centre. In this way, Karous and Pernu (1985) combined the sounding and profiling techniques by using AMN and MNB half-Schlumberger arrays. The Schlumberger apparent resistivity could be derived from the mean of the two halfSchlumberger apparent-resistivity values. The aim of Karous and Pernu (1985) was to detect thin conductors. Schulz and Tezkan (1988) recorded Schlumberger and the corresponding half-arrays simultaneously to examine the dimensionality of the subsurface. Lile et al. (1994) obtained two sets of pole dipole apparent-resistivity values in order to eliminate the dipping effect that occurred in the pseudosection plotting. The aim of the above-mentioned papers that use some combination of the three-electrode arrangements was to increase the sensitivity of the DC method to the targets. However, the 2D inversion using two sets of apparentresistivity data measured by the two-sided three-electrode arrays was not attempted. The aim of this paper is to examine whether the 2D inversion of two-sided three-electrode arrays produces better outputs than that of collinear four-electrode arrays. DESCRIPTION OF THE TWO-SIDED THREEELECTRODE SYSTEM For efficient use of the automated measuring systems, the electrodes are laid out at equally spaced distances over a measurement profile. Many types of apparent-resistivity data could be derived by the selective combination of the current and potential electrodes. In principle, apparent-resistivity values corresponding to all possible electrode arrangements

can be computed from the polepole data by using the superposition principle (see Xu and Noel 1993). The current can be injected through a single electrode and then the resulting voltage in the remaining electrodes can be measured. The process can be terminated when all electrodes have been used as the current source. However, the pole dipole data are less influenced by the noise sources in comparison with the polepole data. The remote potential electrode amplifies the effect of the telluric and other types of noise in the polepole array. The poledipole array yields a better signal strength than the dipoledipole array. However, it gives less resolution in comparison with the dipoledipole array (Sasaki 1992). To overcome this disadvantage, we examined the 2D inversion of the two-sided three-electrode configuration that is obtained by the combination of two symmetrical poledipole arrays (Lile et al. 1994). This particular layout of the two-sided three-electrode array is shown in Fig. 1. The letters A, B and C represent the current electrodes whilst M and N show the location of the potential electrodes. Any two consecutive electrodes can be used to measure a potential difference. The current electrodes A and B are located on the left- and right-hand sides of a pair of potential electrodes, respectively, while the electrode denoted by the letter C is located along the line perpendicular to the centre of the measuring profile. The potential contributed by this source at the electrodes M and N is relatively small if the electrode C is fixed at a great enough distance that, for practical purposes, is considered as being infinite. Two measurement sequences are performed for each position of the potential electrodes. The first series of apparent-resistivity measurements is conducted by applying a current to the ground through the C and A1, A2, , An electrodes. The measurements are repeated in the same way using the C and B1, B2, , Bn electrodes. The apparent-resistivity values measured using C Aj and C Bj current electrodes in pairs constitute left and right apparent-resistivity values corresponding to the same potential electrode pair (M and N). By shifting the potential electrodes to the next electrode locations and repeating the same measuring sequence, a set of apparent-resistivity values along the measuring line is acquired. The apparent resistivity r aA (AMN) can be defined as

raA kA

DV AMN ; I

where DVAMN is the potential difference between the M and N electrodes caused by the current (I) applied through the source (A) and sink (C) electrodes. The geometric factor, kA,

q 2001 European Association of Geoscientists & Engineers, Geophysical Prospecting, 49, 1325

Detecting small-scale targets by 2D inversion

15

Figure 1 Field layout of the two-sided three-electrode measurement system. A and B denote the current electrodes. C is the current electrode located at a distant point. M and N are the potential electrodes. O is the array centre and a is the distance between two consecutive electrodes.

of the array is given by kA 2p a n n 1 1; n 1; 2; ; 2 where a is the distance between two consecutive electrodes. The apparent resistivity r aB (MNB) corresponding to the case where the current is applied through the B and C electrodes is obtained in a similar way, giving

raB kB

DV MNB ; I

where the geometric factor kB is given by kB 2pann 1 1: 4 Considering the conventional application in which the current is injected through the A and B electrodes, the following equations can be obtained for the apparent resistivity r aAB (AMNB) and for the geometric factor kAB:

electrodes to that between the potential electrodes is large enough. It is equal to the Wenner array when n becomes equal to unity. It is expected that the average resistivity generally has better resolution than the Wenner array and it can be considered as a modified version of the Schlumberger array for the efficient use of multicore measuring systems. All of the r aA, r aB and r aAB apparent-resistivity values should be equal to each other over a one-dimensional earth. Over more complex structures, they will show different behaviours from each other depending on the resistivity distribution of the subsurface. The measuring system described above also provides that the dipoledipole apparent-resistivity values can be obtained from two consecutive poledipole apparent-resistivity values by making use of the superposition principle,

raAB kAB
and

DV MN I

raD

n11 n 1 2rn a 1 nra ; 2

kAB p a n n 1 1: Since 2kAB kA kB ;

6 7

n11 where rn are two successive poledipole apparenta and ra resistivity values measured either on the left- or right-hand side of the potential dipole.

the apparent resistivity r aAB can be computed from the two other apparent-resistivity values by application of the superposition principle without actually measuring the potential difference (Karous and Pernu 1985), yielding

I N V E R S I O N O F S Y N T H E T I C D ATA A comparison test has been applied to examine the resolution obtained with earth models derived from the 2D inversions of three- and four-electrode synthetic data. A computer program that handles two types of electrode array was developed as an adaptation of the algorithm published by Uchida and Murakami (1990). This algorithm is based on the previous works of Sasaki (1981, 1989), but the current program written by Uchida and Murakami (1990) includes the modelling of topography. The forward and inversion schemes

raAB

raA 1 raB : 2

Hence, r aAB can be named as the average apparent resistivity. The behaviour of the average apparent resistivity is similar to the Schlumberger array for large n values satisfying the condition that the ratio of the distance between the current

q 2001 European Association of Geoscientists & Engineers, Geophysical Prospecting, 49, 1325

16

M.E. Candansayar and A.T. Bas okur

utilize the finite-element and damped least-squares methods, respectively. Figure 2 shows a representative model that could be encountered in many archaeological sites in Turkey (in our experience). Two high-resistivity blocks (500 Vm) simulate buried walls or tombs that are currently covered by unconsolidated young sediments (10 Vm). We assume that the walls are buried at a depth of 1.5 m and the distance between the two objects is 2 m. The basement sloping on the left-hand side of the panel has an intrinsic resistivity value of 100 Vm. The apparent-resistivity pseudosections for this model were calculated for a variety of electrode arrays. The station spacing is 2 m and the location of the stations with respect to the target bodies is illustrated on the top of the model. AB/2 values of 3, 5, 7, , 17 m were used at each measurement station for the data acquisition. Figure 3(a,b) show apparentresistivity values corresponding to the poledipole arrays of AMN and MNB, respectively. Figure 3(c,d) show apparentresistivity values corresponding to the four-electrode array (AMNB) and the dipoledipole array (ABMN) obtained from (8) and (9), respectively. All electrode arrays clearly indicate the existence of a sloping boundary except the MNB poledipole array. However, the qualitative interpretation made by visual inspection of the pseudosections gives no indication of the shallow resistive targets. Hence a quantitative interpretation becomes necessary for an accurate interpretation of the data. Figure 4(a) illustrates the model obtained from the 2D inversion of the four-electrode data. The initial guess was a 100 Vm half-space. An RMS value of 0.088 was reached after six iterations. The calculated resistivity values of the targets are 22.1 and 20.6 Vm. The average resistivity of the surrounding medium is about 15 Vm. These results indicate that it would be difficult to distinguish the target from the

surrounding medium in real exploration conditions because of noise contamination of the data and a more complicated resistivity distribution in the subsurface. Figure 4(b) shows the resistivity model inverted from the dipoledipole data. The number of iterations and the RMS value are 4 and 0.31, respectively. The calculated intrinsic resistivity values of the target blocks are 17.6 and 16.7, respectively. The average resistivity of the surrounding medium is fairly estimated as 12 Vm. The individual inversions of r aA and r aB poledipole apparent-resistivity data did not produce better results than the inversion of the dipoledipole data and are therefore not shown here. This is in agreement with the inversion results of borehole data obtained by Sasaki (1992) showing that the poledipole array has less resolution than the dipoledipole array. Figure 4(c) shows the result of the 2D inversion of the twosided three-electrode data. An RMS value of 0.069 was achieved after seven iterations. The resulting RMS value is less than that of the previously mentioned arrays. The calculated intrinsic resistivity values of the targets are 49 and 46 Vm, respectively. The average resistivity of the host medium is accurately calculated as being 10 Vm. The contrast between the computed resistivity value of the targets and the host medium helps to differentiate the two target blocks from the surroundings. The sloping high-resistivity structure on the left-hand side of the model is identified more precisely in comparison with the solution produced by the 2D inversion of the other arrays. As mentioned above, the dipoledipole data have better resolution than the pole dipole data (Sasaki 1992). However, the data obtained from the two-sided poledipole data contain more information about the subsurface in comparison with the dipoledipole data, and consequently produce more precise inversion results. This is a representative example of several test runs. The other synthetic data inversions have also proved that the two-sided three-electrode array resolves small-scale targets better than do the conventional four-electrode arrays. FIELD STUDY Field data were collected from an archaeological site known as Alaca Hoyuk, in the northeastern part of central Turkey. The archaeological site consists of four cultural periods, Calcolitithic (35003000 bc), early Bronze Age (30002000 bc), Hittite (20001200 bc) and Phrygian. In the Hittite period, the site was an important religious centre surrounded by a city wall and it was close to Khattausha, the capital city of the Hittite Empire. The location of the city wall was not known although the

Figure 2 The two-dimensional model used for the calculation of the synthetic apparent-resistivity data.

q 2001 European Association of Geoscientists & Engineers, Geophysical Prospecting, 49, 1325

Detecting small-scale targets by 2D inversion

17

Figure 3 Apparent-resistivity pseudosections for (a) the left-hand-side poledipole, (b) the right-hand-side poledipole, (c) the four-electrode array and (d) the dipoledipole array, calculated from the model shown in Fig. 2 by the finite-element method.

archaeological excavations were begun in 1935 and the majority of the temple area has been uncovered. Based on archaeological information, an area having dimensions of 20 m by 28 m was selected for detailed investigations. An abandoned village built on the archaeological site occupied the terrain. The inhabitants of this village had been moved to another place by the state authorities in 1935. The soil covering the survey area is not natural and it consists of pebbles, sands and the remains of sun-dried bricks that were used in the construction of the adobe houses in the last centuries. Therefore, the cover was expected to have high-resistivity values. The buried wall was expected to strike along the northsouth direction and to be buried at a depth of about 24 m. The field measurements were made between 1 and 20

August 1997 over 15 parallel lines, each 20 m long, and spaced 2 m apart from each other. The distance between stations was also set at 2 m, thus there were 11 measuring stations along each survey line. The measurement direction was selected to be from west to east and the current electrodes A and B were located on the western and eastern sides of the array centre, respectively. A total of 2640 apparent-resistivity values were obtained for AB/2 spacings of 3, 5, 7, 9, 11, 13, 15 and 17 m (n 1 to n 8). Figure 5 shows representative apparent-resistivity pseudosections for profiles 12 and 14. The qualitative interpretation is difficult; one gets the impression that an undulating resistive top layer overlies a conductive basement. The dipole dipole apparent-resistivity values show the well-known inverse V-shaped artefacts. Figure 6 shows representative

q 2001 European Association of Geoscientists & Engineers, Geophysical Prospecting, 49, 1325

18

M.E. Candansayar and A.T. Bas okur

Figure 4 The final models inverted from (a) the four-electrode, (b) the dipoledipole and (c) the two-sided three-electrode synthetic apparentresistivity data.

apparent-resistivity maps plotted for electrode spacings of AB/2 7 m and AB/2 9 m for the four-electrode array and the left- and right-hand sides of poledipole arrays. The dipoledipole data have been plotted for n 3 and n 5 levels. Although the pseudosection presentation and the apparent-resistivity maps give useful information about the location of possible archaeological targets, the size and depth of the target bodies could not be estimated by a qualitative interpretation. Hence the 2D inversion method was applied to all profiles to obtain a more reliable picture of the subsurface. Since the archaeological work should be carried out with extreme care, a limited area that could be examined in a few excavation seasons was required by the archaeologists as an outcome of the electrical survey. In view of the 2D inversion results, two excavation sites were pinpointed. The suggested areas enclosed with rectangles are marked as EA1 and EA2 in Fig. 6. The area EA1 was excavated in 1998 and a part of the city wall was exposed as expected from the

geophysical interpretation. The exposed wall is buried at a depth of 1.8 m. The width of the wall decreases from 2.5 m to 0.5 m from north to south. The height of the wall is 1.5 m at the northern side, but it gradually becomes thinner towards the south and decreases to 0.5 m. The area EA2 was excavated in the summer of 1999 and a room having a kiln was exposed. Views of the exposed city wall and the room are shown in Fig. 7. We now discuss the results of the geophysical interpretation that was made without knowing the results of archaeological excavations. The forward modelling mesh consisted of 79 blocks. These model blocks were subdivided into 2400 finite-element cells. The resistivity of the bottom 18 model blocks was fixed in the inversion for rapid convergence of the algorithm. The size of the model blocks was adjusted in view of the information obtained from the two-sided gradient transformation (Karous and Pernu 1985; Candansayar, Bas okur and Peksen 1999). The number of apparent-resistivity data at each profile was

q 2001 European Association of Geoscientists & Engineers, Geophysical Prospecting, 49, 1325

Detecting small-scale targets by 2D inversion

19

Figure 5 Pseudosections of apparent-resistivity data measured at the Alaca Hoyuk archaeological site. The left- and right-hand panels show the data measured on profiles 12 and 14, respectively. The type of electrode configuration is shown between the panels.

176, 88 and 77 in the two-sided three-electrode, fourelectrode and dipoledipole arrays, respectively. Two independent iteration processes were applied using a homogeneous half-space model having intrinsic resistivities of 100 and 10 Vm, respectively. These differing initial guesses produced the same final model. The required misfit and the maximum limit of iterations were set at relative error values of 0.001 and 10, respectively. However, all inversions were terminated before the above conditions were satisfied. The convergence criterion stopped the algorithm since no further improvement in the data misfit would be achieved by performing extra iterations.

Figure 8 shows a representative example for the estimated model calculated from the data measured over line 12. The inversion of the four-electrode data was stopped at the ninth iteration with an RMS error of 0.91 (Fig. 8a). The location of the wall is well estimated, but the depth of the target appears to be incorrectly determined giving an impression that the target is very close to the surface. The block corresponding to the wall has a lower inverted resistivity value (118 Vm) than the block on top of it (204 Vm). The inversion of the dipole dipole data was stopped at the fourth iteration giving an RMS value of 2.79. The results are slightly better than those of the four-electrode configuration (Fig. 8b). The inversion of

q 2001 European Association of Geoscientists & Engineers, Geophysical Prospecting, 49, 1325

20

M.E. Candansayar and A.T. Bas okur

Figure 6 Plan views of apparent-resistivity data measured at the Alaca Hoyuk archaeological site. The left- and right-hand panels show the data measured with AB/2 7 m and AB/2 9 m for four-electrode and poledipole arrays, n 3 and n 5 for the dipoledipole array, respectively. The type of electrode configuration is shown between the columns.

the two-sided three-electrode data was stopped at the sixth iteration giving an RMS value of 0.83 (Fig. 8c). The location of the target is correctly determined in the inversion of all types of electrode array. However, initial examination of the inversion results gave an impression that the size of the target was larger than its actual dimensions. Again, the resistivity of the lower block (130 Vm) is slightly less than the inverted resistivity value of the top block (200 Vm). The majority of the inversions performed for the remaining profiles produced similar results. Since the inversion results clearly indicated the existence of a high-resistivity surface layer, we decided to

check the validity of the inversion results using a synthetic model. The model consisted of a very high-resistivity block that is embedded in a low-resistivity host medium. A horizontal layer overlay the target and the host medium. The resistivity of the surface layer was less than that of the target, but it was very high in comparison with the resistivity of the host medium. The synthetic data obtained from this model were computed and were then input to the 2D inversion algorithm. The results of the inversions were similar to the model derived from the inversion of the field data. The resistivity of the target derived from the four-electrode array

q 2001 European Association of Geoscientists & Engineers, Geophysical Prospecting, 49, 1325

Detecting small-scale targets by 2D inversion

21

Figure 7 (a) A view of the exposed city wall from the southern side (area EA1); (b) a view of the exposed room and kiln (area EA2).

was estimated to be less than that of the surface layer (similar to Fig. 8a). The two-sided three-electrode array produced intrinsic resistivity values that were slightly less than or close to the resistivity of the surface layer. This indicated the better resolution of this array in comparison with its four-electrode counterpart. Then, with the experience obtained from the inversion of the synthetic data, we interpreted the highresistivity block beneath the surface, having a bottom depth of 3 m, as an indication of an ancient wall, while the inversion results of the four-electrode and dipoledipole arrays indicated that the target was embedded in a highresistivity surface layer. The results of the 2D inversion can be presented as trueresistivity maps by contouring the intrinsic resistivity values of blocks, corresponding to the same depth range. The purpose of this type of presentation is to examine the lateral resistivity variations at a specified depth. Figure 9(a) shows

the resistivity distribution for the blocks that occupy the depth range between 0.41 m and 1.83 m. This depth range is represented by high-resistivity values and thus it is interpreted as the effect of the covering soil. Figure 9(b) illustrates the block depths ranging from 1.83 m to 3.11 m. The existence of the wall is clearly indicated in this depth range. This image also allows depth estimates to be made. The estimated depths of the top and bottom of the target are 1.83 m and 3.11 m, respectively. Examination of the highly resistive zone suggests that the wall exists from the first profile until profile 16. However, the results of the 2D inversion of the four-electrode and dipoledipole data suggest that the wall terminates at profile 14 (Fig. 10a,c). Since we assume that the detecting capability of the two-sided three-electrode array is superior to the four-electrode array, we outline the wall as shown in Fig. 9(b). The expected width of the target is 2 m as inferred from Fig. 9(b). The yellow dashed lines indicate the exact

q 2001 European Association of Geoscientists & Engineers, Geophysical Prospecting, 49, 1325

22

M.E. Candansayar and A.T. Bas okur

Figure 8 The final models inverted from (a) the four-electrode, (b) the dipoledipole and (c) the two-sided three-electrode apparent-resistivity data measured at the Alaca Hoyuk archaeological site. The letter W and the white outlined box indicate the position and size of the exposed city wall.

position of the city wall and help to check of the validity of the interpretation. The resistivity distribution in the next block range (3.11 5.20) is shown in Fig. 9(c). The resistivity values do not vary significantly except in a high-resistivity zone to the southwest of the survey area. The lack of resistivity variation indicates that the bottom of the wall lies within the block of depth range 1.833.11 m. However, the high-resistivity zone

between profiles 18 and 22 may indicate the existence of a new target having dimensions of 5 m by 5 m (marked EA2). The depth of the top and bottom of this building are estimated from the resistivity image as being 3.11 m and 5.2 m, respectively. Since archaeological excavations have exposed the targets, it is possible to compare the exact dimensions and depths of the targets with the 2D inversion results. The yellow dashed

q 2001 European Association of Geoscientists & Engineers, Geophysical Prospecting, 49, 1325

Detecting small-scale targets by 2D inversion

23

Figure 9 Plan views of the final model inverted from the two-sided three-electrode apparent-resistivity data measured at the Alaca Hoyuk archaeological site. The figure shows the variation of the intrinsic resistivity values inside the blocks within the same depth range. The resistivity maps correspond to the depth ranges (a) 0.411.83, (b) 1.833.11 and (c) 3.115.20, respectively. The excavation areas are outlined by black rectangles (EA1 and EA2). Yellow dashed lines indicate the exposed wall and room. The green lines mark the geophysical interpretation.

lines outline the excavated archaeological objects while the green lines show the targets as interpreted before the excavation. The width of the wall varies between 2.5 m and 0.5 m from north to south. In the central part of the excavation area (EA1), its width is approximately 1 m, which is less than our interpretation. The correct determination of the width would be possible if we took into account very high resistivity values as represented by the dark blue colour in Fig. 9(b). The depths of the top and bottom of the wall are exactly the same as our interpretation in the northern side of the area. The direction change in the left-hand side of the excavated area is also well estimated from the 2D inversion of

the two-sided three-electrode data. However, the exposed wall is continuous throughout the excavation area although the 2D inversions of the two-sided three-electrode arrays and the other arrays predict the termination of the wall at 16 m and 14 m, respectively. There are two possible reasons for this. First, the wall becomes thinner towards the southern side and its height reaches a minimum value of 0.5 m. Secondly, some limestone slabs support the wall on both sides and a mixture of soil and pebbles fills the core of the wall. It is expected that the resistivity of the core is not much higher than that of the host medium. The resistivity contrast between the host medium and the wall is mainly due to the

q 2001 European Association of Geoscientists & Engineers, Geophysical Prospecting, 49, 1325

24

M.E. Candansayar and A.T. Bas okur

Figure 10 Plan views of the final model inverted from the four-electrode (left-hand panels (a) and (b)) and dipoledipole data (right-hand panels (c) and (d)). The figure shows the variation of the intrinsic resistivity values inside the blocks within the same depth range. The top panels (a) and (c) correspond to the depth range 1.833.11 while the bottom panels (b) and (d) correspond to the depth range 3.115.20. Yellow dashed lines indicate the exposed wall and room. The green lines mark the geophysical interpretation. The resistivity scale is the same as in Fig. 9.

limestone slabs. These may be explanations of the discrepancies encountered in the estimation of the extension of the wall towards the south. In the excavation area of EA2, the 2D inversion of twosided three-electrode data was very successful in determining the location, depth and size of the room (Fig. 9c). However, the inversion of the four-electrode and dipoledipole data gives no clear idea about the existence of this small room (Fig. 10b,d). CONCLUSION The four-electrode data measured with a fixed pair of potential electrodes contain no information on whether a shallow small-scale target between two consecutive electrodes is located on the left- or right-hand side of the array centre. Information about the location of the target is obtained by shifting the centre of the electrode array along the measurement profile. However, individual three-electrode

sounding data that comprise a two-directional apparentresistivity data set would contain information about the location of the target. The shifting of the measurement system provides additional information. This is the main advantage of two-sided three-electrode arrays. The 2D inversions of the synthetic and measured apparentresistivity data also prove that the two-sided three-electrode array has better resolution for the detection of small-scale targets in comparison with its four-electrode counterpart and dipoledipole data. Although the results do not differ drastically from each other, the improved results produced by the two-sided three-electrode array could become important in detecting relatively small targets. It should be noted that our tests are not sufficient to compare the detecting capabilities of the above-mentioned arrays for deriving largescale structural information. However, we feel that the choice between three-electrode systems and conventional arrays should be made depending on the purpose of the survey, since the two-sided three-electrode arrangement requires twice as

q 2001 European Association of Geoscientists & Engineers, Geophysical Prospecting, 49, 1325

Detecting small-scale targets by 2D inversion

25

many measurements as conventional arrays. For example, if the intention is to obtain a general structural view of the survey area, then conventional arrays are preferred because fewer measurements are required. We recommend a twosided three-electrode array for a relatively difficult survey problem such as closely spaced targets and resistive objects underlying a resistive surficial layer. ACKNOWLEDGEMENTS Constructive comments by Associate Editor P. Valla and anonymous referees were helpful in improving the paper. This work was supported by the Scientific and Technical Research Council of Turkey (TUBITAK) under grant no. YDABCAG553. We thank Professor Dr Aykut Cnaroglu, the chief archaeologist of the Alaca Hoyuk excavation, for his continued contributions and interest. We also thank all the archaeology students who helped with the field measurements.

REFERENCES
Acworth R.I. and Griffiths D.H. 1985. Simple data processing of tripotential apparent resistivity measurements as an aid to the interpretation of subsurface structure. Geophysical Prospecting 33, 861887. Candansayar M.E., Bas okur A.T. and Peksen E. 1999. Detecting small-scale targets by the two-sided gradient transformation. Journal of the Balkan Geophysical Society 2, 100111. (Available on-line at http://bgs.ankara.edu.tr/.) Dahlin T. 1996. 2D resistivity surveying for environmental and engineering applications. First Break 14, 275283.

Griffiths D.H. and Turnbull J. 1985. A multi-electrode array for resistivity surveying. First Break 3, 1620. Karous M. and Pernu T.K. 1985. Combined sounding profiling resistivity measurements with the three-electrode arrays. Geophysical Prospecting 33, 447459. Lile O.B., Backe K.R., Elvebakk H. and Buan J.E. 1994. Resistivity measurements on the sea bottom to map fracture zones in the bedrock underneath sediments. Geophysical Prospecting 42, 813824. Loke M.H. and Barker R.D. 1996. Rapid least-squares inversion of apparent resistivity pseudosections by a quasi-Newton method. Geophysical Prospecting 44, 131152. Morris M., Rnning J.S. and Lile O.B. 1997. Detecting lateral resistivity inhomogeneities with the Schlumberger array. Geophysical Prospecting 45, 435448. Pelton H.W., Rijo L. and Swift C.M., Jr. 1978. Inversion of twodimensional resistivity and induced-polarization data. Geophysics 43, 788803. Sasaki Y. 1981. Automatic interpretation of resistivity sounding data over two-dimensional structures (I). Geophysical Exploration (Butsuri-Tanko) 34, 341350 (in Japanese). Sasaki Y. 1989. Two-dimensional joint inversion of magnetotelluric and dipole-dipole resistivity data. Geophysics 54, 254262. Sasaki Y. 1992. Resolution of resistivity tomography inferred from numerical simulation. Geophysical Prospecting 40, 453463. Schulz R. and Tezkan B. 1988. Interpretation of resistivity measurements over 2D structures. Geophysical Prospecting 36, 962975. Uchida T. and Murakami Y. 1990. Development of Fortran code for the two-dimensional Schlumberger inversion. Report. Geological Survey of Japan. Van Overmeeren R.A. and Ritsema I.L. 1988. Continuous vertical electrical sounding. First Break 6, 313324. Xu B. and Noel M. 1993. On the completeness of data sets with multi-electrode systems for electrical resistivity survey. Geophysical Prospecting 41, 791801.

q 2001 European Association of Geoscientists & Engineers, Geophysical Prospecting, 49, 1325

You might also like