【文章內容簡介】 array surveying with multiple current electrode positions was employed. The maximum depth of investigation in the inverted depth sections was around 50m, which was more than enough to reach the bed rock. The number of data points collected for each line were around 2,500 for the long line with 5 m spacing, around 2,900 for the detailed line with 2 m spacing and 850–1,050 each for the three lines crossing the dam. Evaluation of the field data was done via 2D inverse numerical modelling (inversion),using the software Res2dinv1, the inversion,2D structures are assumed。 ., the ground properties are assumed constant perpendicular to the line of the profile, while the current electrodes are modelled as 3D sources .This is valid for cross section measurements as topography is taken into account. However, evaluation of surveying along the dam will result in severe 3D effects, but nevertheless the approach is valuable for detecting anomalous zones. The inversion is done through the generation of a finite element model of the resistivity distribution in the ground, which is adjusted iteratively to ?t the data so that the differences between the model response and the measured data(the model residuals)are minimised. This can be done by either minimising the absolute values of the differences (inversion with L1norm or robust inversion), or minimising the squares of the differences (inversion with L2norm or smoothnessconstrainedleastsquares inversion) (Loke et al. 2021). Robust (L1norm) inversion is more capable of handling sharp boundaries in the model and was used for all measurements, due to the expected large contrasts in electrical properties of the involved materials. Resistivity data from the inverted models were plotted with the software Erigraph by using linear interpolation between neighbouring cell values. Erigraph es with the ABEM equipment. The resistivity measurements along the dams as well as the cross lines provided data of high quality, resulting in low model residuals ( 1–3%) for the inverted sections. The upper part of Fig. 2 shows an inverted section from the 5 m spacing survey along the entire dam XY. Data from the 2 m spacing and the 5 m spacing survey along the central part of the dam were bined and inverted jointly (Fig. 2). A highresistive bottom layer is evident throughout the section, at depths that correlate well with the somewhat sparse and uncertain in formation available about variations in bedrock level. The highresistivity bottom layer es close to the surface at the ends of the line. The central part the line exhibits three principal layers above the highresistive base layer. Starting from the top。 a couple of metres thick layer of around 100 W m, a mediumresistive layer of a few hundred O hmm down to approximately 10 m depth, and below that a layer around 100 W m reaching circa 30 m depth. The cross section at chainage 0/471 m (Fig. 3) shows a sequence of zones with different resistivity with steep to vertical interfaces, going from low ( 40 W m) over intermediate ( 200 W m) and rather low(100 W m) to very high(1,000 W m).These zones correspond well with that which can be expected from the way the dam is built. Starting from the upstream side on the left in the diagram, the lowresistive zone that continues up to 20 m corresponds to saturated sand, and with a resistivity of 10 W m of the pore water, a resistivity of 40 W m is reasonable if Archie’s law is applied to the high porosities that can be expected for wellsorted sand. Thanks to the wellsorted character of the sand, it is possible to identify the groundwater surface in the diagram, being at the surface in the upstream edge of the diagram and decreasing to are latively stable level of 185 masl going in the downstream direction. This fits perfectly with the observation wells at chainage 0/500 m that show a level of 185–186 masl. The higher resistivity above the level 185 masl corresponds to the tailings sand above the groundwater level. The zone with higher resistivity next to the lowresistive zone matches up with the position of the upstream support fill of the dam, with a water level at 185 masl visible as a decrease in resistivity. The next zone, of relatively low resistivity, corresponds to the core of the dam, where the inclined shape at the top followed by a more vertical part is well visible. There is also an indication of a reverse slope for the lowest part of the zone that can be interpreted as the core, however, not as inclined as on the drawing (Fig. 1).The highly resistive downstream zone corresponds to the downstream rockfill, whereas the filters cannot be readily identified. The cross section at chainage 0/492 m (Fig. 3) agrees well with the one at 0/471 m, but the top highresistive part corresponding to the upper part of the upstream support fill is not so well developed. The latter explains the highresistive zone at 0/471m in the sections measured along the dam (Fig. 2).The extra lowresistive part of the zone interpreted as the core might be an indication of a change in the properties connected to the sink hole development. The ground water level was recorded at elevation 165 masl at +3 m on the cross section (chainage 0/500 m). The cross section at chainage 0/310 m (Fig. 3) is similar to the one at0/471 m with one important exception. The highresistivity zone corresponding to the upstream filter is interrupted by a large zone (around 10 m high) with resistivity below 100 W m. This can be interpreted as an effect of anomalous seepage through the dam, which fits well with the occurrence of old sink holes. Measured groundw