Identification of sliding surface using electrical-resistivity tomography for landslide mitigation: A case study of the Cibitung Landslide

Landslide mitigation efforts require a knowledge of the geometry and depth of the sliding surface present in a landslide body. Electrical resistivity tomography is the most common geophysical method used in landslide investigation. This paper presents the results of a series of electrical resistivity tomography surveys performed using a dipole-dipole array configuration to identify the sliding surfaces within a landslide body located in the Cibitung landslide area. The ERT images parallel the landslide body suggest that the sliding surface located below the crown is characterized by a bedrock layer with very low resistivity values. This bedrock layer continues down-slope at a maximum depth of 9 m below the ground surface. In contrast, the landslide body is characterized by a higher resistivity value. Based on the interpretation of the ERT images, the landslide involved a non-circular deep sliding surface. The results of this study have been used, in combination with the geotechnical drilling data, to construct the landslide cross-section necessary to analyze landslide stability and subsequently to recommend a landslide stabilization measure.


Introduction
Landslide is the most common natural disaster in Indonesia, especially in West Java.Landslides commonly occur during and after a heavy rainfall period.The occurrences of rainfall-induced landslides are very common in volcanic hillslopes with deep groundwater levels.The development of groundwater level is of course controlled by the presence of a low permeability rock layer.Such a rock layer may act as a sliding surface.The information on sliding surfaces is required for landslide stability analysis and stabilization work.Thus, it is necessary to have a comprehensive knowledge of the depth and geometry of a sliding surface within the landslide masses.
A deep-seated landslide disaster took place on the Cibitung slope on 5 May 2015 (figure 1), which destroyed a geothermal pipeline, and several houses and claimed nine lives [1].The landslide occurred following a long period of rainfall from March to April 2015, producing an antecedent rainfall of 900 mm.The landslide occurrence was initiated by the development of tension cracks at the hilltop top few months before the event.To stabilize the landslide area, some landslide stabilization works were conducted by the geothermal company, including the removal of large amounts of landslide materials, installation of 2-m high gabion walls at the landslide toe, and reconstruction of slope geometry.Despite the soil removal, much landslide materials are still present in the landslide area.Thus, the landslide materials may still have a potential to slide especially during heavy rainfall, and thus introduce a landslide hazard to the newly constructed geothermal pipeline [1].Several geophysical exploration methods can be utilized to estimate the geometry and depth of sliding surfaces within a landslide mass.The most common method is the electrical resistivity tomography method which has been used in many landslide studies worldwide [2][3][4][5][6].However, the results of this method still require validation using geotechnical data.This paper aims to present the results of a series of electrical resistivity surveys to identify the sliding surfaces within the Cibitung landslide body.The study also involved validation of the estimated sliding surface using some geotechnical drilling data.The objectives of this study were to (a) obtain the electrical resistivity profiles in the landslide body, (2) estimate the geometry and the depth of the sliding surface, (3) compare with the geotechnical drilling data, and (4) recommend a landslide stabilization measure.

Geology of the study area
As seen in figure 2, the geology of the Cibitung landslide area primarily consists of Quaternary volcanic rocks [7].The volcanic rocks originated from the Malabar-Tilu volcanic rocks (Qmt), consisting of volcanic tuff, and laharic breccia, the undifferentiated eflata deposits of old volcanic rock (Qopu), and the Waringin-Bedil andesites (Qwb).Field mapping suggests that the landslide materials involved mainly the Malabar-Tilu volcanic rocks (Qmt) and the sliding surface is located at the Waringin-Bedil andesites [1].Two faults, namely the SW-NE trending strike-slip fault and the N-S striking normal fault, also exist in the landslide area, crossing adjacent to the center of the landslide crown area.The field evidence associated with these two faults is the presence of hot springs in the vicinity of the study area.The hydrological condition of the landslide area is mainly governed by the spring water and the shallow groundwater table.Spring water notably appears to seep out from the landslide crown area after heavy rainfall, flowing into the collection pond and then directed to the main ditch.Hot springs are also observable at several locations in the vicinity of the landslide area.

Methodology
To achieve the objectives, this study involved two methods, namely geophysical survey and numerical modeling.The geophysical tomography was carried out using the electrical resistivity method, while the numerical modeling was performed to calculate the slope stability using the computer program SLOPE/W.These study methods are described in detail below.

Electrical resistivity method
The Supersting R8/IP equipment was used to conduct electric resistivity measurements in the landslide area (figure 3).The electrical resistivity equipment was connected, using a multi-core cable, to 56 stainless steel electrodes, laid in a straight line at equal distances apart.Figure 3 shows the survey lines in the study area.The resistivity measurements were conducted in two ERT lines parallel to the slope direction (E-W), and two lines perpendicular to the slope direction (N-S).The electrode spacing was set at 3 m for E-W lines and 4 m for N-S lines to obtain the twodimensional ERT profiles up to a maximum depth of 40 m.To provide comprehensive data resolutions, the dipole-dipole array configuration was selected for data protocol acquisition (figure 4).The elevations of each electrode were recorded using a total station.

Numerical modeling
The computer program SLOPE/W [8] was used to perform a slope stability analysis on a selected slope profile to validate the sliding surface estimated from the geophysical survey and to evaluate the stability of the landslide masses under the effect of a pre-existing groundwater level.The hillslope model was established using some geotechnical borehole and laboratory soil properties data [1].The generalized limit equilibrium (GLE) method was chosen to determine the safety factor of the landslide.

Electrical resistivity images in the E-W direction
The results of the electrical resistivity measurements along the lines in the W-E direction are shown in figure 5.The resistivity values range from 2 to 400 -m, indicating that the landslide area is composed of different soil/ rock with different degrees of saturation.Referring to this figure, the brecciated lava (see figure 2) has low resistivity values of less than 20 -m, suggesting that the brecciated lava has high water content.In contrast, the soils deposited above the brecciated lava have higher water content, characterized by resistivity between 20 and 100 -m.The layer of deposited soils is likely to represent the landslide materials.From these results, a clear boundary of the resistivity contrast exists between the brecciated lava and the deposited soil.This resistivity boundary is likely to represent a sliding surface (black broken line in figure 5).Some patches of low resistivity zone are also noticeable in the landslide materials in Line-01, indicating the presence of localized perched water table.Below the brecciated lava, the rock layer of unknown origin has higher resistivity values between 30 and 100 -m.From this figure, the depth of the sliding surface is at a maximum depth of 9 m.

Comparison with borehole data
To validate the location and geometry of the sliding surface estimated from the resistivity survey, a landslide cross-section was constructed using some geotechnical borehole data.Figure 7 presents the constructed soil/ rock stratification of the landslide area for Line-01.This figure shows the presence of 9-m thick landslide materials deposited above the brecciated lava as seen in BH-15.Thus, a sliding surface is very obvious to exist at a maximum depth of 9 m, and in good agreement with the resistivity tomography images of Line-01 and 02 shown in figure 6. Numerical modeling was also performed using a computer program SLOPE/W [8] to validate the geometry of the sliding surface (figure 8). Figure 8(a) shows that the sliding surface developed in the landslide model is in good agreement with that of the ERT survey.The numerical modeling also shows that the safety factor of the landslide area is at a critical value, suggesting the landslide material is still unstable under an existing high groundwater table.The result of numerical modeling shown in figure 8(b) indicates that lowering the groundwater table increases the landslide stability.Thus, this numerical modeling recommends the application of any method of groundwater control such as a siphon drain, or sub-horizontal drain to stabilize the landslide area.

Conclusions
According to the results of the resistivity survey conducted using dipole-dipole array configuration, the resistivity values of between 20 and 100 -m are associated with landslide materials, while the brecciated lava layer has resistivity values lower than 20 -m.The boundary of resistivity contrast between landslide materials and the brecciated lava is identified as the location of the non-circular sliding surface of the Cibitung landslide, with a maximum depth of 9 m.Based on this estimated location and geometry of the sliding surface and the geotechnical data, this study shows that the landslide materials are still in unstable condition, and thus lowering the groundwater level is recommended to increase its stability.

Figure 1 .
Figure 1.Aerial photos of the Cibitung landslide area: source area and landslide crown (left), and the landslide material deposition area (right)

Figure 3 .
Figure 3. Map of resistivity survey locations showing four survey lines (Line-01 to Line-04) and the outline of the landslide area (white line).

Figure 4 .
Figure 4. Dipole-dipole array configuration used in this study.

Figure 5 .
Figure 5. Electrical resistivity tomography images for Line-01 (left) and Line-02 (right) in the E-W direction.The sliding surface is indicated by a black broken line.

Figure 6
presents the results of the electrical resistivity measurements along the lines in the N-S direction.Similar to the resistivity images in the E-W direction, the resistivity values range from 2 to 400 -m.The brecciated lava is also represented by a low resistivity layer, extending from North to South.This figure also presents a clear resistivity contrast located below the ground surface, demonstrating the location and geometry of the sliding surface.

Figure 6 .
Figure 6.Electrical resistivity tomography images for Line-03 (left) and Line-04 (right) in the N-S direction.The sliding surface is indicated by a black broken line.

Figure 7 .
Figure 7. Landslide cross-section based on geotechnical borehole data in the E-W direction.

Figure 8 .
Figure 8. Results of numerical modeling of slope stability (a) under a high groundwater level condition, (b) low groundwater level condition.