Beneath the Surface: Identifying Subsurface Caves in "Gua Pandan" Using Integrated Electrical Profiling Method

One of the geopark tourism areas in East Lampung Regency, "Gua Pandan," has run into rock subsidence on the surface. As part of the subsidence prevention, indirect electrical methods between resistivity and chargeability profiling were applied to identify the presence of a subsurface cave in the study area. Two measurement lines were carried out with Wenner Alpha and Wenner Schlumberger arrays. Because the depth target is shallow (approximately 10 m) and to obtain a better resolution, each line has a stretch length of 70 m and 2 m electrode spacing. A line was measured over a known underground cave to produce a desired outcome, and the other was in an area with no cavities. Based on the results from each profile of resistivity and chargeability, an air-filled target has a value of over 5,000 𝛺𝑚 and under 6 ms, respectively. Then, integrated processing of both methods generated a metal factor (MF) profile to view the presence and estimated shape of the cave/ cavities. The result represents that an MF value under 1.5 ms/Ωm is a cavity, and solid rock is over 1.5 ms/Ωm. Also, the MF level from both configurations d elineates a similar section. However, a modest difference occurs in estimating the cavity shape geometry, 14 𝑚 × 4 𝑚 for Wenner Apha and 22 𝑚 × 6 𝑚 for Wenner Schlumberger. Furthermore, this study can be an initial step in safety assessment in the area.


Introduction
"Gua Pandan" is a cave formed from igneous rock and has become one of the natural diversity in East Lampung Regency, Indonesia [1]. The cave is only a few meters (2-3 m) below the surface, so the cover is mostly rock. According to direct observation, it has a dimension of 9 m width and 3 m height. Hence, several stone collapses were found at some points in the area. As part of the tourism objectives, safety becomes necessary for people visiting there. Thus, a preliminary study is required to identify the underground cavity's presence as an initial safety assessment step.
The second one is the induced polarization method, which is generally applied to mineral exploration, for instance, gold [18][19][20][21][22][23][24][25], copper [26,27], and galena [28,29]. However, only a few researchers utilized it for underground cave detection, all deployed dipole-dipole arrays [12][13][14]. The result is based on a chargeability value; it depends on the content of the cavity or rock mineralization so that it can be lower or higher [12]. A study indicated that a cave has a lower chargeability value [13].
Since the rock structure has been found on a surface in the area of this study, there is possible that the resistivity level demonstrates a higher value from the surface to the subsurface. Theoretically, it may be ambiguous to differentiate between a very dense stone and air (in a cavity); qualitatively, they have the same level of resistivity property [30,31]. However, they have an unequal level of dielectric property [30,31], which can lead to a chargeability response. A study occupied both techniques to depict the lava caves (similar to the cave's structure in this study) containing rock minerals with a dipole-dipole arrangement [12]. Therefore, this study will approach a similar mode [12] with disparate arrays of the Wenner Alpha and Wenner Schlumberger to view their effectiveness and compare their outcomes in detecting a cave. Considering the study area is roughly covered by rock entirely, these configurations are appropriate to deploy because their signal strengths are better than others [32]. Then, the processing output of each method will be integrated to avoid misinterpretation in reassuring the presence of an air-filled subsurface cavity and its geometry.

Methodology
An electrical profiling method is a 2D geophysical technique that visualizes some electrical properties of rock, such as resistivity and chargeability. In principle, it utilizes an electrical current flowing to the subsurface through a current electrode, and the response produces an electrical potential [31]; refer to Figure 1 for illustration. In practice, the electrode arrangement can be viewed in Figure 2 to get an electrical resistivity value.
with Δ is a potential difference between the electrode 1 and 2 , is an electrical current flowing from the electrode 1 to 2 , and is a geometrical factor (all units are in an international system). Figure 2. A general configuration of in-line electrodes to measure the electrical properties in the subsurface (modified from [31]).
On the other hand, when the current flows to a conductive material, the material will be polarized (charging state and potential level will reach a peak), like a capacitor, see Figure 3. Then, the discharging condition occurs when the current is turned off. In this form, the potential rate will gradually decrease until it reaches an equilibrium state at zero level (before being induced by an electrical current). . Charging and discharging concepts on a material; and a chargeability enumeration scheme shown in a shaded colour (modified from [31]).
In the discharging form, the observed chargeability ( ) is computed by applying an equation from [30] where is a peak potential value when the electrical current is turned on, and is a potential decay between 1 and 2 at its current shuts down (all units are in an international system).
Acquiring a subsurface profile from both techniques requires a particular arrangement of the electrodes. In this study, Wenner Alpha and Wenner Schlumberger configurations will be employed. Based on Figure 2, a fixed interval among the electrodes should be maintained for the Wenner Alpha array. Because the minimum depth target is 10 m, based on equation (4) according to [33], a minimum stretch length of 57.8 m needs to deploy. Then an electrode spacing of 2 m is implemented to get a better image resolution since the target has approximately a width of 8-10 m. Occupying 36 electrodes achieves a 70 m length of the dimension line.
with is a median investigation depth of the Wenner Alpha shape, and is a total length of a line (all units are in the international system). This study utilized an automatic resistivity-IP system to expedite the data acquisition process. Figure 4(a) presents a measurement sequence description of the shape.
The second array is the Wenner Schlumberger form. It is combined between Wenner and Schlumberger arrays. The Wenner shape is when a variable "n" equals 1 (the same as the Wenner Alpha form in Figure 4(a) when n =1). The Schlumberger shape has a requite when a distance of C 1 P 1 (or P 2 C 2 ) is greater than P 1 P 2 . Then its profiling technique should be when n >1. Its sequence illustration can be noted in Figure 4(b) with the same electrode arrangement as the first array. Because of its dissimilarity in the form, the estimated investigation depth is dissimilar too. The following equation is the formula to enumerate the investigation depth of the Wenner Schlumberger configuration (Z WS ) [33].
Each result between resistivity and chargeability properties will be integrated to generate a metal factor ( ) profile. It is a ratio of those properties, indicating a dependent change in the rock conductivity [34].
where the variables and , are chargeability and resistivity values of the rock. Equation (6) is a time domain metallic factor with its unit of ms/Ωm. In this study, this parameter aims to estimate the target geometry, which is not enough through those electrical properties analyses. Since the objective is an air-filled cavity, theoretically, it should have a high resistivity and low chargeability, so its metal factor may produce an insignificant content.
(a) (b) Figure 4. The actual array description applied in this study is (a) Wenner Alpha and (b) Wenner Schlumberger (adapted from [33]).
This study conducts two measurement lines, with the first line spread on the surface over a known air-filled cave and the other in a location without a cavity but close to the end of an underground cave track. This scheme is to validate the outcomes. To be clear, Figure 5 shows the survey design of this study. In the map, some subsidences occur in the area. Three subsidence spots and a cave mouth are north of line 1, and it crosses the cave track on the top ground. Then the second line is in the area of the cave discontinuity but about 10 m away from the dead end of the cave way, perpendicularly. This line 2 aims to confirm the calculation and cave interpretation; thus, in that line should not be a presence of an air-filled cavity.

Results and Discussion
The results from both configurations have a high resistivity level (top images in Figure 6 to Figure 9), delineating the resistivity distribution from about 1,500 Ωm to more than 10,000 Ωm. Based on the information from [1], geologically, the rock formation in the area is lava basaltic. Its high resistivity indicates the material structure is almost entirely rocks. The images of both arrangements have a similar profile in the first line ( Figure 6 and Figure 7), especially on a very high resistivity level. Meanwhile, in the second line, the product of the Wenner Schlumberger displays more contrast resistivity levels than the Wenner Alpha array because the former form has better horizontal [33] and image [32,35] resolutions than the latter.
To explain those outcomes, determine the median of the resistivity range; it attained a value of 5,000 Ωm. Assuming the value under 5,000 Ωm is the solid rock, the possibility of more than that is a highly dense rock or an air-filled cavity matched to the anomaly marked with the dashed lines in Figure  6 to Figure 9 (top image in each figure). Since in the study area have happened many subsidences, the location is similar to [12] in that there are many underground cavities. The subsidence occurs because the cave ceiling structure cannot withstand its load, causing the falling stones. This case has also happened in this study area. If so, a resistivity level over 5,000 Ωm can be defined as an air-filled cave.
To clarify this assumption, proceed to the next step, a chargeability analysis. The value distribution from both arrays ranges from nearly one to just over 18 ms (bottom images of Figure 6 to Figure 9). A significant chargeability level appoints mineral content in the rock, which can be carbonate or other mineral properties [12]; thus, a cavity filled with air should be a low-level value. Only the Wenner Schlumberger display from the first line shows a low chargeability anomaly with a value under 6 ms (dotted line in the bottom image of Figure 7), in line with a highly resistive anomaly (top image of Figure  7). There is possibly a targeted anomaly, an airfilled cavity. This result becomes a reference for a targeted cavity presence, verifying its profile has a better image resolution than the Wenner Alpha display [32,35]. In contrast, the chargeability profile in the first line of the Wenner Alpha arrangement is roughly horizontally layered (bottom image of Figure 6). The chargeability distribution gradually increases vertically from the top to the bottom layer, depicting no sign of a targeted anomaly. It seems the high chargeability level aligns with the highly resistive zone; it may be a very dense rock. This alignment is also on the right side of the Wenner Schlumberger shape (Figure 7), denoting a very compacted stone. Even though the cross-sections from both configurations ( Figure 6 and Figure 7) are not precisely similar, the electrical property distributions on the right side are relatively constant. At a glance, the profile in Figure 6 seems different from the first-line shape of Wenner Schlumberger, producing ambiguity in interpreting a targeted cavity.   On the other hand, the chargeability results for both configurations in the second line (view the bottom images of Figure 8 and Figure 9) produce almost similar profiles, notably a high value in several particular zones. Still, no low-level anomaly aligns with a very resistive zone. Instead, the high chargeability distributions align with the highly resistive zones (their anomalies are highly dense rocks). Overall, either resistivity or chargeability distributions view roughly identical profiles for both arrays (Figure 8 and Figure 9).
Then, to minimize the ambiguity, the integrated processing applies between the resistivity and chargeability products for both arrangements. By occupying equation (6), the MF distribution is computed, then the profile is generated, which can be noticed in Figure 10 (line 1) and Figure 11 (line 2). The MF cross-section of each array is presented in each figure; the top image is for the Wenner Alpha shape, and the rest is the Wenner Schlumberger form. To specify a targeted anomaly boundary, the authors decide it according to the geological condition in the field since there are no similar previous studies as references there.
According to the measurement lines ( Figure 5), the first line result should represent an air-filled cave, while the product of the second line should describe a fully dense rock. The authors define two classifications of an MF value to draw a desirable image based on the condition. First, a low-level zone, the value below 1.5 ms/Ωm as a cavity filled with air, and then, more than that, as a high-level area, which can be a solid stone or a highly dense rock. The first result can be viewed in Figure 10, the first line profile, which correctly displays the presence of an underground cavity filled with air. It fits the actual geological state that the techniques can reassure its existence according to its position in the subsurface perpendicular to the data acquisition line on the ground. Figure 11 is the second outcome, which presents the product of line 2. It shows the entire image is rock and matches the classification performed.
Although the Wenner Alpha shape of resistivity and chargeability distributions do not depict an alignment between very high levels of resistivity and a low chargeability zone (Figure 6), the MF profile draws a contrasting view between low-and high-level zones (top image of Figure 10). Accordingly, the MF cross-section of the Wenner Alpha form is similar to the Wenner Schlumberger form ( Figure 10). The difference in those chargeabilities profiles (bottom images of Figure 6 and Figure 7) causes variance computing of the MF distribution, resulting in disparate cave geometry products. The cave shape of the Wenner Schlumberger array nearly doubles from the outcome of the Wenner Alpha configuration. Even so, both delineate an anomaly of an air-filled subsurface cavity. From the results (Figure 10), the cavity ceiling is at a depth of about 2-3 m from the surface. The basement depth ranges from 6 m (Wenner Alpha) to 8 m (Wenner Schlumberger), and its width is approximately 14 m (Wenner Alpha) and 22 m (Wenner Schlumberger).
Also, the measurement line length covers the target's investigation depth, referring to equations (4) and (5) for each array. The Wenner Schlumberger array has a deeper depth of investigation than the Wenner Alpha because of the Figure 9. The profile of line 2 with Wenner Schlumberger array, a resistivity profile at the top, and a chargeability profile at the bottom. disparate depth penetration mathematically and more coverage data vertically (see Figure 4).
Nevertheless, the Wenner Alpha has slightly better sensitivity in depth in detecting a basement of high  resistivity anomaly than the former array [32]. Even though the estimated geometry from both shapes ( Figure 10) does not accurately match the actual cave shape (Figure 12), they are appropriately effective as expected. Thus, the integrated profiling method between resistivity and induced polarization is respectable in confirming the presence of an underground air-filled cave. In this study case, the Wenner Alpha form depicts a targeted anomaly much better than the Wenner Schlumberger shape.

Conclusion
An air-filled subsurface cave as a targeted zone is accurately and effectively identified by the integrated method between resistivity and chargeability profiling. The cave geometry is successfully described through a metal factor profile. The value under 1.5 ms/Ωm is classified as a targeted anomaly and, over that, is entirely solid rock. The obtained result employs the Wenner Alpha array presenting an approximate dimension of 14 m width and 4 m height, nearly close to an actual geometry (9 m × 3 m), much better than the Wenner Schlumberger configuration with 22 m width and 6 m height. For further research, a geochemistry study is required to analyze the chemical content within the stone to figure out why the ceiling structure of the cave cannot withstand its load.

Acknowledgment
This study was funded through a research grant program by the Research and Community Service Institution (LPPM) Institut Teknologi Sumatera with a contract number B/765b/IT9.C1/PT.01.03/2022. Also, it was supported by the Watershed and Protected Forest Management Agency (BPDASHL) and Giri Mulyo-East Lampung Government. The authors thank the Department of Geophysical Engineering and Geological Engineering students for assisting with the data acquisition.