Mapping Underground River Flows in karst Areas with the VLF-EM Method (Case Study of the Krawak Region, Singgahan Tuban)

The Krawak Springs is a source of springs from the cracks of river rocks in the middle of a teak forest in Guwoterus Village, Montong District, Tuban. This spring is a source of irrigation water for agricultural land in the District of Montong. Based on its urgency, it is necessary maintaining its sustainability by identifying recharge areas and underground river flows. VLF is quite advantageous for use in the regional mapping underground rivers in karst areas because can describe the structure of the karst environment and predict interconnected pathways in karst. Data acquisition uses four tracks with an average length of about 300 m. Data processing was made qualitatively using MATLAB software and quantitatively using Inv2DVLF software. Qualitative results show that the conductive zone is identified as a subsurface river flow. From the processing results, it is known that the subsurface river flow zone can be identified as a zone that is conductive or has low resistivity. The underground river flow is interpreted to be located on Track 3 and Track 4 and has a direction of continuity to the southwest


Introduction
The Tuban Regency comprises limestone rocks and young reefs rich in karst springs [1].One of the karst springs in Tuban is Krawak Springs, located in Guwoterus Village, Montong District, Tuban Regency, East Java.The Krawak Spring is a spring that comes out of a gap.Krawak is in the middle of the teak forest in Guwoterus Village, District Montong, Tuban.The Krawak Spring is a source of irrigation water for land agriculture in the Montong District area, Tuban.In addition, Krawak is a source spring water from the natural tourist attraction Ngirip Waterfall.Based on this, it is known that this spring has an impact on many sectors, so it is necessary to maintain the sustainability of the ecosystem.However, complex problems often occur in karst areas, generally related to clean water availability.Karst hydrogeological systems generally have unique characteristics.Karst landforms have secondary porosity caused by water dissolution.The presence of karst causes the formation of subsurface water flows so that the karst hydrology on the surface is less developed and less dominant [2].This is because the cavities in karst areas will cause water to enter underground, accumulate in baseflow, and form underground rivers.As a result of this process, the surface in the karst area will look dry because water tends to be underground compared to the surface [3].So, in this case, a geophysical method is needed to determine the recharge area and underground river flow.The VLF-EM method is a geophysical method for measuring rock conductivity by identifying the characteristics of secondary electromagnetic waves.These secondary waves are produced from the induction of primary electromagnetic waves emitted from radio transmitters with very low frequencies, namely 10-30 KHz [3].This method is commonly used in underground river exploration by exploiting the conductive properties of water and can cover steep terrain and broad regional conditions in karst areas [4].The previous researchers (Iswahyudi et al., 2023) used the VLF-EM method to search for groundwater river patterns in karst areas.Based on the research results, the direction of the flow pattern and the depth of the groundwater river can be determined from the distribution of resistivity values resulting from VLF-EM data processing [5].This research uses the VLF-EM method to determine the possibility of subsurface rivers in karst areas in the Montong District.

VLF-EM
The Very Low Frequency (VLF) method is a geophysical method with a passive measurement technique that utilizes the magnetic component of the electromagnetic field generated by a radio wave transmitter with a shallow frequency, namely around 15-30 kHz [6].In general, this method is used to explore conductive objects.This is because the VLF method utilizes changes in field components due to conductive variations, which are then used to determine subsurface structures.The VLF-EM method measures the results of horizontal and vertical local magnetic field components with two orthogonal induction coils.The local magnetic field HR result is the superposition of the primary field HP and the secondary field HS, where HP>HS.In underground conductors, the total VLF field is elliptically polarized.Furthermore, the results of VLF-EM are the in-phase (natural) and quadrature (imaginary) parts of the ratio (HRz/HRy).The natural and imaginary components are expressed as a percentage of the main total field of the VLF transmitter.The natural part of the tipper is sensitive to objects with low resistivity.In contrast, the quadrature part of the tipper is sensitive to variations in the electrical properties of the earth [7].The wave propagation often found in the survey area is sky waves [9].Sky waves are waves radiated by an antenna into the ionosphere layer in the upper atmosphere and deflected back to the earth.Direct

Data Processing
In this process, there are two methods, namely qualitative and quantitative analysis.Qualitative analysis uses Fraser and Karous-Hjelt filters, while quantitative analysis uses inversion.In qualitative analysis, the data will be processed with NA-MEMD filtering using MATLAB software to remove noise in the data.After that, it continues with the Fraser and Karous-Hjelt filter, which has a discrete filtering principle which provides apparent current density at different depths so that it will cause the magnetic field to be the same as the measurement results.The process has output in the form of in-phase and quadrature intersection graphs as well as 2D cross-sections of equivalent current density [10].Meanwhile, in quantitative analysis, the inversion process uses Inv2DVLF software to obtain a quantitative cross-sectional model of the subsurface layer.The results of the inversion process were modeled using Surfer with output in the form of 2D subsurface resistivity [11].In VLF-EM data processing, filtering needs to be done to produce data with a trim noise level.One of the commonly used filtering methods is NA-MEMD [12].In figure 4, the results of the filtering process provide a significant difference between the data before and after processing.Data will have good processing results if the filtered data still follows the initial data graph.Qualitative processing is done by filtering the data using NA-MEMD.After the NA-MEMD is done, the following filtering process is using the Fraser filter to focus the scattered anomalies into smaller and IOP Publishing doi:10.1088/1755-1315/1307/1/0120057 more detailed ones to produce an intersection graph between in-phase and quadrature to determine conductive and resistive anomalies and the Karous-Hjelt filter to describe in 2D the current density of the measurement area at a certain depth.The third track is the result of the Fraser and Karous-Hjelt filters.From the Fraser filter graph between in-phase and quadrature, if the quadrature value is in the minimum position while the in-phase value is in the maximum position, it can be concluded that there is a resistive or conductive anomaly.Meanwhile, the function of the Karous-Hjelt filter is to show the current density value as a function of depth and measurement distance.In this process, high current density values are coloured orange/red, while low current density values are blue.In determining the depth of the anomaly, this filter uses pseudosection identification of current density anomalies at horizontal locations with variations in current density at the same depth as the measurement point.So, on the third track, the conductive zone indicated as subsurface river flow (marked with a red box) is at a depth of around 1 to 38 meters at a distance of 80 to 120 meters.

Quantitative analysis
Quantitative VLF data processing uses the Inv2DVLF and MATLAB programs.The data used in the inversion process has gone through the NA-MEMD filter process.In the inversion process, 100 iterations were used with an initial resistivity value of 750 Ωm because the rock layer measurement area is dominated by karst or limestone rocks (500-10 7 ) [13].Apart from that, this processing uses the Lagrange parameter with a value of 0.03, widely used during the inversion process [14].A 2D model will be made after the inversion process is complete and can be done using Surfer software or MATLAB.The overlay results carried out horizontally using Google Earth or SketchUp in Figures 11, figures 12, and figure 13 show the same results.This research aims to determine the underground river flow zone (high conductive or resistive zone).The possibility of underground river flow is currently interpreted as being in the third and fourth paths because the conductive values between the two paths are continuous (marked with red arrows).The estimated direction of the subsurface river is from northeast to southwest.Meanwhile, in Figure 12, there is also a red circle, which is interpreted as a conductive anomaly but is not thought to be an underground river flow because, in that area, there is no water flow at all, while the resistive anomaly resulting from processing is of low value.

Conclusion
From the results of VLF-EM data processing using quantitative and qualitative analysis, it can be seen that underground river flows can be identified through conductive zones with low resistivity values.From the results of the interpretation of the underground river flow pattern, it is located on the third and fourth tracks from northeast to southwest.

Figure 1 .
Figure 1.Various types of waves emitted by the transmitting antenna and waves received by the receiver [8].
waves are signals emitted by a transmitting antenna that are received directly by the receiver without experiencing reflection.In contrast, reflected waves are reflected off the surface and received by the receiver.Meanwhile, trapped waves are waves that penetrate subsurface rock layers, which provide information on the propagation time of the wave until the receiver receives it.1307 (2024) 012005 IOP Publishing doi:10.1088/1755-1315/1307/1/0120053 3. Methodology3.1.Data acquisistionThe VLF-EM measurements were located in the Krawak Springs area and its surroundings in Guwoterus Village, Montong District, Tuban Regency, East Java.In this research, there are a total of four tracks, each track has a length of 250-300 meters.On each track, data will be collected at intervals of 3 meters.This measurement uses three different frequencies, including 19.8 KHz, 23 KHz and 24.1 KHz.The data measured in VLF-EM acquisition are inphase, quadrature, tilt, and totalfield.

Figure 5 .Figure 6 .Figure 7 .
Results of the Fraser and Karous-Hjelt filter at a frequency of 19.8 KHz in (a) first track, (b) second track, (c) third track, (d) fourth track Results of the Fraser and Karous-Hjelt filter at a frequency of 23 KHz in (a) first track, (b) second track, (c) third track, (d) fourth track Results of the Fraser and Karous-Hjelt filter at a frequency of 24,1 KHz in (a) first track, (b) second track, (c) third track, (d) fourth track

Figure 8 .Figure 9 .Figure 10 .
Figure 8. Results of the inversion model at a frequency of 19.8 KHz in (a) first track, (b) second track, (c) third track, (d) fourth track

Figure 11 .Figure 12 .Figure 13 .
Figure 11.Horizontal overlay using Google Earth on the first track and second track