Subsidence Due to Groundwater Withdrawal in Kathmandu Basin Detected by Time-series PS-InSAR Analysis

Abstract

In recent years, subsidence due to excessive groundwater withdrawal is a major problem in the Kathmandu Basin. In addition, on 25 April 2015, the basin experienced large crustal displacements caused by Mw 7.8 Gorkha earthquake. In this study, we applied StaMPS- Persistent Scatterer InSAR (StaMPS PS-InSAR) technique to estimate the spatio-temporal displacements in the basin after the mainshock. 34 Sentinel-1 C-band SAR data are used for measuring subsidence velocity during 2015-2017. We found the maximum subsidence velocity of about 9.02 cm/year and mean subsidence rate of about 8.06 cm/year in the line of sight direction, respectively, in the central part of the basin.

1. Introduction

Land subsidence due to groundwater depletion is a global phenomenon occurred due to over-exploitation of sub-surface groundwater for domestic and industrial supply. In recent years, due to the rapid development of major cities in terms of industries and population, both surface and groundwater are becoming inadequate to satisfy the industrial, domestic and irrigational water needs. In the Kathmandu basin, the capital of Nepal having dense population has undergone severe groundwater extraction in the recent years (Pandey et al., 2010). Previous studies have reported that the groundwater extraction rate is much higher than the aquifer’s recharge rate in the basin (Pandey et al., 2010). These trends may lead to the land subsidence in the sedimentary basin.

Land subsidence due to the groundwater extraction can be monitored using several techniques. Ground-based methods such as Leveling (Chen et al., 2007) and Global Positioning System (GPS) (Abidin et al., 2001) are able to measure accurate ground motions. However, Leveling requires adequate manpower and time consuming to observe displacements over sparse points. Whereas, permanent GPS stations can provide reliable vertical and horizontal displacements continuously only at the station for a period of time. In recent years, space-borne earth observations using interferometric synthetic aperture radar (InSAR) technique is applied to measure the ground motions caused by various geohazards (Tomas and Li, 2017). Previous studies (Pandey et al., 2012; Pandey and Kazama, 2011) have monitored the land subsidence caused by over-exploitation of groundwater. In this study, we applied time-series persistent scatterer interferometric analysis (PS-InSAR) to monitor the land subsidence in the Kathmandu basin, Nepal.

2. Study Area

The Kathmandu basin located in the lesser Himalayas is a densely populated city in Nepal (Fig. 1). The basin is mainly composed of fluvio-lacustrine and fluvio-deltaic sediments, which consists of peat, clay, carbonaceous clay, sand, gravel and boulders (Moribayashi and Maruo, 1980). Based on the gravimetric survey carried out by (Cresswell et al., 2001; Moribayashi and Maruo, 1980), the thickness of fluvio-deltaic sediments is about 650 m (Fig. 1). The basin is divided into the following lithostratigraphic units (Fig. 1): The Lukundol and Itachi in the south, the Gokarna, and Thimi in the north and Bagmati, Kalimati and Patan in the center. There are two active faults are located around the Kathmandu valley. Considering the hydrogeological conditions in the basin, groundwater for domestic and industrial use is mainly extracted from shallow and deep aquifers separated by a thick clay layer (aquitard) (Fig. 1).

Fig. 1. Study area showing the location and the boundary of the Kathmandu Basin, Nepal geological formations, monitoring wells, and GPS stations.

3. Data used and InSAR processing

Table 1. List of SAR data used in this study with baseline information corresponding to the single master scene (2017/07/11)

Table 2. Description about SAR data used in this study

We applied StaMPS PS-InSAR (Hooper et al., 2004) technique in this study to estimate the time-series displacements in the Kathmandu basin. In this processing, the master scene is chosen based on minimum geometric and temporal decorrelation. Multi-looked (10 × 2) differential interferograms stack were generated with the single master scene (2017/07/11). The perpendicular baseline (Bperp) and temporal baseline (Btemp) information for each InSAR pair can be seen in Fig. 2. All the interferogram pairs are having perpendicular baseline less than 150 m. In order to remove topographic phase contribution in the interferometric phase, 30 meter SRTM digital elevation model is used in this processing. Since most of the parts in the study area consists of urban features, the coherence was maintained very well. However, the coherence has been greatly reduced in the mountainous regions. Subsequently, the time-series InSAR processing was performed by applying StaMPS algorithm (Hooper et al., 2004) as shown in Fig. 3.

Fig. 2. Plot showing perpendicular baseline (m) and temporal baseline (days) with respect to the master scene (2017/07/11).

Fig. 3. StaMPS PS-InSAR Processing flow used in this study (Hooper et al., 2004).

4. Results and Discussion

The cumulative surface displacement map is presented in Fig. 4, spanning from the year 2015 to 2017. The central part of the Kathmandu basin is found to be subsided significantly in the LOS direction i.e., Location A, B in Fig. 4. The obtained PS-InSAR results in Fig. 4 and 5 are with respect to the reference point (blue square), selected close to the KKN4 GPS station, which is located away from urban part of Kathmandu basin. Because KKN4 GPS station is appears to be nearly stable compared to other part of the basin. Therefore, the mean velocity at the reference point will be zero. Apart from KKN4 station, NAST and KIRT GPS stations are affected by severe ground motions and discontinuous in the observation. Therefore, cumulative displacement, subsidence rate in the LOS direction are obtained with respect to the KKN4 GPS station. The maximum line of sight displacement of -18 cm is observed in the central basin (Fig. 4). In addition, the southern part of the basin has been identified as subsiding for 12 cm. The negative sign indicates the line of sight movement away from the satellite. However, apart from locations A and B, the entire basin is also experiencing minor deformation s shown in Fig. 4.

Fig. 4. Cumulative surface LOS displacement map (2015/08 - 2017/12).

Fig. 5. Mean LOS Displacement rate map (cm/year) from (2015/08 ~ 2017/12).

Similarly, the mean subsidence velocity observed from the year 2015 to 2017 using StaMPS PS-InSAR processing is shown in Fig. 5. The maximum LOS subsidence rate of about 9.5 cm/year is observed in region A, whereas approximately 7 cm/year LOS subsidence rate is identified in the southern part of the Kathmandu city (region B) (Fig. 5). Both the subsidence regions are highly correlated with the clay deposition (fluvio-lacustrine) in the basin. Moreover, the deep aquifer with a maximum thickness of 200 m, located toward the central and southern part of the basin, is mainly composed of thick compressible layers.

According to the previous studies, the groundwater extraction in these regions has already exceeded the recharge rate (Gautam and N Prajapati, 2014). The subsidence pattern in these regions is centralized in smaller regions, which suggests the over-exploitation of groundwater by anthropogenic activities. The groundwater table level in many parts of the basin is observed to continuously lowered in the past decade (Pandey et al., 2012). Therefore, we consider that the groundwater extractions in the thick compressible layers are the possible cause for the subsidence in the Kathmandu basin.

It is worth noting that apart from significant subsidence in the central basin, we have observed minor subsidence about 1.5 cm/year with reference to KKN4 GPS station over the whole basin. We anticipate that this regional deformation is possibly due to the post-seismic relaxation following the large co-seismic crustal motion about 1.2 meters (uplift) (Elliott et al., 2016).

The average time-series displacement observed for the region A is illustrated in Fig. 6. The mean subsidence rate in the region A is estimated as 8.06 cm/year along the LOS direction. It is very important to compare the InSAR results with groundwater level data. However, due to the lack of detailed hydrological information such as water-table data and pumping rate during the study period, a comparison cannot be performed. Despite the limitation, with available information on groundwater trend and geological information, we anticipate that the subsidence in the Kathmandu basin is possibly due to groundwater withdrawal from the clay deposits in the basin.

Fig. 6. Time-series Line of Sight displacements in the region A.

5. Conclusion

In this study, we applied StaMPS PS-InSAR technique to monitor the land subsidence in the Kathmandu basin. The time-series analysis revealed Korean Journal of Remote Sensing, Vol.34, No.4, 2018 the subsidence in the central part of the basin with mean subsidence rate about 8.06 cm/year in the region A. The maximum subsidence rate about 9.5 cm/year is observed in region A with reference to KKN4 GPS station. The subsidence pattern is highly correlated with the local geological formation, which is mainly the lacustrine clay deposits. However, the major limitation of this study is the absence of groundwater level in subsiding regions.