PSI and GNSS derived ground subsidence detection in the UNESCO Heritage City of Ahmedabad, Western India

Abstract Presently, ground subsidence in urban areas is becoming a peril to the infrastructure and society. In this study, we conducted a detailed analysis of the results, based on the Persistent Scatterer Interferometry (PSI), Global Navigation Satellite System (GNSS) and groundwater in the ‘World Heritage City’ of Ahmedabad, a major metro city in western India. The PSI technique is applied to measure the ground subsidence using the Sentinel 1 A dataset for the period 2017 to 2020. The results reveal a subsidence, maximum up to the level of 25 mm/yr in the areas of the southeast and west parts of the city. Analysis derived from GNSS results indicates a significant amount of ground subsidence based on the 2009–2020 dataset. Groundwater data (1995–2019) were analyzed to identify factors likely to cause ground subsidence and the results are in close agreement with the PSI and GNSS results.


Introduction
Groundwater depletion is becoming a global concern due to the excessive use of groundwater resources. Population growth, rapid urbanization, and industrial expansion are the main reasons for the increasing demand for water extraction per capita (Godfray et al. 2010;Klein Goldewijk et al. 2010;Wada et al. 2014;Hanasaki et al. 2018).The use of water for domestic and agricultural purposes has increased in recent decades, which ultimately leads to the over-use of groundwater. Due to water extraction, land subsidence is becoming a major threat to some cities (Hung et al. 2012;Modoni et al. 2013;Pacheco-Mart ınez et al. 2015). The spontaneous water removal leads to vertical compaction of aquifer sediments by reducing pore pressure and soil compaction resulting in land subsidence. The land subsidence promotes micro-level topographic changes, cracks and fissures in the surface and ultimately causing heavy damage to the infrastructures and becomes hazardous to the civilization. Ancient hidden water ways were also identified as a major factor for urban land subsidence (Krassakis et al. 2021). Therefore, monitoring of ground subsidence is an important challenge for city planners. The monitoring of land subsidence is primarily carried out by techniques, i.e. levelling surveys, borehole extensometers; but due to the technological advancement and for better precision now day's techniques such as Interferometric Synthetic Aperture Radar (InSAR) and Global Navigation Satellite System (GNSS) are being applied (Wright et al. 2004;Karila et al. 2013;Bon ı et al. 2015;Zhang et al. 2015;Eriksen et al. 2017;Li et al. 2017;Zhu et al. 2017;Choudhury et al. 2018;Riel et al. 2018;Liu et al. 2019;Rateb and Kuo 2019;Tangdamrongsub et al. 2019;Yang et al. 2019). The GNSS provides precise position of a preferred point on the earth's surface using the GNSS satellites and technique is useful to measure the deformation with mm level of accuracy. However, the InSAR technique measures the ground deformation with an accuracy of sub cm level and covers a large surface area.
To estimate the spatio-temporal changes over a large area the Persistent Scatterers Interferometry (PSI) come in to existence and the main output of the PSI technique is the line of sight (LOS) displacement for each persistent scatterers (PS) (Ferretti et al. 2000(Ferretti et al. , 2001. The PSI technique has the edge over the typical InSAR studies by having the corrections for orbital, atmospheric and topographic errors to precisely achieve an accuracy upto the mm level for a Persistent Scatterers (PS) on the desired area (Ferretti et al. 2001(Ferretti et al. , 2007Werner et al. 2003;Crosetto et al. 2008Crosetto et al. , 2016. PSI technique has shown remarkable results in measuring crustal deformation even in the low strain provinces (B€ urgmann et al. 2006;Massironi et al. 2009;Vilardo et al. 2009;B ejar-Pizarro et al. 2010;Bell et al. 2011;Grandin et al. 2012;Peyret et al. 2013;Dumka et al. 2020Dumka et al. , 2021. The InSAR technique is useful in the studies like geomorphology, hydrology, and landslides, etc. and successfully detect even a small ground deformation (Alsdorf et al. 2000;Smith et al. 2002;Hooper et al. 2012;Perissin et al. 2012;Wasowski and Bovenga 2014;Tessari et al. 2017;Zhou et al. 2017;Floris et al. 2019). Several aspects of the geosciences can be thoroughly investigated using this technique (Ouchi 2013;Crosetto et al. 2016;Biggs and Wright 2020). Road/railway, bridge network loops are detected very commonly using this technique (Perissin et al. 2012;Shafieardekani and Hatami 2013). The space based InSAR technique provides precise measurements in ground deformation with a wide surface coverage (Ferretti et al. 2001;Psimoulis et al. 2007) and it has become a current tool for the research related to the tectonics and earthquakes (Gens and Van Genderen 1996;Klees and Massonnet 1998;Atzori et al. 2009;Merryman Boncori et al. 2015;Fiaschi et al. 2017;Takada et al. 2018;Bacques et al. 2020).
In Indian cities, the land subsidence is very common and it is believed that the extraction of groundwater for drinking is to be a major factor in the subsidence as about 80 to 85% of drinking water in the cities of India comes from groundwater (IE 2018;TI 2018). The PSI based study of the National Capital Region (NCR) of Delhi indicates subsidence due to the groundwater extraction (Malik et al. 2019;Garg et al. 2020). Likewise, the city of Kolkata has also observed ground subsidence due to the overuse of the groundwater (Chatterjee et al. 2006;Bhattacharya et al. 2011;Sahu and Sikdar 2011;Suganthi and Elango 2020). Annual Groundwater level reductions of 7-10 feet were observed in the southern Indian city of Chennai (Raju et al. 2008). A study based on GPS data from 2009 to 2015 indicates land subsidence due to groundwater extraction in Gandhinagar city, the capital of Gujarat in western India (Choudhury et al. 2018).
A survey by the Central Groundwater Board (CGWB) found that Gujarat had 59% of wells showing a decrease in groundwater level between 2007 and 2016 (Tiwari 2016;www.cgwb.gov.in). Likewise, a study from Ahmedabad district classified the northeastern part of the district, including Ahmedabad and Dascroi Taluka, as overexploited based on groundwater status (www.cgwb.gov.in). So, in this study, we used an advanced remote sensing technique, PSI, to detect the subsidence of the city of Ahmedabad, which is considered to be the commercial capital of the state of Gujarat in the western India. The ground subsidence is estimated using the 71 images of Sentinel 1 A dataset acquired in descending passes from 16 October, 2017 to 30 September, 2020 by applying the PSI technique. Similarly, the GNSS dataset of an urban site as well as three nearby locations in the Cambay basin are analyzed to estimate the vertical displacement. In addition, the groundwater dataset of eleven wells (CGWB), from 1995 to 2019, is used to calculate ground subsidence due to the water extraction. Finally, a combined analysis is performed to locate the areas of maximum land subsidence in the city of Ahmedabad.

Study area
The present study area includes a large city (Ahmedabad), located between latitude 22 54 0 0 00 N to 23 8 0 0 00 N and longitude 72 23 0 0 00 E to 72 42 0 0 00 E, in western India, also known as the industrial capital of the state of Gujarat (Figure 1). Based on the historical significance of the city, it became the first Indian city to be designated a 'UNESCO World Heritage City'. According to the 2011 census, the population of Ahmedabad is 5.6 million (censusindia.gov.in). The city of Ahmedabad is located in the Cambay basin, in the mainland region of Gujarat. Geologically, the Tertiary sediments around the study area are overlain by a thick pile of quaternary sediments (Pathak and Balsubramanayan 1971; Kaila et al. 1990;Kumar et al. 1998;Dwivedi et al. 2017Dwivedi et al. , 2020. The alternate layers of sand, silt gravel and inter-mixed granular materials and sticky yellow to grey clay form most of the quaternary sediment (Gupte 2011). The city is divided into east and west by the Sabarmati River and most of the study area lies under the floodplains of the Sabarmati River and the sediments mostly comprises silt and sand exposed on the surface (Goel 2001;Dwivedi et al. 2020). Some sand dunes as well as residual highlands and some terraces have also been reported in the study area (Dwivedi et al. 2020). The geotechnical investigation reveals a low water table and high compaction in the city of Ahmedabad (Dwivedi et al. 2017). Groundwater mapping based on the well-log data reveals the availability of groundwater in the form of unconfined and confined aquifers and the maximum depth of mapped confined aquifer is up to 300 m in the city (Gupte 2011) ( Figure 2).

Persistent scatterer interferometry (PSI) observations
Persistent Scatterer Interferometry (PSI) is an advanced radar-based remote sensing method of InSAR technique and applied for the periodic measurements of the ground deformation (Gens and Van Genderen 1996;Smith et al. 2002;Hooper et al. 2012;Perissin et al. 2012;Canaslan et al. 2016aCanaslan et al. , 2016bBayer et al. 2017). Similarly, in the present study, we have applied the technique for the measurements of ground subsidence in the city of Ahmedabad, because it provides spatial and temporal measurements with an accuracy of mm level for the chosen points (PSs) and the method relies on the measurements of surface deformation using several stable PSs (Persistent Scatterers) over the desired area (B€ urgmann et al. 2000;Ferretti et al. 2001;Psimoulis et al. 2007;Perrone et al. 2013;Van Leijen 2014;Crosetto et al. 2016;Perissin 2016;Tosi et al. 2016;Biswas et al. 2017;Foumelis et al. 2018;Delgado Blasco et al. 2019;Dumka et al. 2020Dumka et al. , 2021. The normal temporal coherence along with the stable scattering is the must-have properties of PSs (Perrone et al. 2013). The availability of the Sentinel data by the European Space Agency (ESA) (https://sentinel.esa.int/; http://scihub.copernicus.eu) enhanced the use of this technique for various applications.
To measure the surface subsidence in the present study, we applied SNAP S1-toolbox collectively with Stanford Method for Persistent scatterers (Hooper et al. 2012;Veci et al. 2014;Delgado Blasco et al. 2019) Sentinel Application Platform (ESA-SANP). The dataset is Sentinel-1A, descending (Table S1 and Figure S1), as it is the latest set of radar images using a C-band sensor having 12 days cycle (asf.alaska.edu). The frame 514 and 517 of path 34 has been selected to download dataset for the study area ( Figure 3). For the present study area only descending dataset is available so we have not used the ascending dataset. The minimum revisits time of the Sentinel 1 data set provides control over the surface deformation monitoring. The processing is adopted as prescribed in Foumelis et al. (2018). Here, we would like to summarize the steps ( Figure 4) taken in the present study: 1. TOPSAR-split and orbit updates: In the initial processing the orbital information has been updated by splitting of Sentinel-1 IW SLC core products. The splitting of TOPS master and slave scene is carried out after defining the sub-swath, polarization and burst for the study area. The common burst between master and slave has been considered during further processing (Foumelis et al. 2018). The precise orbital information downloaded in the SNAP has been updated. 2. Co-registration and Interferogram generation: The geometric co-registration (Backgeocoding) and Enhanced Spectral Diversity (ESD) refinement are included in this step. The inverse least square method is applicable that connects the master image to the slave image. The step of Interferogram generation starts based on the phase difference between master and slave images. The Shuttle Radar Topography Mission (SRTM) elevation model at 1 arc second resolution, acquired by NASA in the year 2000 (Farr et al. 2007), is used in the study for the modelling and removal of the topographic phase before the generation of Interferograms. To obtain spatially continuous images deburring of SLCs and differential interferograms is required (Foumelis et al. 2018).  3. SNAP to StaMPS: The generated differential interferograms along with the co-registered master-slave SLCs have been exported to StaMPS format from the SNAP. 4. StaMPS processing: The processing here starts with the selection of PS pixel candidates after looking at the amplitude dispersion index (DA) typically 0.4-0.42. The lower the value of DA, the higher is the amplitude stability of the pixel. The estimation of phase noise value for each pixel candidate in all the interferograms is used to determine the stability of PS candidate. The noise characteristic is looked at for the selection of pixels. The PS pixel candidates were corrected for spatially uncorrected look angle error to eliminate the phase due to deformation. The wrapped phase is converted into an unwrapped phase by applying a 3D unwrapping method (Chen and Zebker 2000) on the selected PS pixels. The spatial look angle error (SCLA) with master atmosphere error (Bekaert et al. 2015) and orbit error is subtracted from the unwrapped phase ( Figure S2).

GNSS observations
To observe vertical displacement in the region, data from four Continuously Operating Reference Sites (CORS), GNSS sites, was processed. Among them, a GNSS site (AMBD) is located to the southeastern part of Ahmedabad city and another GNSS site (GNDC) is north of Ahmedabad city. Two other GNSS sites (SIPU and SATL) located in the Cambay Rift Basin (CRB) are also taken to observe the vertical deformation in the basin (see Figure 1). The GPS data of site AMDB started in the year 2017 and site GNDC started in the year 2009 (Table 1). The GNSS data of the sites, having a sampling interval of 30 seconds, were analyzed from 2009 to 2020. The TEQC (Translation, Editing and Quality Check) program by UNAVCO (University NAVSTAR Consortium) (Estey and Meertens 1999) was utilized for the check of high cycle slip of data, poor multipath data and to minimize the shorter length data during the pre-processing of the data. The GAMIT/GLOBK 10.7 software (King and Bock 1998;Herring et al. 2010) is applied for the post-processing of data. To generate the results in the International Terrestrial Reference Frame (ITRF) (http://itrf. ensg.ign) 21 IGS sites (Table S2) located globally were used during the processing (Beutler et al. 1994;King and Bock 1998;Herring et al. 2010;Altamimi et al. 2016) for the calculation of Indian plate motion in ITRF14. To minimize the effect of ocean tides and solid earth tidal displacement, FES2004 and IERS model have been applied, while the GPT2 model is applied for the meteorological corrections (Letellier et al. 2004;McCarthy and Petit 2004;Lagler et al. 2013;Carr ere et al. 2016;Lyard et al. 2021). After processing the velocity of sites are estimated in ITRF14 reference frame in WGS84 datum (Table 1).

Water level observations
Global research using the SAR technology, in recent years, has highlighted the role of groundwater in urban subsidence (Zhou et al. 2017;Aimaiti et al. 2018;Aslan et al. 2018;  Bui et al. 2020;Khorrami et al. 2020;Suganthi and Elango 2020). So, to determine the role of groundwater depletion on land subsidence, we obtained water level data from 11 water wells available for the Ahmedabad (www.cgwb.gov.in). These wells have data observations of various time-periods from 1996 to 2020 (Table 2). In addition, we used the water level data of the city provided by Gupte (2011), which refers to the decrease in groundwater levels between 1960 and 1995 ( Figure  5). The variation in the groundwater level is measured by applying relative measurements in the observation wells (www.cgwb.gov.in).

Results and discussion
The results derived after analyzing the PSI dataset showed the existence of displacement (LOS) in Ahmedabad ( Figure 6). Two major patches of negative displacement were identified in the PSI results, located to the west and southeast of the heritage city. Negative displacement represents the ground subsidence in the study area. Maximum ground  subsidence (LOS) of 20-25 mm/yr was observed near the Ghodasar, Vatva and Hathijan areas in the southeast of the city. However, the second patch near the Ghuma and Bopal area, to the west of the city, indicates an annual subsidence of 15-22 mm. Furthermore, the close observation of the results revealed some additional patches of negative displacement/subsidence, with a rate of 2-8 mm/yr, in the central-west and central-east of the city. The magnitude of ground subsidence in the central area is less than in the areas with large subsidence. Based on the ground subsidence, we categorized the whole city into three zones; high, moderate and low subsidence zone. The high subsidence zone comprises the southeast and western parts of the city. The moderate subsidence zone comprises east-central, while the low subsidence zone comprises the central part of the city. The standard deviation in the processing up to the level of 2.5 mm/yr for the study area ascertains the analysis ( Figure 6). After looking into the high amount of displacement in several areas of the heritage city we decided to find out the exact reason for the subsidence. There are two main possibilities believed to be responsible for the subsidence, one is tectonic and other is the non-tectonic. Tectonically, the area is in the Cambay Rift Basin (Figure 7) and the seismological data-based BIS (Bureau of Indian Standards) map kept Ahmedabad into zone 3, which is considered as a zone of moderate risk (www.bis. gov.in). The Global Seismic Hazard Assessment Program (GSHAP) (Zhang et al. 1999) classifies Ahmedabad city as a low-hazard area (www.asc-india.org). The earthquake records based on the local network of the last 2 decades indicate insignificant seismicity near the study area (Chopra et al. 2008). Further, the GPS-derived results in this part of the Indian plate reveal a very low amount of deformation (Dumka and Rastogi 2013;Jade et al. 2014Jade et al. , 2017Gahalaut et al. 2019;Dumka et al. 2019aDumka et al. , 2019bDumka et al. , 2020Dumka et al. , 2021. Moreover, to access the tectonic activity in the area, we have processed the GNSS dataset of almost a decade (2009-2019) for this part. The results reveal an average plate motion of 49.2 ± 1.0 mm annually for this part and after removing the plate motion using the Euler pole for the Indian plate (Jade et al. 2017), the average annual horizontal deformation recorded 0.77 ± 0.4 mm for this region (Figure 7). That is understandably low in the Intra-plate part as compared to the plate boundary region (Bettinelli et al. 2006;Ader et al. 2012;Mahesh et al. 2012;Jade et al. 2014Jade et al. , 2017Dumka et al. 2014aDumka et al. , 2014bDumka et al. , 2019aDumka et al. , 2019bStevens and Avouac 2015). The GPS site in Ahmedabad (AMBD) reveals   4.0 ± 1.0 mm from 2009 to 2019. Comparatively, two other sites (SIPU & SATL) in the same CRB showed no subsidence (Figure 8). Further, the sites (IISC, HYDE) in the tectonically stable part of the Indian plate (Jade 2004;Bettinelli et al. 2006;Ader et al. 2012;Mahesh et al. 2012;Jade et al. 2014Jade et al. , 2017Jade et al. , 2020 in southern India do not vertical displacement. Recent past geodetic studies indicate that the interior of the Indian plate exhibits very less ( 1 mm/yr) internal deformation (Bilham et al. 1998;Paul et al. 2001;Jade 2004;Bettinelli et al. 2006;Ader et al. 2012;Mahesh et al. 2012;Jade et al. 2014Jade et al. , 2017Jade et al. , 2020Dumka et al. 2014aDumka et al. , 2014bDumka et al. , 2019aDumka et al. , 2019bStevens and Avouac 2015) and is considered to be the tectonically stable part of the Indian plate. A study by Jade (2004) showed that the motion of IISC GPS sites is equal to that of the Indian plate and considers the peninsular India as part of the rigid plate.
Hence, the negative vertical displacement as indicated by the sites (AMBD, GNDC) indicates a significant amount of subsidence in and around the city of Ahmedabad. The ground subsidence due to groundwater depletion is already reported in the Gandhinagar area (Choudhury et al. 2018). Therefore, on the basis of GNSS, seismicity, BIS observation, it can be established that the tectonic activity in this part is very low and therefore vertical displacement arising in the city may not have the major tectonic connection.
To find the correlation between the ground subsidence and groundwater depletion, we then analyzed well water level data available through CGWB (www.cgwb.gov.in) ( Table  2). The water level data from wells around the city mainly indicates a drop in the water table based on the dataset from 1994 to 2019 (Figure 9). Maximum annual drops in groundwater level of about 1.6 m and 1.3 m have been reported by Sola and the Shilaj well located west of the city, respectively. The Vatva well, east of the city, recorded an annual declining of 1.04 m. The remaining wells in the city recorded an average decline of 0.24 m/yr (see Figure 9), (Table 2). Besides, a study based on water level data from 1960 to 1995 reveals an average decline rate of 2 m annually (Gupte 2011) (see Figure 4). An analysis of groundwater depletion carried out between 2005 and 2014, observed a drop in water level of about 2 m in Ahmedabad district (Verma 2014). Hydrological observation wells indicate 0.6 to 2.7 m of annual decline in the water level (CGWB 1990(CGWB , 2000(CGWB , 2007. Hence, based on the dataset of almost 50 years it can be summarized that the groundwater level is continuously decreasing in the city of Ahmedabad. We therefore deduce that groundwater decline may be responsible for the ground subsidence in Ahmedabad and this is also confirmed by PSI and GNSS observations of the present study. To see the direct relationship between groundwater depletion and ground subsidence, we estimated the rate of subsidence due to the falling water level. To estimate the annual rate of ground subsidence due to the groundwater decline we adopted the method as described by (Wang 2000;Choudhury et al. 2018;Fernandez et al. 2018) due to the change in pore pressure by water depletion.
Where, x, y, z and x 1 , y 2 , z 3 ¼ coordinates of observation and source; u z ¼ surface subsidence; Δp ¼ pore pressure change of aquifer; , u ¼ drained and undrained Poisson's ratios; B ¼ pore pressure coefficient; l ¼ shear modulus (Geertsma 1973;Segall 1985;Du and Olson 2001;Chen 2011;Fernandez et al. 2018). After looking at the insignificant horizontal motion of GPS sites (GNDC and AMBD), we have calculated vertical displacement at each well site based on the depletion rate of the respective well. The maximum water table depth of around 300 m is reported in the Ahmedabad region (Gupte 2011), therefore, in the present study an average depth of 150 m has been considered for the calculation. The parameters including B and l were considered same as calculated for the Gandhinagar region (Choudhury et al. 2018).
The estimates after the calculation reveal variation in the ground subsidence rate in the study area. Maximum ground subsidence of 9.8 mm/yr is estimated by the Sola well, followed by a subsidence rate of 8.2 mm/yr at Shilaj. The Vatva well indicates a subsidence rate of 8.1 mm/yr (Table 2). Significant subsidence of up to the level of 3.0 mm/yr is also estimated in the middle part. The water wells of Bopal, Arbudnagar, Hazipur garden and Ghuma indicate a subsidence of about 2.0 mm/yr. The prevailing subsidence trend due to the groundwater removal almost matches the PSI-derived LOS displacement trend for the city (Figure 10). The displacement generated by PSI is estimated to be the maximum near the locations of Vatva (southwest) and Bopal and Ghuma (west) of the city. Similarly, the subsidence estimated by groundwater depletion is also recorded maximum in similar locations in the city. For better understanding of PSI derived displacement and groundwater depletion derived subsidence rate, a mean of around 50 PS points at each water well location has been taken for correlation (Figure 10b). The coefficient of determination (r 2 ) value of 0.66 indicates close correlation among the PSI displacement and groundwater depletion derived subsidence for the city.
In order to compare the displacement of PSI and GNSS, we generated an average time-series of PS points within a radius of 100 m from GNSS site, and found a good match of subsidence trend ( Figure 11). Although, we believe that the difference in estimates is mainly due to the comparison of LOS (PSI) and vertical (GNSS) displacement. Since, we analyzed the descending Sentinel dataset, because the study area does not have an ascending dataset, we cannot convert the LOS displacement to vertical displacement. Therefore, after considering the corresponding subsidence trend derived via PSI, GNSS and ground water depletion in the city we can conclude that groundwater depletion is the main cause of ground subsidence in Ahmedabad. There may be some reasons for the Figure 10. (a) The distribution of the subsidence in the city as derived by PSI (background map) is almost matching with the subsidence due to groundwater depletion (contour lines) in the city of Ahmedabad. The maximum subsidence is visible to the southwest and west part of the city. The numbers on the contour indicates subsidence caused due to water level decline. (b) Comparison of PSI derived displacement and groundwater depletion derived subsidence rate.
disparity, firstly we have measured the subsidence only due to the decline of water level but the subsidence may also enhance due to the overloading of the infrastructures in the urban areas; secondly, the process of ground subsidence is slow as compared to the water level decline. Thus, the actual subsidence may be greater in the near future in areas where water levels are now rapidly declining.
Therefore, based on the results derived after analyzing the PSI dataset, GNSS dataset and water level dataset, we can conclude that the ground subsidence takes place at Ahmedabad and that the main cause of ground subsidence is the rapid decrease of groundwater levels. There can be numerous reasons for the drop in groundwater level, i.e. excessive groundwater withdrawal and poor groundwater recharge. Even so, a faster rate of water level decline (2 m/yr) was observed from 1960 to 1995 and an average drop of 1 m/yr was observed from 1995 to 2019 for the area. A report from the GWYB also mentions that the dependence on groundwater is moderately decreasing in the city (Tiwari 2016) and that the authorities as well as the government are constantly working to recharge the groundwater through various techniques. In addition, the use of groundwater has declined moderately over past decade due to the progressive measures taken by the government by providing surface water, i.e. water from the Narmada canal for the use of drinking, industrial and agricultural purposes. However, supporting hydrogeology also plays a vital role in ground subsidence. Long-term water uptake results in ground subsidence mainly due to pore pressure reduction in the aquifer system and further due to surface loading by the developmental/construction activities in the urban areas. Groundwater extraction for drinking purpose is not the only causes of ground subsidence in the present study area, as the southeast portion of land subsidence also includes a large portion of agricultural fields from Daskroi Taluka and the western area near Bopal/adjacent also includes a good area of agricultural land. Thus, ground subsidence in Ahmadabad is the combined outcome of groundwater extraction for drinking, Industrial and Agricultural purposes.

Conclusions
The current study, which is based on Persistent Scatterer Interferometry (PSI), the Global Navigation Satellite System (GNSS), and ground water depletion, indicates ground subsidence in the World Heritage city Ahmedabad, western India. The PSI results, based on 71 Sentinel 1 A datasets from 2017 to 2020, revealed three zones of subsidence in the city. The south-eastern part of the city and parts of the neighbouring rural areas experienced high levels of subsidence. The west-central and east-central part of the city reveal moderate and low subsidence zone, respectively. The maximum annual displacement (LOS) observed using PSI technology reaches a level of 25.0 ± 2.5 mm. The GNSS-derived displacement of 10 ± 1 mm/yr observed in the southeast of the city supports the PSI observations. Estimated ground subsidence using water level data shows a maximum subsidence of 9 mm/yr. The results of the PSI and GNSS in this study have been considerably verified by the ground subsidence caused by the decline of groundwater. Comparison of the extent of subsidence with the PSI results concludes that the ground subsidence in the city is caused by the decrease in the groundwater levels.
In general, the current research on the basis of PSI, GNSS and groundwater data concludes that the ground subsidence takes place in Ahmedabad city and the main cause of the subsidence is the extensive use of groundwater for drinking, industrial and agriculture.