remote sensing

Article Ground Deformation and Its Causes in City, from Sentinel-1A Data and MT-InSAR

Naeem Shahzad * , Xiaoli Ding , Songbo Wu and Hongyu Liang Department of Land Surveying and Geo-Informatics, The Hong Kong Polytechnic University, Hong Kong, China; [email protected] (X.D.); [email protected] (S.W.); [email protected] (H.L.) * Correspondence: [email protected]

 Received: 24 September 2020; Accepted: 15 October 2020; Published: 20 October 2020 

Abstract: Land subsidence, as one of the engineering geological problems in the world, is generally caused by compression of unconsolidated strata due to natural or anthropogenic activities. We employed interferometric point target analysis (IPTA) as a multi-temporal interferometric synthetic aperture radar (MT-InSAR) technique on ascending and descending Sentinel-1A the terrain observation with progressive scans SAR (TOPSAR) images acquired between January 2015 and December 2018 to analyze the spatio-temporal distribution and cause of subsidence in Abbottabad City of Pakistan. The line of sight (LOS) average deformation velocities along ascending and descending orbits were decomposed into vertical velocity fields and compared with geological data, ground water pumping schemes, and precipitation data. The decomposed and averaged vertical velocity results showed significant subsidence in most of the urban areas in the city. The most severe subsidence was observed close to old Karakorum highway, where the subsidence rate varied up to 6.5 cm/year. − The subsidence bowl profiles along W–E and S–N transects showed a relationship with the locations of some water pumping stations. The monitored LOS time series histories along an ascending orbit showed a close correlation with the rainfall during the investigation period. Comparative analysis of this uneven prominent subsidence with geological and precipitation data reflected that the subsidence in the Abbottabad city was mainly related to anthropogenic activities, overexploitation of water, and consolidation of soil layer. The study represents the first ever evidence of land subsidence and its causes in the region that will support the local government as well as decision and policy makers for better planning to overcome problems of overflowing drains, sewage system, littered roads/streets, and sinking land in the city.

Keywords: land subsidence; Sentinel-1A; multi-temporal InSAR; interferometric point target analysis (IPTA); Pakistan

1. Introduction Ground subsidence is a common geological hazard worldwide, which is mostly linked to natural (e.g., earthquake, compression of soil, and change in aquifer) or anthropogenic (e.g., underground construction activities, groundwater extraction) processes [1]. Subsidence at the local scale specifically in urban areas can be a serious threat to the economy, such as damages to infrastructures (buildings, roads, bridges, canals), and to viable and sustainable economic development [2]. Changes in geological settings, mining, construction of subways, and ground water extraction are among the main causes of land subsidence in most cities in the world [3,4]. Other than these causes, the subsidence showed high correlation with changing geology in terms of strata lithology, formation, and structure [5]. The areas with unconsolidated surficial deposits generally form the most prolific aquifers and subsidence in such areas is mostly due to the extensive extraction of ground water [6]. This extensive use of underground water resources and increase in the pumping stations due to increase in the population is causing

Remote Sens. 2020, 12, 3442; doi:10.3390/rs12203442 www.mdpi.com/journal/remotesensing Remote Sens. 2020, 12, 3442 2 of 18 significant loss in small and large aquifers in most parts of the world, creating problems related to drainage system, flooding during rainy seasons, and the functioning of structures such as canals [6–8]. Numerous events of localized land subsidence have been reported in the past couple of decades in various parts of the world, especially due to extensive groundwater discharge [9]. Variation in seasonal timings and change in precipitation patterns have disturbed the runoff and recharge in aquifer systems. This change in runoff and recharge is also linked with increased demand of water in cities, which is one of the hurdles in achieving the sustainable goals of any country. Therefore, the demand for spatio-temporal monitoring of this hazard has increased to take necessary measures and to achieve sustainable goals. A number of traditional studies describing different methodologies and effective solutions are available in the literature to monitor and study the ground deformation/subsidence processes. These traditional point-based monitoring studies include the following: global positioning system (GPS), leveling, and geological monitoring methods [10,11]. These methods lack in providing enough samples for ground-based subsidence monitoring and are very difficult to use in areas surrounded by steep slopes. Interferometric synthetic aperture radar (InSAR) has been proven to be the most efficient method, widely used to monitor land subsidence all over the world [1,12,13]. This technique is capable of detecting small changes at an unprecedented level of spatial details (centimeter to millimeter) in land surface elevation at low cast under ideal conditions [14–16]. So far, the problem incorporated in InSAR approach due to change in scatterer with time is mainly due to atmospheric and ionospheric disturbances, which lead to the introduction of decorrelation in the signal [17,18]. This is a common problem associated with the repeat-pass InSAR technique as a result of discrepancies that arise between two related SAR signals over the same location at different epochs [19]. Many advanced InSAR techniques has been developed to overcome this limitation by using multi-temporal InSAR (MT-InSAR) techniques [20]. MT-InSAR techniques work on the principal of simultaneously processing multi-SAR images to accurately extract deformation information. Zhang et al. [21] and Crosetto et al. [22] presented very good reviews of these techniques. These MT-InSAR techniques have been widely applied to monitor ground subsidence on different SAR data sets, concluding that these techniques have the ability to overcome (at least partially) the weaknesses of the conventional InSAR approach [23–25]. In this study, a step-wise and iterative technique, i.e., interferometric point target analysis (IPTA) multi-baseline technique of persistent scatterer interferometry (PSI), is implemented on Sentinel -1 (4 season 4 years) data acquired between January 2015 and December 2018 along ascending and × descending orbits in order to investigate, highlight, and understand the causes of subsidence and its temporal evolution in response to increasing demand of ground water and changes in soil properties in Abbottabad city of Pakistan (Figure1). Based on iterative process for calculating high-precision deformation on point scatterers (PSs), IPTA has been proven to hold strong power in monitoring deformation using a large number of SAR images [26–29]. The study area has various types of natural conditions surrounded by mountains and an attractive place thanks to its natural beauty. This is why the city has experienced rapid urban growth in the past decade. In order to facilitate this increased urban settlement in the area, the Government of Pakistan (GoP) has implemented various drinking water supply schemes/projects [30]. As a result of contamination in the surface water and upper level of groundwater, the need to drill deeper borewells has increased with the increased number of tubewells in the area. This study is the first ever attempt to examine the spatio-temporal deformation patterns and their causes in the region. The results obtained by implementing the advanced InSAR technique will support the local government for better decision making and planning to overcome problems such as overflowing drains, disturbed sewerage system, sinking land, and littered roads and streets. Remote Sens. 2020, 12, 3442 3 of 18 Remote Sens. 2020, 12, x FOR PEER REVIEW 3 of 19

Figure 1. Location map of study area. 2. Study Area and Datasets 2. Study Area and Datasets 2.1. Characteristics of the Study Area 2.1. Characteristics of the Study Area Abbottabad, formerly known as “the city of the maple trees” and a famous in the lower ,Abbottabad, is the administrativeformerly known headquarters as “the city of theof the Hazara maple division trees” ofand Khyber a famous Pakhtunkhwa hill station province in the lower Himalayas, is the administrative headquarters of the Hazara division of of Pakistan (Figure1). Geographically, the study area extends from 34 ◦14047.0000 N, 73◦11056.0000 E province of Pakistan (Figure 1). Geographically, the study area extends from 34°14′47.00″ N, to 34◦ 7046.0000 N, 73◦16056.0000 E. Endowed with scenic and green mountains located about 125 km north73°11′56.00 of ″ E to 34° (Capital 7′46.00″ ofN,Pakistan) 73°16′56.00 through″ E. Endowed with Highway,scenic and the green area mountains is a very famouslocated hillabout resort 125 kilometers in summer. north With of Islamabad an altitude (Capital of 1260 of m Pakistan) (4120 ft), through the area Karakoram also serves Highway, as a gateway the area to Galliyatis a very regions famous in hill west. resort Geologically, in summer. the With central an altitude part of the of study1260 m area (4120 consists ft), the of area Quaternary also serves Alluvial as a (unconsolidatedgateway to Galliyat surficial regions deposits in west. of sandGeologically, and gravel) the surroundedcentral part byof Hazara,the study Tanawal, area consists Lokhart, of andQuaternary Hazira formationAlluvial (unconsolidated [31]. surficial deposits of sand and gravel) surrounded by Hazara, Tanawal,Being Lokhart, the part and of Orash Hazira Valley, formation rainfall [31]. occurs throughout the year with the highest rainfall during monsoonBeing seasons, the part makingof Orash the Valley, region rainfall prone tooccurs flashflood. throughout According the year to thewith recent the highest census rainfall report, theduring population monsoon of theseasons, city has making increased the region at a rate prone of 3.61% to /flashflood.year from 1998According to 2017 to [32 the]. recent census report, the population of the city has increased at a rate of 3.61%/year from 1998 to 2017 [32]. 2.2. Datasets 2.2. Datasets The selection of Sentinel-1 data was based on the fact that this mission provides long-term, timelyThe data selection with satisfactory of Sentinel-1 resolution data was free based of cost. on the These fact datathat arethis amission reliable provides source to long-term, study the deformationtimely data with and itssatisfactory evolution overresolution time, whichfree of can cost. be usedThese to data predict are futurea reliable scenarios source [33 to]. Therefore,study the indeformation this study, and we its used evolution Sentinel over -1A time, data which along can the be ascending used to predict and descending future scenarios orbits (incident[33]. Therefore, angle ~33.81in this study, and 33.15 we degrees,used Sentinel respectively). -1A data along Interferometric the ascending wide and swath descending (IW) single orbits look (incident complex angle (SLC) ~ the33.81 terrain and 33.15 observation degrees, with respectively). progressive Interferometric scans SAR (TOPSAR) wide swath images, (IW) operating single look in C-band complex (~5.6 (SLC) cm wavelength),the terrain observation acquired fromwith progressive European Space scans Agency SAR (TOPSAR) (ESA) distributed images, operating under Copernicus in C-band program, (~5.6 cm werewavelength), used to mapacquired the spatial from distributionEuropean Space of ground Agency deformation (ESA) distributed and time under series Copernicus deformation program, patterns. Thewere 84 used acquired to map ascending the spatial orbit imagesdistribution (track 100)of ground were between deformation 4 January and 2015 time and series 2 December deformation 2018, patterns. The 84 acquired ascending orbit images (track 100) were between 4 January 2015 and 2 December 2018, and the 82 descending orbit images (track 107) were between 17 January 2015 and 27 December 2018. All the acquired images were preprocessed for orbital refinement using precise orbit

Remote Sens. 2020, 12, 3442 4 of 18 and the 82 descending orbit images (track 107) were between 17 January 2015 and 27 December 2018. All the acquired images were preprocessed for orbital refinement using precise orbit determination (POD) files released by ESA. Moreover, 1 arc second (~30 m) shuttle radar topography mission (SRTM) digital elevation model (DEM) was downloaded from United States geological survey (USGS) and used to simulate and remove topographic phases. For processing and co-registration, the acquired images on 8 August 2017 for ascending data and 13 December 2016 for descending data were selected as master images. This selection of the master image was based on the joint correlation method presented in [34]. Google Earth was used to interpret and analyze the data at different stages. Advanced InSAR processing software, GAMMA-2017®, developed by GAMMA Remote Sensing AG, Switzerland, was used to define and implement the IPTA processing chain for the mapping of ground subsidence in the area.

3. Methodology The use of the InSAR technique for ground deformation mapping and monitoring depends on a number of factors, including operational band (C, X, or L), type of incident surface, topography and geography of the area, number of acquisitions, and temporal intervals between repeated images. These factors play an important role in achieving high efficiency of the InSAR technique [35]. To achieve such efficiency of the results, the IPTA approach was employed to analyze and monitor the ground subsidence in the area. Although the phase model used for point target analysis is the same as that used in conventional interferometry, its step-wise iterative process is very helpful in refining the parameters and to separate the deformation phase from all non-deformation phase components of Equation (1) [36]. k = k + k + k + k ϕi,unw ϕi,de f o ϕi,noise ϕi,topo ϕi,atm (1)

k where ϕi,unw is the unwapped phase of the ith point target in the kth interferogram and is expressed as the k k k sum of ϕi,atm (atmospheric path delay phase), ϕi,noise (decorrelation phase noise), ϕi,topo (phase residues k due to the digital elevation model (DEM) inaccuracies), and actual ϕi,de f o deformation phase. k The term ϕi,topo is the phase attributed to the errors introduced by DEM as elaborated in Equation (2) by [35],  k  4π B  k  i,perp  ϕ =  ∆ε (2) i,topo  λ RSinθ 

k where ∆ε is the height error (m) increment; Bi,perp is the perpendicular baseline; and λ, R, and θ are the radar wavelength, slant range, and look angle, respectively. k The term ϕi,de f o is actual deformation phase that can be further divided into linear and non-linear components [35]. The linear component of this deformation phase is presented in Equation (3).

4π  k = k ϕi,de f o lin t vi (3) − λ

k where t is the time difference of the kth interferogram and vi is the linear deformation velocity at the ith point target. Moreover, the efficiency of the models and data storage capacities is one of the main concerns of the researchers working with limited computer resources. The vector format data structure used in this approach has provided a great benefit to generate results even with the limited system resources. The overall processing sequence adopted for IPTA processing in this research study is presented in Figure2. RemoteRemote Sens.Sens. 20202020,,12 12,, 3442x FOR PEER REVIEW 55 of 1819 Remote Sens. 2020, 12, x FOR PEER REVIEW 5 of 19

Figure 2. Processing workflow for interferometric point target analysis (IPTA). Figure 2. ProcessingProcessing workflow workflow for interferomet interferometricric point target analysis (IPTA). 3.1. Generation of Point List and Differential Interferograms 3.1. Generation of Point List and Differential Differential Interferograms In order to minimize the contribution of non-deformation phases elaborated in Equation (1), generationIn order of toa good minimize quality the of contributioncontripointsbution is important. of non-deformation In this study, thphases e strategy elaborated used to in indetermine Equation Equation good (1), (1), generationquality point of of a atargets, good good quality qualityi.e., persistent of of poin pointsts isscatterers important. is important. (PS), In thiswas In thisstudy, based study, th one strategy thetwo strategy approaches, used usedto determine toi.e., determine temporal good goodqualityvariability quality point and pointtargets, spectral targets, i.e., diversity i.e.,persistent persistent [36]. scatterers The scatterers ratio (PS),of the (PS), was mean was based based(temporal on on two two average approaches, approaches, of backscattering) i.e., i.e., temporal to variabilitythe standard and deviation spectral diversity(sigma) of [[36].36 backscatter]. The ratio from of the this mean average (temporal was used average for oftemporal backscattering) variability. to theFor standardspectral diversity,deviation an(sigma) averaged of of backscatter coherence from fromfrom thisthis each average average SLC was was was used used used to for for select temporal temporal the PS variability. variability. with low Forspectral spectralspectral diversity. diversity,diversity, In order an an averaged averaged to have coherence enough coherence PS, from thfrome each point each SLC targets SLC was usedwasof both toused selectthe to approaches theselect PS withthe werePS low with spectralmerged low diversity.spectraltogether. diversity. For In order these In to points, order have enough theto have SLC PS, enoughvalues the pointare PS, extracted th targetse point ofand targets both written the of both approaches in the the vector approaches were data merged SLC were point together.merged data Fortogether.stack. these Taking For points, these advantage the points, SLC of valuesthe the SLC long-term are values extracted aretime extracted seri andes written data, and the written in maximum the vector in the scenevector data SLCdifference data point SLC of datapoint three stack. data (3) Takingstack.scenes Taking advantagewas considered advantage of the with long-termof the perpendicular long-term time series time baseline data,series the data,check maximum the of maximum200 scenem and di scenetemporalfference difference ofbaseline three of (3) threecheck scenes (3) of wasscenes100 considereddays was to considered generate with perpendicular interferometric with perpendicular baseline pairs. baseline checkBased ofon check 200 the m ofbaseline and 200 temporal m chartand temporal baseline(Figure 3), checkbaseline these of 100checkPS dayswere of to100translated generate days to interferometrictogenerate generate interferometric point-based pairs. Based pairs.DEM, on theBase leadin baselined ong theto chart calculatingbaseline (Figure chart3 ),the these (Figure point-based PS were3), these translated differential PS were to generatetranslatedinterferogram. point-based to generate DEM, point-based leading to calculatingDEM, leadin theg point-basedto calculating diff erentialthe point-based interferogram. differential interferogram.

FigureFigure 3.3. BaselineBaseline vs.vs. time plots (left)) ascendingascending orbitorbit datadata ((rightright)) descendingdescending orbitorbit data.data. 3.2. InterferometricFigure 3. PointBaseline Target vs. time Processing plots (left) ascending orbit data (right) descending orbit data. 3.2. Interferometric Point Target Processing 3.2. InterferometricIPTA is based Point on aTarget 2D linear Processing regression model, in which the phase difference between point IPTA is based on a 2D linear regression model, in which the phase difference between point targets is analyzed in the temporal domain. This 2D regression is based on the linear dependency of targetsIPTA is analyzed is based inon the a 2D temporal linear regressiondomain. This mode 2D l,regression in which isthe based phase on difference the linear betweendependency point of the topographic phase with the perpendicular baseline (Equation (2)) and the linear dependency of targetsthe topographic is analyzed phase in the with temporal the perpendicular domain. This base 2Dline regression (Equation is based (2)) and on the linearlinear dependencydependency ofof the deformation phase with time (Equation (3)). This 2D linear regression was applied to generate thethe topographicdeformation phase phase with with the time perpendicular (Equation (3)). base Thisline 2D (Equation linear regression (2)) and the was linear applied dependency to generate of initial estimates of deformation phase, residual topography, and phase standard deviation. This phase theinitial deformation estimates phaseof deformation with time phase,(Equation residual (3)). Thistopography, 2D linear and regression phase standardwas applied deviation. to generate This initialphase estimatesstandard deviationof deformation represented phase, a residualquality ch topography,eck between and the modelphase standardfit and differential deviation. phase. This phase standard deviation represented a quality check between the model fit and differential phase.

Remote Sens. 2020, 12, 3442 6 of 18 standard deviation represented a quality check between the model fit and differential phase. In addition to these terms, the phase model also generated the linear regression phase residual, which was comprised of phase variations related to atmospheric artifacts as well as non-linear deformation phase component with decorrelation noise (Equation (4)).

k = k + k + k ϕi,res ϕi,non lin ϕi,atm ϕi,noise (4) − The residual phase at this stage exhibited phase jumps related to unwrapping errors that were removed iteratively using the minimum cost flow (MCF) algorithm [37]. Because atmospheric phase screen (APS) and non-linear deformation phases are spatial low-pass signals, we applied the linear least square spatial filtering method to the unwrapped residual phase in order to calculate the phase attributes related to APS. The estimated APS is then subtracted from the point-based differential interferogram stack for all of the PS candidates. As presented in Figure2, this process was iterated several times to generate improved quality estimates of the height correction, APS, and refined unwrapped differential phase. More technical details on processing strategies using the IPTA framework can be found in [36,38].

3.3. Deformation Rate and Time Series Retrieval The final deformation rate and deformation time series histories were retrieved for each point by implementing the multi-baseline (mb) approach [39]. These final deformation histories and average annual deformation rates corresponding to point information are converted to line of sight (LOS) displacement and equiangular (EQA) projection for further analysis.

3.4. Decomposition of LOS Displacements into 2D Displacement Fields The mean LOS velocity fields against each PS, extracted from two different orbits, i.e., ascending and descending, were decomposed into vertical and east–west components. Assuming the negligible north–south movement [40], the vertical (dv) and east–west (dh) components were computed using Equation (5) [41,42]. asc ! asc sinθasc ! ! dlos cosθ cos∆α dv dsc = dsc dsc (5) dlos cosθ sinθ dh where θasc and θdsc represent the incident angles of ascending and descending orbits, respectively, and ∆α is the difference between headings of ascending and descending orbits.

3.5. Causal Factors of Deformation The cause of the subsidence is normally considered and interpreted as an individual/single factor, but urban subsidence especially in areas surrounded by hills with variety of geological units may be caused by the interaction of two or more factors together. This interaction between causal factors is very clear and reported in many studies [43]. If changes in such casual factors are left unnoticed, then the subsidence can lead to significant disaster in the area. Abbottabad, being one of the major affected cities in the October 2005 earthquake, has attracted a number of researchers for several reasons. One reason is its location in an intermountain basin with thick Quaternary deposits. Apart from the number of causal factors of subsidence, we gathered data about natural and anthropogenic factors that influence land subsidence. Geological layers (Figure4) were extracted from the map of Geological Survey of Pakistan (GSP) [44] and refined using the map published in [31]. Remote Sens. 2020, 12, 3442 7 of 18 Remote Sens. 2020, 12, x FOR PEER REVIEW 7 of 19

Figure 4. GeologicalGeological map of the study area.

4. ResultsPrecipitation data were gathered from Kakul Station of Pakistan Meteorological Department (PMD) from the year 2015 to 2018. Moreover, information related to the increase in drinking water The criteria for the generation of interferometric pairs on the basis of multi-baseline can supply schemes was collected from various reports [45–47] and compiled for the analysis. Because of guarantee that all of the Sentinel 1A images are involved. These interferometric pairs were assessed the unavailability of ground water discharge data at these water supply schemes, the analysis with by applying a standard two-pass D-InSAR algorithm to generate differential interferograms. During ground water discharge was done with the help of the implemented plans of water supply under such D-InSAR processing, it was observed that the interferograms with relatively longer temporal schemes. The pattern of subsidence and the locations of high subsidence zones were compared and intervals showed the phase variations in some part of the study area. The phase variations could be analyzed with these developments in the area. related to topographic residuals and perpendicular baselines. Such phase variation cannot be categorized4. Results to a specific phase signal unless multiple D-InSAR products based on time series data are analyzed. For this purpose, IPTA was employed in Abbottabad city and adjacent surrounding areas to extractThecriteria deformation for the information. generation of interferometric pairs on the basis of multi-baseline can guarantee that all of the Sentinel 1A images are involved. These interferometric pairs were assessed by applying 4.1.a standard Average two-passSubsidence D-InSAR Rate Map algorithm to generate differential interferograms. During D-InSAR processing, it was observed that the interferograms with relatively longer temporal intervals showed the phaseFigure variations5 shows the in average some part velocities of the studyin cm pe area.r year The along phase LOS variations directions could of both be ascending related to (left)topographic and descending residuals (right) and perpendicular data, derived baselines.by implementing Such phase the IPTA variation technique. cannot These be categorized LOS results to werea specific geocoded phase in signal WGS unless84 coordinate multiple system D-InSAR and productssuperimposed based onto on time the seriesshaded data relief are map analyzed. of the studyFor this area. purpose, Color IPTAramps was from employed blue to red in Abbottabad represent the city movement and adjacent of land surrounding along the areasLOS direction. to extract Thedeformation negative information.deformation values indicate the movement away from the sensor (i.e., subsidence) and positive deformation values are related to movement towards the sensor (i.e., uplift). Owing to lower positive4.1. Average values Subsidence (Figure Rate 5), Mapit can be assumed that there is no movement towards the sensor (i.e., uplift); therefore, the term “deformation” is represented as “subsidence” in this research paper. It can Figure5 shows the average velocities in cm per year along LOS directions of both ascending (left) be seen in Figure 5 that the results extracted from both orbits’ data exhibit similar spatial patterns. and descending (right) data, derived by implementing the IPTA technique. These LOS results were These similar patterns represent the presence of dominant vertical deformation and negligible geocoded in WGS 84 coordinate system and superimposed onto the shaded relief map of the study area. horizontal velocity [42]. Color ramps from blue to red represent the movement of land along the LOS direction. The negative The results of average vertical velocity obtained after decomposing LOS velocities of both orbits are shown in Figure 6. It was observed in Figure 6 that the averaged vertical deformation (subsidence) rate varies up to −6.5 centimeter (cm) per year in the area. Figure 6 also represents a prominent, un-

Remote Sens. 2020, 12, 3442 8 of 18 deformation values indicate the movement away from the sensor (i.e., subsidence) and positive deformation values are related to movement towards the sensor (i.e., uplift). Owing to lower positive Remote Sens. 2020, 12, x FOR PEER REVIEW 8 of 19 values (Figure5), it can be assumed that there is no movement towards the sensor (i.e., uplift); therefore, the term “deformation” is represented as “subsidence” in this research paper. It can be seen in Figure5 even subsidence pattern and the most prominent pattern with values from −6.5 to −2 cm/year was that the results extracted from both orbits’ data exhibit similar spatial patterns. These similar patterns identified along the old (locally known as road Nowshera-Abbottabad road, represent the presence of dominant vertical deformation and negligible horizontal velocity [42]. N35), which is settled in the foothills of the Himalayas, including the area of the city.

Figure 5. AverageAverage line of sight (LOS) velocity maps: ( aa)) ascending ascending and (b)) descending descending orbits’ data.

Moreover,The results Figure of average 6 also vertical represents velocity three obtained main subs afteridence decomposing regions, i.e., LOS A, velocities B, and C. of A both and orbitsB are theare regions shownin with Figure high6. population It was observed density in Figure and exhibi6 thatt thehigher averaged subsidence vertical rates, deformation whereas region (subsidence) C falls intorate the varies medium up to to 6.5low centimeter subsidence (cm) region per with year a in relatively the area. scattered Figure6 population. also represents The subsidence a prominent, in − regionun-even A subsidenceincorporates pattern residential and the societies most prominent of Hassan pattern town, withAmir values alam awan from town.6.5 to The2 cm subsidence/year was − − variesidentified from along −6.5 thecm/year old Karakoram to −0.5 cm/year highway in this (locally region. known Region as road B lies Nowshera-Abbottabad in Mir Alam Town and road, Jhangi N35), areawhich with is settled a higher in thecentered foothills value of theof − Himalayas,6.5 cm/year including up to −1 cm/year the cantonment at the outer area ofedges the city.of the region. Region C is the northern part of the main city, and lies in Kaghan and Sir Syed Colony with relatively higher altitudes than regions A and B. The subsidence rate varies between −3.0 and −1 cm per year in this region.

Remote Sens. 2020, 12, 3442 9 of 18 Remote Sens. 2020, 12, x FOR PEER REVIEW 9 of 19

Figure 6. 6. MeanMean vertical vertical velocity velocity map map in incm/year. cm/year. The The red redcolor color represents represents down downwardward movement movement (i.e., subsidence).(i.e., subsidence).

4.2. SubsidenceMoreover, Time Figure Series6 also represents three main subsidence regions, i.e., A, B, and C. A and B are the regions with high population density and exhibit higher subsidence rates, whereas region C falls into theThe medium spatio-temporal to low subsidence evolution regionin the deformation with a relatively pattern scattered using population.ascending orbit The data subsidence at six (6) in differentregion A stages incorporates overlaid residential on averaged societies multi-looked of Hassan intensity town, Amir (MLI) alam image awan in town.RADAR The geometry subsidence is shownvaries fromin Figure6.5 cm 7./ yearAnalyzing to 0.5 cmthe/ yearincreasing in this region.and expanding Region B trend lies in Mirof the Alam deformation Town and Jhangiin the − − eastward/westwardarea with a higher centered and southward/westward value of 6.5 cm/year directio up tons,1 respectively, cm/year at the four outer typical edges PS of (P1, the P2, region. P3, − − andRegion P4) Cwere is the selected northern for partdetailed of the regional main city, time and seri lieses analysis in Kaghan (Figure and Sir8a–e). Syed Time Colony series with trends relatively show thehigher deformation altitudes thanevaluation regions extracted A and B. using The subsidence ascending rate data varies and scattered between at3.0 different and 1 locations cm per year over in − − thethis deforming region. area. The locations of these points are shown in Figure 8a.

4.2. Subsidence Time Series The spatio-temporal evolution in the deformation pattern using ascending orbit data at six (6) different stages overlaid on averaged multi-looked intensity (MLI) image in RADAR geometry is shown in Figure7. Analyzing the increasing and expanding trend of the deformation in the eastward /westward and southward/westward directions, respectively, four typical PS (P1, P2, P3, and P4) were selected for detailed regional time series analysis (Figure8a–e). Time series trends show the deformation

Remote Sens. 2020, 12, 3442 10 of 18 evaluation extracted using ascending data and scattered at different locations over the deforming area. The locations of these points are shown in Figure8a. Based on LOS deformation obtained using ascending data, the temporal variations in the subsidence values of P1 vary from 0 to 25 cm, with an average rate of 6.63 cm/year. P2, which lies − − in the upper part of subsidence zone A, exhibits an average rate of 5 cm/year and the trend varies − from 0 to 18 cm. P3, which lies in the upper part of subsidence zone B, varies from 0 to 16 cm, − − with an average rate of 4.3 cm/year. P4, which lies in northern part of study area and in the center of − subsidence zone C, shows lower values varying from 0 to 12 cm, with an average rate of 3 cm/year. − − The temporal variations of these points are shown altogether in Figure8b–e. These variations show a non-linear trend in the subsidence. The variations in time series patterns show that deformation slows down and is sustained for a certain period of time. One of the possible reasons for this variation could be the seasonal changes in the underground water levels, particularly after the monsoon season

Remoteevery Sens. year. 2020, 12, x FOR PEER REVIEW 10 of 19

Figure 7. SixSix different different stages stages of ofcumulative cumulative land land subsidence subsidence pattern pattern in RADAR in RADAR geometry. geometry. The Thebackground background image image is the is averaged the averaged intensity intensity image image of the of thestudy study area. area. One One color color cycle cycle corresponds corresponds to 10to 10cm/year cm/year displacement. displacement.

Based on LOS deformation obtained using ascending data, the temporal variations in the subsidence values of P1 vary from 0 to −25 cm, with an average rate of −6.63 cm/year. P2, which lies in the upper part of subsidence zone A, exhibits an average rate of −5 cm/year and the trend varies from 0 to −18 cm. P3, which lies in the upper part of subsidence zone B, varies from 0 to −16 cm, with an average rate of −4.3 cm/year. P4, which lies in northern part of study area and in the center of subsidence zone C, shows lower values varying from 0 to −12 cm, with an average rate of −3 cm/year. The temporal variations of these points are shown altogether in Figure 8b–e. These variations show a non-linear trend in the subsidence. The variations in time series patterns show that deformation slows down and is sustained for a certain period of time. One of the possible reasons for this variation could be the seasonal changes in the underground water levels, particularly after the monsoon season every year.

Remote Sens. 2020, 12, 3442 11 of 18 Remote Sens. 2020, 12, x FOR PEER REVIEW 11 of 19

FigureFigure 8.8. TimeTime series series evaluation evaluation of of averaged averaged LOS LOS deformation deformation presented presented using using ascending ascending data: data: (a) (alocation) location of ofPSs, PSs, (b) ( bP1,) P1, (c) (P2,c) P2, (d) ( dP3,) P3, and and (e) (P4.e) P4.

5.5. DiscussionDiscussion andand AnalysisAnalysis BeingBeing part part of of the the valley valley and and foothills, foothills, the citythe exhibitscity exhibits heavy heavy flash floods.flash floods. Several Several drainage drainage systems weresystems introduced were introduced in the area in to the move area the to floodedmove the water flooded from water the city from to thethe nearby city to riverthe nearby channels river via Darkhanchannels Katha via Darkhan into Dor Katha River. into With Dor the River. passage With oftime, the passage the situation of time, of floodedthe situation and standing of flooded water and in thestanding city is becomingwater in the more city severe, is becoming which resultsmore severe, in paralysing which theresults city in with paralysing traffic jams, the especiallycity with traffic during rainfalljams, especially seasons [ 48during]. To better rainfall understand seasons [48]. this scenario,To better theunderstand subsidence this profiles scenario, along the two subsidence transects, westprofiles to east along (W–E) two and transects, south towest north to east (S–N), (W–E) were and analyzed. south to It wasnorth observed (S–N), were that analyzed. the magnitude It was of LOSobserved subsidence that the along magnitude these transects of LOS decreases subsidence in value along while these moving transects from decreases W–E or S–N,in value keeping while the maximummoving from values W–E in or the S–N, centers keeping of the the transects maximum and valu causinges in athe bowl-shaped centers of the profile, transects particularly and causing along thea bowl-shaped W–E transect, profile, and some particularly major dipsalong in the the W–E subsidence transect, profile and some along major the S–Ndips transectin the subsidence (Figure9). Theprofile comparison along the of S–N these transect subsidence (Figure profiles 9). The withcomparis the actualon of these ground subsidence elevation profiles profiles with can the be seenactual in Figureground9. Itelevation is clear fromprofiles Figure can9 bethat seen these in bowl-shapedFigure 9. It is regions clear from of subsidence Figure 9 that zones these in this bowl-shaped part of the regions of subsidence zones in this part of the city are basically responsible for the standing water

Remote Sens. 2020, 12, 3442 12 of 18

Remote Sens. 2020, 12, x FOR PEER REVIEW 12 of 19 city are basically responsible for the standing water during the rainfall season, limiting human activity; navigation;during the andrainfall severe season, destruction limiting to human roads, streets,activity; and navigation; buildings, and making severe them destruction prone to to seepage roads, asstreets, well. and buildings, making them prone to seepage as well.

FigureFigure 9.9. SubsidenceSubsidence profileprofile alongalong thethe S–NS–N andand W–EW–E transects.transects.

InIn general,general, mostmost subsidencesubsidence areasareas inin AbbottabadAbbottabad belongbelong to residential area and commercial area. TheThe subsidencesubsidence inin thesethese areasareas isis ofof coursecourse duedue toto somesome undergroundunderground activitiesactivities relatedrelated toto variousvarious anthropogenicanthropogenic (e.g., (e.g., ground ground water water withdrawal) withdrawal) and and natural natural (e.g., (e.g., soil consolidation)soil consolidation) processes. processes. In most In cases,most cases, the land the subsidence land subsidence pattern pattern represents represents an integrated an integrated effect of effect both of of both these of factors. these factors. In such In cases, such thecases, anthropogenic the anthropogenic factors actfactors as a speedact as upa speed tool for up the tool naturally for the occurring naturally eventsoccurring such events as subsidence such as andsubsidence degradation. and degradation. Because of the Because increase of inthe population increase in with population time, this with subsidence time, this and subsidence degradation and candegradation be related can to thebe related increase to inthe the increase demand in the of the demand underground of the underground natural resources natural (water resources discharge). (water Thedischarge). detailed The relationships detailed relationship of land subsidences of land with subsidence the categorized with the factors categorized are discussed factors are below. discussed below. 5.1. Soil Consolidation 5.1. SoilUnconsolidated Consolidation soil is postulated as a primary driver of land subsidence in most of the major citiesUnconsolidated and areas of the worldsoil is [6postulated]. Figure4 showsas a primary that most driver of the of central land subsidence part of the studyin most area of isthe located major incities alluvial and sedimentsareas of the with world grey [6]. brown Figure shale 4 shows and minorthat most limestone of the incentral the west; part dolomite,of the study dark area grey is shale,located light in alluvial grey medium sediments to thick with bedded grey brown limestone shal ine and the south;minor andlimestone dolomite in the limestone west; dolomite, comprised dark of sandstonegrey shale, and light shale grey in themedium north [31to ].thick Figure bedded 10 shows limestone the relationship in the south; between and these dolomite geological limestone units andcomprised average of vertical sandstone subsidence and shale rates. in Itthe can north be observed [31]. Figure in Figure 10 shows 10 that the mostrelationship of the subsidence between these area belongsgeological to unconsolidatedunits and average deposits vertical (alluvial subsidence surfaces). rates. It can be observed in Figure 10 that most of the subsidenceKeeping in area view belongs the importance to uncons ofolidated unconsolidated deposits deposits, (alluvial whichsurfaces). makes mattresses of aquifers, the three-layered structure analysis as presented by [46] showed that the first layer is comprised of clay and silt along the inter-beds of gravel, with a thickness of 30 to 40 m in the west and 70 to 80 m in the central part of the subsidence area; the second layer majorly consists of gravel with a conspicuously increased thickness of about 40 m in the center and southern part of the subsidence area; and the third layer of the unconsolidated soil in the region is underlying bedrock, represented by shale in the west and dolomite and limestone in the east. The thickness of the second layer increases while moving in a westerly direction along the transect W–E, which follows exactly the western extent of the subsidence region. This structural analysis showed a strong correlation with the overall extent subsidence. Although most of the land subsidence occurred in alluvium deposits, the direct correlation between this type of soil with subsidence may not be solely possible, but the direct correlation of this type of geological unit with the existence of aquifer and its exploitation (changing the unconsolidated

Remote Sens. 2020, 12, 3442 13 of 18 surface to consolidated surface) could provide a better understanding of the land subsidence process inRemote this Sens. populated 2020, 12, partx FOR of PEER the REVIEW city. 13 of 19

Figure 10.10. Relationship between geological unitsunits and average subsidencesubsidence rate.rate.

5.2. EKeepingffect of Ground in view Water the importance Withdrawal of unconsolidated deposits, which makes mattresses of aquifers, the three-layeredAnthropogenic structure factors mostlyanalysis include as presented infrastructure by [46] development,showed that the mining first activities,layer is comprised and ground of waterclay and exploitation. silt along the Abbottabad inter-beds of city gravel, and thewith surrounding a thickness of area 30 areto 40 mostly m in the dependent west and on70 surfaceto 80 m waterin the and central ground part water. of the The subsidence increase in area; the contamination the second layer of water majorly due toconsists mixing of of mineralsgravel with from a surroundedconspicuously hills increased was bringing thickness water-borne of about diseases40 m in intothe center the community and southern and becomingpart of the more subsidence severe witharea; timeand the [49]. third In order layer to of handle the unconsolidated this situation, soil Government in the region bodies is underlying implemented bedrock, a number represented of safe drinkingby shale waterin the schemeswest and in dolomite the area. and One limestone of the major in the and east. most The recent thickness schemes of wasthe aidedsecond by layer the Japaneseincreases Governmentwhile moving and in a implemented westerly direction by Japan along International the transect Cooperation W–E, which Agency follows (JICA) exactly after the conductingwestern extent a survey of the in subsidence 2009 [46]. Underregion. thisThis project, structural combined analysis use showed of surficial a strong and correlation underground with water the wasoverall proposed extent subsidence. by repairing Although the old tubewells most of the and land building subsidence new tubewellsoccurred in in alluvium the areas deposits, with dense the population.direct correlation For this between purpose, this atype network of soil of with water subsidence supply lines may connectingnot be solely surface possible, water but reservoirs the direct andcorrelation tubewells of this was type developed of geological (Figure unit 11 with), implemented, the existence and of aquifer handed and over its toexploitation the Government (changing in Octoberthe unconsolidated 2015 [50]. surface to consolidated surface) could provide a better understanding of the land subsidenceBy comparing process the in this subsidence populated of the part area of withthe city. the existing and implemented water supply scheme, it can be seen in Figure9 that the crests along profiles W–E and S–N belong to the locations of tubewells. 5.2. Effect of Ground Water Withdrawal Similarly, it can be seen in Figure 11 that the extent and pattern of the subsidence zone follow the distributedAnthropogenic locations offactors tubewells mostly/boreholes. include Theinfrastructure edges of subsidizing development, zones mining are at relatively activities, higher and altitudesground water than theexploitation. central part Abbottabad of the city city and and follow the thesurrounding locations ofarea other are tubewellsmostly dependent in this area. on Assurface elaborated water in and Section ground 5.1, thewater. aquifers The inincrease the study in areathe arecontamination identified in theof water first and due second to mixing layers ofof theminerals underground from surrounded soil and groundwater hills was inbringing these aquifers water-borne moves withdiseases varying into water the levelcommunity from 1190 and to 1210becoming m mean more sea levelsevere (M.S.L); with therefore,time [49]. it wasIn order emphasized to handle by the this JICA situation, team during Government the handing bodies over ofimplemented the projectto a number monitor of the safe water drinking discharge water in scheme order tos avoidin the majorarea. One future of issues,the major such and as most subsidence recent inschemes the area. was The aided increase by of the 3.5 percentJapanese population Government per year and since implemented 1999 in 2017, by including Japan anInternational increase of 17Cooperation percent from Agency 2009 to(JICA) 2015 after [47], ultimatelyconducting increased a survey thein 2009 drinking [46]. Under water demandthis project, in the combined area. use of surficial and underground water was proposed by repairing the old tubewells and building new tubewells in the areas with dense population. For this purpose, a network of water supply lines connecting surface water reservoirs and tubewells was developed (Figure 11), implemented, and handed over to the Government in October 2015 [50].

Remote Sens. 2020, 12, 3442 14 of 18 Remote Sens. 2020, 12, x FOR PEER REVIEW 14 of 19

FigureFigure 11. Location 11. Location of tubewells of tubewells and andsupply supply lines lines overla overlaidid on average on average vertical vertical velocity velocity map. map.

Nevertheless,By comparing the the area subsidence of higher of subsidence the area rateswith agreesthe existing well with and the implemented tubewell in thewater study supply area; thescheme, deformation it can be mostly seen in started Figure with 9 that an excessivethe crests increasealong profiles of water W–E pumping and S–N from belong developed to thetubewells locations duringof tubewells. the investigating Similarly, it period.can be seen One in of Figure the other 11 possibilitiesthat the extent of and this subsidencepattern of the could subsidence be the water zone leakagefollow the problem distributed in the undergroundlocations of pipes,tubewells/bo as therereholes. have been The more edges than of 800 subsidizing differentcases zones reported are at regardingrelatively higher water leakagealtitudes issues than inthe the central area. part Moreover, of the city illegal and connections follow the locations and uncontrolled of other tubewells pumping hoursin thiscould area. As also elaborated be responsible in Section for the 5.1, changing the aquifers water in table the study in the area area. are identified in the first and secondIn orderlayers toof verifythe underground the cause of soil this and change groundwater in the area, in these we analyzedaquifers moves the temporal with varying correlation water betweenlevel from IPTA-InSAR 1190 to 1210 derived m mean time sea series level subsidence (M.S.L); attherefore, investigated it was PSs, emphasized i.e., P1, P2, P3,by andthe P4,JICA and team the precipitationduring the handing (Figure over 12). of The the collected project to precipitation monitor the water data at discharge the nearby in order Kakul to station avoid (themajor location future ofissues, the station such as is subsidence shown in Figure in the 11area.) showed The increase a strong of correlation3.5 percent withpopulation time series per year trend since (Figure 1999 12 in). The2017, seasonal including variation an increase in the groundof 17 percent water particularlyfrom 2009 to after 2015 monsoon [47], ultimately rainfall (June increased through the September) drinking justifieswater demand the cause in ofthe the area. subsidence in the area. The monsoon rainfall refills the aquifers, balancing the undergroundNevertheless, water the discharge area of higher and stabilizing subsidence the rates process agrees of well subsidence. with the Thistubewell clearly in the demonstrates study area; thethe processdeformation and causemostly of started subsidence with in an the excessive study area. increase Zhang of et water al. [51 pumping] and Galloway from developed et al. [52] showedtubewells a similarduring relationship the investigating and joint period. effect One of natural of the andother anthropogenic possibilities of conditions this subsidence with subsidence. could be Moreover,the water leakage the identification problem in of the high underground subsidence particularly pipes, as there in the have peripheries been more of tubewellthan 800 locationsdifferent showedcases reported a good abilityregarding of Sentinel water 1Aleakage data and issues the IPTAin the approach area. Moreover, for land subsidence illegal connections mapping in and this partuncontrolled of the region. pumping hours could also be responsible for the changing water table in the area. In order to verify the cause of this change in the area, we analyzed the temporal correlation between IPTA-InSAR derived time series subsidence at investigated PSs, i.e., P1, P2, P3, and P4, and the precipitation (Figure 12). The collected precipitation data at the nearby Kakul station (the location

Remote Sens. 2020, 12, x FOR PEER REVIEW 15 of 19

of the station is shown in Figure 11) showed a strong correlation with time series trend (Figure 12). The seasonal variation in the ground water particularly after monsoon rainfall (June through September) justifies the cause of the subsidence in the area. The monsoon rainfall refills the aquifers, balancing the underground water discharge and stabilizing the process of subsidence. This clearly demonstrates the process and cause of subsidence in the study area. Zhang et al. [51] and Galloway et.al. [52] showed a similar relationship and joint effect of natural and anthropogenic conditions with subsidence. Moreover, the identification of high subsidence particularly in the peripheries of Remotetubewell Sens. 2020locations, 12, 3442 showed a good ability of Sentinel 1A data and the IPTA approach for15 land of 18 subsidence mapping in this part of the region.

FigureFigure 12.12. TimeTime seriesseries LOSLOS subsidencesubsidence ofof PSPS pointspoints P1,P1, P2,P2, P3,P3, andand P4P4 versusversus dailydaily precipitationprecipitation atat KakulKakul Station.Station.

6.6. ConclusionsConclusion and and Future Future Work Work InIn thisthis study,study, thethe multi-temporalmulti-temporal InSARInSAR approachapproach ofof IPTAIPTA waswas employedemployed onon multi-trackmulti-track SentinelSentinel 1A1A TOPSARTOPSAR images images collected collected between between January January 2015 2015 and Decemberand December 2018 to 2018 analyze to analyze the spatio-temporal the spatio- distributiontemporal distribution and trend and of subsidencetrend of subsidence in Abbottabad in Abbottabad City and City the and surrounding the surrounding area. Steparea. byStep step by processingstep processing chain chain of iterative of iterative IPTA IPTA approach approach was employed was employed and the and causes the causes of the of subsidence the subsidence were investigatedwere investigated by analyzing by analyzing the relationship the relationship between natural between (soil consolidation)natural (soil consolidation) and anthropogenic and factorsanthropogenic (ground factors water extraction)(ground water in detail. extraction) The average in detail. subsidence The average rate subsidence map showed rate that map the showed central partthat ofthe the central subsiding part of area, thewhich subsiding is close area, to which Karaokram is close highway to Karaokram and cantonment, highway and including cantonment, major localitiesincluding such major as localities Hassan town, such as Amir Hassan Alam town, Awan Amir town, Alam Mir Awan alam town,town, andMir Jhangialam town, Area, and is sinking Jhangi atArea, a maximum is sinking rate at ofa maximum ~6.5 cm/year, rate varying of ~6.5 fromcm/yea diffr,erent varying rates from up to different 1 cm/year. rates Time up seriesto 1 cm/year. history graphsTime series showed history a close graphs correlation showed with a close the rainfall correlation during with the investigationthe rainfall during period the and investigation subsidence profilesperiod and along subsidence the W–E and profiles S–N transectsalong the showed W–E an varyingd S–N creststransects along showed the transects varying that crests are relatedalong the to thetransects locations that of are tubewells. related to The the overall locations pattern of tu ofbewells. the subsidence The overall was investigated,pattern of the and subsidence it was found was thatinvestigated, it follows and the developmentit was found madethat it forfollows providing the development safe drinking made water for to theproviding community safe drinking under a recentlywater to completed the community project under of JICA. a recently The aquifer completed in this partproject of theof JICA. Abbottabad The aquifer basin in is decreasingthis part of as the a resultAbbottabad of extensive basin wateris decreasing discharge as in a recentresult years,of extensive resulting water in the discharge consolidation in recent of soil, years, further resulting leading in tothe the consolidation creation of severalof soil, further local problems leading to in the the creation area. Building of several dikes local/bowls problems after rainfallin the area. or flood Building are ultimatedikes/bowls responses after torainfall the subsidence or flood processesare ultimate and responses the concerned to the departments subsidence were processes unaware and of thisthe phenomenonconcerned departments in the study were area. unaware This study of highlightsthis phenomenon the first everin the evidence study area. of land This subsidence study highlights and its causesthe first in ever the region,evidence and of theland results subsidence obtained and by its implementing causes in the theregion, advanced and the InSAR results technique obtained can by provide support to the local government as well as decision and policy makers for better decision making and planning to overcome problems such as availability of drinking water, overflowing drains, sewage system, littered roads/streets, and sinking land in this part of the city. Because in situ observations were not available, some decorrelation noises may be present in the final deformation maps, but they have been greatly minimized during 2D regression quality checks at each iteration. Therefore, the LOS deformation rates generated for each point target are basically reliable for mapping the land subsidence in the area, and showed consistency in the results of ascending Remote Sens. 2020, 12, 3442 16 of 18 and descending data sets. Despite our success of monitoring of subsidence in Abbottabad with IPTA, other MT-InSAR methods such as temporary coherent scatterer (TCP) and the Stanford method for persistent scatterers (StaMPS) could also be investigated in order to highlight the performance ability of these techniques in the area. Correlation of InSAR generated results with hazards related to land subsidence using GPS time series data and underground water level measurement can also be included in future research.

Author Contributions: All authors contributed to the manuscript and discussed the results. This research was supervised by X.D. Conceptualization, N.S. and X.D.; Data curation, N.S., H.L., and S.W.; Methodology, N.S. and S.W.; Formal Analysis, N.S. and H.L.; Funding acquisition, X.D. In addition, all authors contributed to the final revision of manuscript. All authors have read and agreed to the published version of the manuscript. Funding: The first author is supported by a Hong Kong Ph.D. Fellowship of the Research Grant Council (RGC). The work was partly supported by the RGC (Grants Numbers: PolyU152164/18E and PolyU152233/19E), The Hong Kong Polytechnic University (project numbers: ZE1F and 1-ZVN6), and Research Institute for Sustainable Urban Development (RISUD) of The Hong Kong Polytechnic University (1-BBWB). Acknowledgments: The authors wish to acknowledge the ESA for releasing and arranging the Sentinel-1A and its POD data. We greatly appreciate the editors and the reviewers for their constructive suggestions and insightful comments that helped us greatly to improve this manuscript. Conflicts of Interest: The authors declare no conflict of interest.

References

1. Zhou, L.; Guo, J.; Hu, J.; Li, J.; Xu, Y.; Pan, Y.; Shi, M. Wuhan surface subsidence analysis in 2015–2016 based on Sentinel-1A data by SBAS-InSAR. Remote Sens. 2017, 9, 982. [CrossRef] 2. Deltares. Sinking Cities: An Integrated Approach Towards Solutions; Deltares—Taskforce Subsidence: Delft/Utrecht, The Netherlands, 2015. 3. Wang, H.; Feng, G.; Xu, B.; Yu, Y.; Li, Z.; Du, Y.; Zhu, J. Deriving spatio-temporal development of ground subsidence due to subway construction and operation in delta regions with PS-InSAR data: A case study in Guangzhou, China. Remote Sens. 2017, 9, 1004. [CrossRef] 4. Ding, X.L.; Liu, G.X.; Li, Z.W.; Li, Z.L.; Chen, Y.Q. Ground subsidence monitoring in Hong Kong with satellite SAR interferometry. Photogramm. Eng. Remote Sens. 2004, 70, 1151–1156. [CrossRef] 5. Higgins, S.A.; Overeem, I.; Steckler, M.S.; Syvitski, J.P.M.; Seeber, L.; Akhter, S.H. InSAR measurements of compaction and subsidence in the Ganges-Brahmaputra Delta, Bangladesh. J. Geophys. Res. Earth Surf. 2014, 119, 1768–1781. [CrossRef] 6. Holzer, T.L.; Johnson, A.I. Land subsidence caused by ground water withdrawal in urban areas. GeoJournal 1985, 11, 245–255. [CrossRef] 7. Schmid, W.; Hanson, R.T.; Leake, S.A.; Hughes, J.D.; Niswonger, R.G. Feedback of land subsidence on the movement and conjunctive use of water resources. Environ. Model. Softw. 2014, 62, 253–270. [CrossRef] 8. Galloway, D.L.; Burbey, T.J. Review: Regional land subsidence accompanying groundwater extraction. Hydrogeol. J. 2011, 19, 1459–1486. [CrossRef] 9. Radutu, A.; Nedelcu, I.; Gogu, C.R. An overview of ground surface displacements generated by groundwater dynamics, revealed by InSAR techniques. Procedia Eng. 2017, 209, 119–126. [CrossRef] 10. Ding, X.L.; Huang, D.F.; Yin, J.H.; Chen, Y.Q.; Lau, C.K.; Yang, Y.; Sun, Y.R.; Chen, W. A new generation of multi-antenna GPS system for landslide and structural deformation monitoring. In Advances in Building Technology; Anson, M., Ko, J.M., Lam, E.S.S., Eds.; Elsevier: Oxford, UK, 2002; pp. 1611–1618, ISBN 978-0-08-044100-9. 11. Lingyun, J.; Qingliang, W.; Shuangxu, W. Present-day 3D deformation field of Northeast China, observed by GPS and leveling. Geod. Geodyn. 2014, 5, 34–40. [CrossRef] 12. Aimaiti, Y.; Yamazaki, F.; Liu, W. Multi-Sensor InSAR Analysis of progressive land subsidence over the coastal city of Urayasu, Japan. Remote Sens. 2018, 10, 1304. [CrossRef] 13. Ikuemonisan, F.E.; Ozebo, V.C.; Olatinsu, O.B. Geostatistical evaluation of spatial variability of land subsidence rates in Lagos, Nigeria. Geod. Geodyn. 2020, 11, 316–327. [CrossRef] 14. Ding, X.L.; Li, Z.; Zhu, J.; Feng, G.; Long, J. Atmospheric effects on InSAR measurements and their mitigation. Sensors 2008, 8, 5426–5448. [CrossRef][PubMed] Remote Sens. 2020, 12, 3442 17 of 18

15. Massonnet, D.; Feigl, K.L. Radar interferometry and its application to changes in the Earth’s surface. Rev. Geophys. 1998, 36, 441–500. [CrossRef] 16. Zebker, H.A.; Villasenor, J. Decorrelation in interferometric radar echoes. IEEE Trans. Geosci. Remote Sens. 1992, 30, 950–959. [CrossRef] 17. Hanssen, R.F. Radar Interferometry: Data Interpretation and Error Analysis (Remote Sensing and Digital Image Processing); van der Meer, F., Ed.; Kluwer Academic Publishers: Norwell, MA, USA, 2001; ISBN 9780792369455. 18. Zhang, B.; Ding, X.L.; Zhu, W.; Wang, C.; Zhang, L.; Liu, Z. Mitigating ionospheric artifacts in coseismic interferogram based on offset field derived from ALOS-PALSAR data. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2016, 9, 3050–3059. [CrossRef] 19. Damoah-Afari, P.; Ding, X.L.; Zhong Lu, Z.L.; Omura, M. Magnitude and extent of six years of land subsidence in shanghai revealed by JERS-1 SAR data. In Geoscience and Remote Sensing New Achievements; Imperatore, P., Riccio, D., Eds.; IntechOpen Limited: London, UK, 2010. 20. Pepe, A.; Calò, F. A review of interferometric synthetic aperture RADAR (InSAR) multi-track approaches for the retrieval of Earth’s surface displacements. Appl. Sci. 2017, 7, 1264. [CrossRef] 21. Zhang, L.; Ding, X.L.; Lu, Z. Ground deformation mapping by fusion of multi-temporal interferometric synthetic aperture radar images: A review. Int. J. Image Data Fusion 2015, 6, 289–313. [CrossRef] 22. Crosetto, M.; Monserrat, O.; Cuevas-González, M.; Devanthéry, N.; Crippa, B. Persistent scatterer interferometry: A review. ISPRS J. Photogramm. Remote Sens. 2016, 115, 78–89. [CrossRef] 23. Hu, J.; Li, Z.W.; Ding, X.L.; Zhu, J.J.; Zhang, L.; Sun, Q. 3D coseismic displacement of 2010 Darfield, New Zealand earthquake estimated from multi-aperture InSAR and D-InSAR measurements. J. Geod. 2012, 86, 1029–1041. [CrossRef] 24. Samsonov, S.V.; Feng, W.; Peltier, A.; Geirsson, H.; d’Oreye, N.; Tiampo, K.F. Multidimensional small baseline subset (MSBAS) for volcano monitoring in two dimensions: Opportunities and challenges. Case study Piton de la Fournaise volcano. J. Volcanol. Geotherm. Res. 2017, 344, 121–138. [CrossRef] 25. Wang, C.; Ding, X.L.; Li, Q.; Shan, X.; Liu, P. Using an integer least squares estimator to connect isolated InSAR fringes in earthquake slip inversion. IEEE Trans. Geosci. Remote Sens. 2016, 54, 2899–2910. [CrossRef] 26. Guangcai, F.; Zhiwei, L.; Xinjian, S.; Bing, X.; Yanan, D. Source parameters of the 2014 Mw 6.1 South Napa earthquake estimated from the Sentinel 1A, COSMO-SkyMed and GPS data. Tectonophysics 2015, 655, 139–146. [CrossRef] 27. San-Ming, L.U.O.; Xin-Jian, S.; Wen-Wu, Z.H.U.; Kai-Fu, D.U.; Wen-Ni, W.A.N.; Hong-Bao, L.; Zhi-Guang, L.I.U. Monitoring vertical ground deformation in the North China Plain using the multitrack PSInSAR technique. Chin. J. Geophys. 2014, 57, 2139–3129. [CrossRef] 28. Zhang, B.; Zhang, L.; Yang, H.; Zhang, Z.; Tao, J. Subsidence prediction and susceptibility zonation for collapse above goaf with thick alluvial cover: A case study of the Yongcheng coalfield, Henan Province, China. Bull. Eng. Geol. Environ. 2016, 75, 1117–1132. [CrossRef] 29. Wegmuller, U.; Walter, D.; Spreckels, V.; Werner, C.L.; Wegmüller, U.; Walter, D.; Spreckels, V.; Werner, C.L. Nonuniform ground motion monitoring with TerraSAR-X persistent scatterer interferometry. IEEE Trans. Geosci. Remote Sens. 2010, 48, 895–904. [CrossRef] 30. Mustafa, U.; Haq, M.; Ahmad, I. Environmental Fiscal Reform in Abbottabad: Drinking Water; Pastakia, F., Ed.; International Union for Conservation of Nature (IUCN): Islamabad, Pakistan, 2009. 31. Hussain, Z.; Khan, S.; Khan, M.A.; Waseem, M.; Ahmed, W. Mapping sediment thickness of the Abbottabad basin, Pakistan. J. Earth Syst. Sci. 2018, 128, 7. [CrossRef] 32. Pakistan Bureau of Statistics. Available online: http://www.pbs.gov.pk/ (accessed on 15 April 2020). 33. Lasaponara, R.; Masini, N. Big Earth data for cultural heritage in the Copernicus era. In Remote Sensing for Archaeology and Cultural Landscapes: Best Practices and Perspectives Across Europe and the Middle East; Hadjimitsis, D.G., Themistocleous, K., Cuca, B., Agapiou, A., Lysandrou, V., Lasaponara, R., Masini, N., Schreier, G., Eds.; Springer International Publishing: Cham, Switzerland, 2020; pp. 31–46, ISBN 978-3-030-10979-0. 34. Chen, Q.; Ding, X.L.; Liu, G.X. Method for optimum selection of common master acquisition for PS-InSAR. Acta Geod. Cartogr. Sin. 2007, 36, 395–399. 35. Mora, O.; Mallorqui, J.J.; Broquetas, A. Linear and nonlinear terrain deformation maps from a reduced set of interferometric SAR images. IEEE Trans. Geosci. Remote Sens. 2003, 41, 2243–2253. [CrossRef] Remote Sens. 2020, 12, 3442 18 of 18

36. Werner, C.; Wegmuller, U.; Wiesmann, A.; Strozzi, T. Interferometric Point Target Analysis with JERS-1 L-band SAR data. In Proceedings of the IGARSS 2003. 2003 IEEE International Geoscience and Remote Sensing Symposium. Proceedings (IEEE Cat. No.03CH37477), Toulouse, France, 21–25 July 2003; Volume 7, pp. 4359–4361. 37. Costantini, M. A novel phase unwrapping method based on network programming. IEEE Trans. Geosci. Remote Sens. 1998, 36, 813–821. [CrossRef] 38. Rabus, B.; Werner, C.; Wegmueller, U.; McCardle, A. Interferometric point target analysis of RADARSAT-1 data for deformation monitoring at the Belridge/Lost Hills oil fields. In Proceedings of the International Geoscience and Remote Sensing Symposium (IGARSS), Anchorage, AK, USA, 20–24 September 2004. 39. Werner, C.L.; Wegmüller, U.; Strozzi, T. Deformation time-series of the lost-hills oil field using a multi-baseline interferometric SAR inversion algorithm with finite difference smoothing constraints. In Proceedings of the AGU Fall Meeting Abstracts, San Francisco, CA, USA, 3–7 December 2012. 40. Wright, T.J.; Parsons, B.E.; Lu, Z. Toward mapping surface deformation in three dimensions using InSAR. Geophys. Res. Lett. 2004, 31.[CrossRef] 41. Samiei-Esfahany, S.; Hanssen, R.F.; Van Thienen-visser, K.; Muntendam-bos, A. On the effect of horizontal deformation on InSAR subsidence estimates. In Proceedings of the Fringe 2009, Frascati, Italy, 30 November–4 December 2009. 42. Aslan, G.; Cakir, Z.; Lasserre, C.; Renard, F. Investigating subsidence in the Bursa Plain, Turkey, using ascending and descending Sentinel-1 Satellite Data. Remote Sens. 2019, 11, 85. [CrossRef] 43. Cui, Z.D.; Yang, J.Q.; Yuan, L. Land subsidence caused by the interaction of high-rise buildings in soft soil areas. Nat. Hazards 2015, 79, 1199–1217. [CrossRef] 44. GSP. Geological Map of Abbottabad area NWFP, Pakistan. Map series Vol. III at 1:50,000 scale, Issued by Director General. Geological Survey of Pakistan (GSP). 2002. Available online: https://www.gsp.gov.pk/ (accessed on 19 October 2020). 45. Japan International Cooperation Agency (JICA). Basic Design Study Report on the Project for Improvement of the Water Supply in Abbottabad, Pakistan; Japan International Cooperation Agency: Tokyo, Japan, 2004. 46. Japan International Cooperation Agency (JICA). Preparatory Survey Report on the Project for the Improvement of Water Supply System in Abbottabad, Pakistan; Japan International Cooperation Agency: Tokyo, Japan, 2009. 47. Japan International Cooperation Agency (JICA). Ex-Post Project Evaluation 2017: Package IV—5 (Pakistan, Bangladesh); Japan International Cooperation Agency: Tokyo, Japan, 2018. 48. Recorder, B. Rain Continues to Paralysis Life in Hazara. Available online: https://fp.brecorder.com/2018/08/ 20180809397986/ (accessed on 20 April 2020). 49. Ahmed, T.; Pervez, A.; Mehtab, M.; Sherwani, S.K. Assessment of drinking water quality and its potential health impacts in academic institutions of Abbottabad (Pakistan). Desalin. Water Treat. 2015, 54, 1819–1828. [CrossRef] 50. Japan Gifts Water Supply System to Abbottabad. Available online: https://www.jica.go.jp/pakistan/english/ office/topics/press151015.html (accessed on 10 May 2020). 51. Zhang, Y.; Liu, Y.; Jin, M.; Jing, Y.; Liu, Y.; Liu, Y.; Sun, W.; Wei, J.; Chen, Y. Monitoring land subsidence in Wuhan city (China) using the SBAS-INSAR method with radarsat-2 imagery data. Sensors (Switzerland) 2019, 19, 743. [CrossRef][PubMed] 52. Galloway, D.L.; Coplin, L.S.; Ingebritsen, S.E. Effects of land subsidence in the Greater Houston Area. In Managing Urban Water Supply; Agthe, D.E., Billings, R.B., Buras, N., Eds.; Springer: Dordrecht, The Netherlands, 2003; pp. 187–203, ISBN 978-94-017-0237-9.

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).