Ground Deformation Analysis of Bolvadin (W

remote sensing

Article Ground Deformation Analysis of Bolvadin (W. Turkey) by Means of Multi-Temporal InSAR Techniques and Sentinel-1 Data

Mumin Imamoglu 1,2,* , Fatih Kahraman 1 , Ziyadin Cakir 3 and Fusun Balik Sanli 2 1 Informatics and Information Security Research Center, TUBITAK, 41470 Gebze, Turkey; [email protected] 2 Department of Geomatic Engineering, Civil Engineering Faculty, Yildiz Technical University, 34220 Istanbul, Turkey; [email protected] 3 Department of Geological Engineering, Istanbul Technical University, 34469 Istanbul, Turkey; [email protected] * Correspondence: [email protected]; Tel.: +90-262-675-3279

Received: 26 February 2019; Accepted: 22 April 2019; Published: 6 May 2019

Abstract: Surface deformations were observed in Bolvadin, located in the province of Afyon (western Turkey) in the last decade which occurred without any destructive earthquakes. In this study, ground deformation of the Bolvadin region is analyzed by means of multi-temporal interferometric synthetic aperture radar (InSAR) technique with Sentinel-1 synthetic aperture radar (SAR) data. Sentinel-1 data acquired in ascending and descending orbits between October 2014 and October 2018 are processed with the Sentinel Application Platform (SNAP) and Stanford Method for Persistent Scatterers (StaMPS) open source software tools. Deformation velocity maps and line-of-sight (LOS) displacement time series are produced and compared with geology, groundwater level and the water surface area of Eber Lake nearby. Deformation velocity maps reveal significant subsidence in most of the town and surrounding regions, which is confirmed by field observations that show severe damage to the settlements and infrastructure of the town. The most severe subsidence is observed to be in the southern part of Bolvadin with rates up to 35 mm/year, which is characterized by the presence of soft alluvial deposits. Composed of slope debris/talus and conglomerate, the northeastern part of the deforming region experiences a relatively lower rate of subsidence. A strong correlation between LOS displacement time series and groundwater level exists both in the short and long term. Moreover, short term variations in LOS displacement time series also show high similarity with seasonal variations in the water surface area of Eber Lake located a few km southeast of the town. We conclude that the primary cause of subsidence is most probably the overexploitation of groundwater and hydrological changes because of (1) the strong correlation of subsidence with lithological units, (2) the similarity between deformation rate and groundwater level changes, (3) the correspondence of seasonal variations in water surface area and short-term deformation rate oscillations, and (4) the absence of InSAR velocity contrast across the active faults.

Keywords: multi-temporal InSAR; time series analysis; sentinel-1; land subsidence; Bolvadin

1. Introduction Land subsidence is a common geological hazard that aﬀects many regions in the world. Depending on the nature of the region and the man-made activities around it, subsidence phenomena may be triggered by anthropogenic activities such as excessive use of groundwater, mining activities and underground construction or by natural processes such as tectonic movements [1]. Land subsidence may result in infrastructure damage, ground surface ruptures, increasing ﬂood risk and adverse

Remote Sens. 2019, 11, 1069; doi:10.3390/rs11091069 www.mdpi.com/journal/remotesensing Remote Sens. 2019, 11, 1069 2 of 17 socio-economic impact in the community [2,3]. Therefore, detecting spatial extent of the deformation pattern and monitoring its temporal evolution is crucial to determine the primary cause of subsidence and conduct preventive activities accordingly to mitigate the negative effects [4]. Over the last two decades, Bolvadin in the province of Afyon (central west of Turkey) has been severely affected by ground deformation such as subsidence, fracturing and fissuring causing severe damage to the settlements and infrastructure. A field study to the region was organized on 12 and 13 February 2019 to observe such deformations in place. Previously reported deformation regions [5] and other potentially damaged areas were visited. Some of the observed surface deformations and their reflections on the buildings are shown in Figure1a–f. The visited regions and the location of each photograph taken are shown in Figure2b. Fractures and fissures are clearly seen on the ground in the regions a and c Although repaving has removed the effects of deformation on the roads in the regions b and d, stone walls and roadside curbs along them reveals the evidence of deformation. No clear surface deformation was observed in the regions e and f. However, the presence of quite a few cracks andRemote openings Sens. 2019 on,the 11, x wall FOR PEER of the REVIEW buildings were observed. 2 of 17

FigureFigure 1. Ground 1 Ground deformations deformations observed observed in Bolvadin and and their their effects eﬀects on on land, land, roads roads and and structures structures.. of the deformation pattern and monitoring its temporal evolution is crucial to determine the primary cause of subsidence and conduct preventive activities accordingly to mitigate the negative effects [4]. Over the last two decades, Bolvadin in the province of Afyon (central west of Turkey) has been severely affected by ground deformation such as subsidence, fracturing and fissuring causing severe damage to the settlements and infrastructure. A field study to the region was organized on 12 and 13 February 2019 to observe such deformations in place. Previously reported deformation regions [5] and other potentially damaged areas were visited. Some of the observed surface deformations and their reflections on the buildings are shown in Figure 1a–f. The visited regions and the location of each photograph taken are shown in Figure 2b. Fractures and fissures are clearly seen on the ground in the regions a and c Although repaving has removed the effects of deformation on the roads in the regions b and d, stone walls and roadside curbs along them reveals the evidence of deformation. No clear surface deformation was observed in the regions e and f. However, the presence of quite a few cracks and openings on the wall of the buildings were observed.

Remote Sens. 2019, 11, 1069 3 of 17 Remote Sens. 2019, 11, x FOR PEER REVIEW 4 of 17

(a)

(b) (c)

Figure 2.2 (a()a Study) Study area area (Bolvadin, (Bolvadin, western western Turkey) Turkey and surrounding) and surrounding regions with region shadeds wi topography,th shaded activetopography fault lines, active and Sentinel-1 fault lines footprints. and Sentinel (b) Visited-1 footprints regions and. (b location) Visited of photographs regions and on location Sentinel-2 of truephotographs color view. on (Sentinelc) Geological-2 true map color of view. Bolvadin (c) Geological taken from map MTA. of Bolvadin taken from MTA.

Bolvadin is located in the middle part of the Afyon-Akşehir Graben system (Figure 2), where the tectonic formation is extremely complex with frequent seismic events [6,7]. In this work, the active fault database for Afyon-Bolvadin (Figure 2) was taken from the renewed active fault map of

Remote Sens. 2019, 11, 1069 4 of 17

This kind of aseismic surface deformation is thought to result from a drastic decrease in ground water table due to excessive pumping or/and drought [5] or creep on nearby active faults [6]. In order to investigate the cause of the surface deformation in the study area, excavations were carried out by [5] in two different areas across fractures that display vertical offsets. Observations on the excavated trench walls reveal that the depths of vertical offsets of 20–30 cm are too shallow (down to 5 m) and offset rate is too high (1.5 to 3 cm/year) to be of tectonic origin, that is, surface creep on active faults buried by the Quaternary alluvium. Consequently, [5] conclude that surface deformations are caused by hydrological changes, not by fault creep or other tectonic motions. However, considering the presence of a complex active fault network [7] and their geodetic observations, [6] suggest that all the surface deformation cannot be explained by ground water table drops only and tectonic creep should also be partly responsible. In order to examine the seasonal behavior of these deformations, a continuous Global Positioning System (GPS) station was established by [6] in January 2015, and collected data for 2 years. For this specific point, the vertical subsidence rate was found to be 7.1 cm/year. In addition to GPS, a leveling network was established by [8] in August 2016. The network was reoccupied in May 2017. Between 18–24 mm of subsidence was observed during 8 months along 9 profiles [8]. Although ground-based tracking techniques, such as GPS, and leveling measurements provide high-precision measurements, they are obtained at predetermined scares points only. They cannot provide detailed and comprehensive information in the spatial domain. Furthermore, operating these techniques in the field is costly in terms of time and labor [9]. Interferometric synthetic aperture radar (InSAR) is a powerful technique that measures ground surface deformation over a wide area with high precision [10,11]. Satellite-based SAR systems provide very high spatial and high temporal resolution (4–6 days nowadays) over a large area at all weather conditions, day and night. Cost of the InSAR technique, in terms of measurement points per square kilometer, is much lower than other deformation measurement techniques such as GPS and leveling [12]. Advanced InSAR techniques enable time series analysis of multi-temporal SAR data and provide a good set of tools for ground deformation measurements with millimeter precision [13–15]. Previous studies used multi-temporal InSAR technique for analysis of land subsidence due to over exploitation of groundwater [16–20], urban area deformation [21–24], mining region stability [25–27], landslide phenomena [28–30], sinkhole detection [31,32] and seismic events [33,34]. The aim of this study is to understand the mechanism of ground deformation in the study area. To do this, we use multi-temporal InSAR technique with Sentinel-1 TOPSAR data. Sentinel-1 products that were acquired in both ascending and descending orbits between October 2014 and October 2018 were processed using Persistent Scatterer Interferometry (PSI) technique. The velocity fields obtained with PSI are compared with geology, groundwater level and water surface area of Eber Lake nearby.

2. Geographic and Geologic Setting of the Study Area The study area, namely Bolvadin, is located on the eastern part of the Aegean sub-region of Turkey. Bolvadin is the largest town in the province of Afyon with a population of over 30 thousand. It is built on a wide fertile plain between Sultanda˘gand Emirda˘gmountains on the Aegean coast. The plain is a rich agricultural and farming area watered by the river Akarçay and Lake Eber. Eber Lake is an important water source for the region, positively aﬀecting economic and agricultural activities. However, the water surface area of Eber Lake varies from year to year due to the intense use of water in irrigated agricultural areas, climatic changes and newly built dams in the region [35–37]. Historically, groundwater has been the primary source of water for irrigation and industrial use in the Bolvadin Region. Due to the soft alluvial sediments surrounding the Akarçay River and Lake Eber, many subsidence phenomena aﬀect the area, making it potentially dangerous for civil infrastructures for the study area [5,6]. Bolvadin is located in the middle part of the Afyon-Ak¸sehirGraben system (Figure2), where the tectonic formation is extremely complex with frequent seismic events [6,7]. In this work, the active Remote Sens. 2019, 11, 1069 5 of 17 faultRemote database Sens. 2019 for, 11 Afyon-Bolvadin, x FOR PEER REVIEW (Figure 2) was taken from the renewed active fault map of Turkey5 of 17 prepared by [7] under the establishment of the Geological Research Department of General Directorate ofTurkey Mineral prepared Research by and [7] Explorationunder the establishment (MTA) of Turkey. of the Geological Research Department of General Directorate of Mineral Research and Exploration (MTA) of Turkey. 3. Materials and Methods 3. Materials and Methods 3.1. Sentinel-1 Dataset 3.1. Sentinel-1 Dataset Sentinel-1 is a twin satellite constellation that carries synthetic aperture radar (SAR) sensors operatingSentinel at C-band-1 is a (~5.6twin cmsatellite wavelength) constellation under that Copernicus carries synthetic Programme aperture coordinated radar (SAR and) managedsensors byoperating the European at C-band Space (~5.6 Agency cm wavelength (ESA). Sentinel-1) under products Copernicus are P distributedrogramme bycoordinated the Copernicus and managed program underby the free European and open dataSpace policy. Agency Sentinel-1 (ESA). ASentinel and Sentinel-1-1 products B together are distrib haveuted imaging by the capability Copernicus of the program under free and open data policy. Sentinel-1 A and Sentinel-1 B together have imaging same region every 6 days at all weather conditions, day and night. Six-day repeat cycle is quite a good capability of the same region every 6 days at all weather conditions, day and night. Six-day repeat temporal resolution for land monitoring services. cycle is quite a good temporal resolution for land monitoring services. The study area is analyzed for ground surface deformation using Sentinel-1 SAR images within The study area is analyzed for ground surface deformation using Sentinel-1 SAR images within the time period between October 2014 and October 2018. Sentinel-1 A covers the study region from the time period between October 2014 and October 2018. Sentinel-1 A covers the study region from the beginning of the analysis period and Sentinel-1 B starting from October 2016 at ascending track the beginning of the analysis period and Sentinel-1 B starting from October 2016 at ascending track 160160 and and at at descending descending track track 65 65 as as shown shown inin FigureFigure2 2a.a. WithWith thethe launch of Sentinel Sentinel-1-1 B B in in mid mid 2016 2016,, temporaltemporal resolution resolution increased increased at at the the second second half half ofof thethe analysisanalysis period from from 12 12 days days to to 6 6 days. days. A A total total ofof 366 366 (183 (183 ascending ascending and and 183 183 descending) descending) productsproducts acquired in in TOPSAR TOPSAR Interferometric Interferometric Wide Wide (IW) (IW) SwathSwath Mode Mode in in Single Single Look Look Complex Complex (SLC) (SLC) formatformat areare processedprocessed independently for for each each track track.. The The singlesingle master master image image is is selected selected for for each eachdataset dataset separatelyseparately by considering temporal temporal and and geometric geometric baselinesbaselines of of the the Sentinel-1 Sentinel-1 A A/B/B products products asas displayeddisplayed inin Figure 33.. MeanMean incidence angle angle of of the the study study areaarea is is about about 35 35°◦ for for ascending ascending dataset dataset and and isis aroundaround 40.540.5°◦ for descending dataset. dataset.

(a) (b)

FigureFigure 3. 3.Baseline Baseline plotplot forfor SentinelSentinel 11 data.data. Red star represents the master image of of all all the the pairs pairs.. ( (aa)) Asc. Asc. tracktrack 160, 160,master master imageimage date is 5 May, 2017. ( (bb)) Desc. Desc. track track 65 65,, master master image image date date is is2 July, 2 July, 2017. 2017.

3.2.3.2. Multi-Temporal Multi-Temporal InSAR InSAR Processing Processing GroundGround deformation deformation in in Bolvadin Bolvadin isisanalyzed analyzed usingusing the PSI technique. Instead Instead of of analyzing analyzing the the wholewhole image, image, only only persistent persistent scatterers scatterers (PS) (PS) points points which which are are stable stable over over time time are analyzed.are analyzed The. The main processingmain processing steps of PSI steps consist of PSI of interferogram consist of interferogram generation, multi-temporalgeneration, multi persistent-temporal scatterers persistent (PS) processingscatterers and (PS) removal processing of atmospheric and removal phase of atmospheric screen (APS) phase [14]. screen The processing (APS) [14]. of The each processing main block of is carriedeach main out as block suggested is carried in [ 21out]. as Single suggested master in interferograms [21]. Single master are generated interferogram usings are Sentinel generated Application using PlatformSentinel (SNAP, Application version Platform 6.0) [38 ].(SNAP, Initially, version data acquired 6.0) [38]. before Initially, March data 2015 acquired are corrected before March for Elevation 2015 Antennaare corrected Pattern for (EAP) Elevation to obtain Antenna appropriate Pattern interferograms.(EAP) to obtain EAPappropriate used for interferograms Sentinel-1 data. EAP upgraded used tofor correct Sentinel the EAP-1 data phase upgraded in addition to correct to the the EAP EAP gain phase in March in addition 2015. to It isthe reported EAP gain that in thisMarch change 2015. may It causeis reported range varying that this phase change oﬀ setmay while cause generating range varying interferogram phase offset with while products generating acquired interferogram before and afterwith [39 products]. After proper acquired sub-swath before and burst after selection,[39]. After all proper Sentinel-1 sub-swath slave images and burst are coregisteredselection, all to a masterSentinel image-1 slave with images Sentinel-1 are coregistered Back Geocoding to a master operator image usingwith Sentinel orbit ﬁles-1 Back [40] andGeocoding digital operator elevation using orbit files [40] and digital elevation model (DEM). Then, interferograms are formed and the topographic phase component is removed. Shuttle Radar Topography Mission (SRTM) [41] with 1

Remote Sens. 2019, 11, 1069 6 of 17 model (DEM). Then, interferograms are formed and the topographic phase component is removed. Shuttle Radar Topography Mission (SRTM) [41] with 1 arc-second (~30 m pixel size) DEM is used for the coregistration and for topographic phase removal from the interferometric phase. VV polarization bands are processed since co-polarized bands achieve higher coherency. The stack of interferograms is passed to the next stage for time series analysis. The Stanford Method for Persistent Scatterers (StaMPS, version 4.1-beta) software package [13,42] is used for selection of PS points within the stack of interferograms and for multi-temporal processing of them. Candidate PS points are initially selected by using the amplitude dispersion index originally proposed by [43] with a relatively higher threshold. Then, spatially-correlated terms are estimated using phase analysis and subtracted from the interferometric phase to obtain more stable PS points that not only depend on signal intensity but also phase stability. In the last stage, the Toolbox for Reducing Atmospheric InSAR Noise (TRAIN) [44] is used to remove atmospheric phase screen. The ERA-I global weather model is used to compute tropospheric phase eﬀects caused by temperature, pressure and humidity variations. Lastly, tropospheric phase delays for each acquisition date are removed from the interferometric phase to obtain the ﬁnal deformation signal.

4. Results

4.1. Multi-Temporal InSAR Results As a result of multi-temporal InSAR analysis, two main products are produced; (1) mean deformation velocity maps in line-of-sight (LOS) direction, and (2) time-series of displacements for measurement points that are provided in the following subsections.

4.1.1. Mean Velocity Maps The mean deformation velocity maps obtained in LOS by multi-temporal analysis of ascending and descending datasets within the time period of October 2014 and October 2018 are shown in Figure4. All measurement values are relative to a common reference point that is considered to be stable over the entire time period. Warm colors (negative values) highlight displacement that occurred away from the sensor while cold colors (positive values) show movement towards the sensor. Although results are displayed in the range of 25 mm/year for visualization purposes, larger displacement rates are present ± at some locations. The velocity maps of the entire study area in Figure4a,b show that subsidence is taking place not only in the town of Bolvadin, but also in almost all parts of the Afyon-Ak¸sehirGraben, including the surroundings of Çay, Çobanlar and Sultanda˘gıtowns. The majority of the PS points is found to be around the urbanized area. There are almost no measurement points over agricultural areas located around the settlement since great variations in the farmlands are expected in such a long analysis period. However, it is quite clear that the entire graben is exposed to severe subsidence. Figure4c,d provides mean velocity maps focused on the Bolvadin region. Ascending and descending velocity maps show quite a similar deformation pattern and rates at most of Bolvadin, which is a sign of pure vertical movement. However, there are some areas like the southern part of the city that LOS deformation is different for ascending and descending dataset. That might be an indication of horizontal movement. For that reason, further processing and decomposition of LOS displacement into vertical and horizontal components is required for easier interpretation of movement. Only the projection of the actual displacement in the LOS direction can be detected by the SAR sensors. Another limitation of deformation measurement is the low sensitivity of SAR sensors to displacement occurred in the N-S direction [45]. In order to obtain 3D movement component, at least 3 different measurements are required obtained by different acquisition geometry. However, the N-S component can be neglected due to the low sensitivity of SAR sensors along that direction. Finally, availability of images obtained at ascending and descending acquisition geometry is enough to calculate vertical and horizontal (E-W) displacement components. Using the same decomposition method outlined in [46] with neglecting the N-S component, vertical and E-W horizontal mean deformation Remote Sens. 2019, 11, 1069 7 of 17 velocity maps are obtained for the Bolvadin region (Figure5). In almost every region of Bolvadin, varying rates of subsidence are observed. The maximum vertical subsidence is observed to be around 35 mm/year in the southern part of Bolvadin. North and north-east parts shows relatively smaller deformation rate compared to south and south-west of the city. Although horizontal E-W movement is negligible in most regions, relatively small movement is observed in a small area in the west and specificRemote Sens. region 2019, in11, thex FOR south. PEER REVIEW 7 of 17

(a) (b)

FigureFigure 4.4 Mean Sentinel LOS velocity maps on ascending track 160 (a) and descending track 65 ((b)) inin centralcentral Afyon-Ak¸sehirgraben.Afyon-Akşehir graben. CloseClose up up view view of of Bolvadin Bolvadin region region on on ascending ascending (c) ( andc) and descending descending (d) tracks.(d) tracks P1,. P2P1, andP2 and P3 pointsP3 points are usedare used in time in time series series analysis analysis shown shown in Figure in Fig 6.ure 6. be around 35 mm/year in the southern part of Bolvadin. North and north-east parts shows relatively smaller deformation rate compared to south and south-west of the city. Although horizontal E-W movement is negligible in most regions, relatively small movement is observed in a small area in the west and specific region in the south.

4.1.2 Time Series Three measurement points are selected from different areas of the city that are exposed to different deformation rates. The points are selected as close as possible to the photo locations that

Remote Sens. 2019, 11, 1069 8 of 17 Remote Sens. 2019, 11, x FOR PEER REVIEW 8 of 17

(a) (b)

FigureFigure 5. 5 DecompositionDecomposition of LOS velocities into vertical and horizontal (E (E-W)-W) components components.. Dashed Dashed red-whitered-white lines lines show show possiblepossible extensionextension of the Bolvadin fault fault (solid (solid red red line) line).. Deformation Deformation rates rates over over profilesprofiles AA’ AA’ to DD’to DD’ are areshown shown in Figure in Figure 8. (a ) 8. Vertical (a) Vertical mean deformation mean deform velocity.ation velocity. Warm colorsWarm (negative colors values)(negative represent values)subsidence. represent subsidence. (b) Horizontal (b) Horizontal E-W mean E- deformationW mean deformation velocity. velocity. Warm colors Warm (negative colors values)(negative represent values) movements represent movements towards west towards direction west and direction cold colors and (positive cold colors values) (positive towards values) east. towards east. 4.1.2. Time Series were taken during the field study. The locations of the selected points are displayed in Figure 4c,d. The Three point measurementP1 is selected points from arethe selectedsouthern from part di offferent the town areas at of athe location city that where are exposed the most to di severefferent deformationsubsidence is rates. observed The. points P2 and are P3 selected points are as selected close as from possible two todifferent the photo regions locations that show that were different taken duringsubsidence the field rate study.s. The Thepoint locations P2 is located of the in selected the western points pa arert of displayed the city, and in Figure the point4c,d. P3 The is located point P1 in is selectedthe eastern from part the southernof the city part where of the relatively town at low a location deformation where therates most are severeobserved. subsidence Figures is6a observed.–c and P2Figure and P3s 6d points–f show are ascending selected from and twodescending different LOS regions time that-series show at all di ffselectederent subsidence points, respectively. rates. The The point P2mean is located of all in points the western on a radiuspart of ofthe 100 city, m and is the calculated point P3 as is locateda LOS displacement.in the eastern partThe oforiginal the city wheredisplacement relatively values low deformation are shown together rates are with observed. the linear Figure fit result6a–f show. ascending and descending LOS time-seriesIn the at long all selectedterm, both points, dataset respectively.s indicate secular The mean movements of all points away on from a radius the ofsensor 100 m at isall calculated points, aswhi a LOSch is displacement. an indication Theof subsidence. original displacement While observed values cumulative are shown displacement together with at the P1 linearin ascending fit result. LOSIn is theapproximately long term, both8 cm, datasets it is about indicate 12 cm in secular the descending movements dataset away. The from difference the sensor of cumulative at all points, whichmeasured is an displacement indication of in subsidence. LOS between While the observedascending cumulative and descend displacementing datasets atcan P1 be in explained ascending LOSwith is the approximately existence of 8horizontal cm, it is aboutmovement 12 cms inin thethat descending region as shown dataset. in TheFigure diff 5erenceb. LOS of cumulative cumulative measureddisplacement displacements at P2 decrease in LOS to about between 7 cm the for ascending both dataset ands. P3 descending is a relatively datasets more stable can be point explained that withshows the about existence 2 cm of of horizontal LOS cumulative movements displacement in that regionin both as datasets shown. inSimilar Figure displacement5b. LOS cumulative values and patterns at P1 and P2 are indicative of pure vertical deformation. Additionally, short-term displacements at P2 decrease to about 7 cm for both datasets. P3 is a relatively more stable point oscillations are observed in all selected points. As the general deformation trend decreases, the that shows about 2 cm of LOS cumulative displacement in both datasets. Similar displacement amount of oscillation decreases from P1 to P3. Effects of seasonal variations on oscillation will be values and patterns at P1 and P2 are indicative of pure vertical deformation. Additionally, short-term analyzed in details in the following subsections. oscillations are observed in all selected points. As the general deformation trend decreases, the amount In order to illustrate how the rate of deformation changes between two different regions during of oscillation decreases from P1 to P3. Effects of seasonal variations on oscillation will be analyzed in the analysis period, all displacement time series along the profile line between the points P2 and P3 details in the following subsections. are combined. For the sake of simplicity, only the linear approximations of LOS displacements are shown in Figure 6 c,d for ascending and descending datasets. Obviously, all measurement points over the P2-P3 line are exposed to subsidence. However, there has been an increase in subsidence

Remote Sens. 2019, 11, 1069 9 of 17 Remote Sens. 2019, 11, x FOR PEER REVIEW 9 of 17

(a) (b) (c)

(d) (e) (f)

(g) (h)

FigureFigure 6. 6LOS LOS displacement displacement time time series series at selectedat selected points points are shownare shown together together with linear with fit linear for Asc.-160 fit for (aAsc.) P1-160 (b) P2(a) (P1c) P3(b) and P2 Desc.-65(c) P3 and (d Desc.) P1 (e-65) P2 (d ()f )P1 P3. (e Linear) P2 (f) fitP3 of. Linear LOS displacement fit of LOS displacement time series ontime the profileseries on between the profile points between P2 and points P3 are P2 displayed-for and P3 are displayed (g) Asc.-160-for ( andg) Asc. (h)-160 Des.-65, and (h tracks.) Des.-65, Points tracks. and profilePointslocations and profile are locations shown in are Figure shown4c,d. in Figure 4c,d. over the years from point P3 to P2. While about 2 cm of cumulative displacement is observed at the In order to illustrate how the rate of deformation changes between two different regions during thepoint analysis P3, there period, is a allpproximately displacement 7 c timem of series subsidence along theat the profile point line P2 between in LOS thedirection. points P2Cumulative and P3 are combined.displacement For thevalues sake in of LOS simplicity, correlate only wi theth linear ascending approximations and descending of LOS datasetdisplacements since negligible are shown horizontal E-W movement is observed in the region. in Figure6c,d for ascending and descending datasets. Obviously, all measurement points over the P2-P3 line are exposed to subsidence. However, there has been an increase in subsidence over the years 4.2. Comparison with Lithology from point P3 to P2. While about 2 cm of cumulative displacement is observed at the point P3, there is approximatelyAs previously 7 cm ofstated subsidence, Bolvadin at the is point located P2 in in LOS the direction.central part Cumulative of the Afyon displacement-Akşehir valuesGraben in LOSsystem correlate, and mainly with ascending characterized and by descending the existence dataset of alluvial since negligiblesediments. horizontal The lithological E-W movement units in the is observednorth and in the the north region.-east area of Bolvadin are composed of slope debris/talus and conglomerates. The outer regions of the town also contain limestones and schists. Mean deformation velocity maps 4.2.and Comparison geological withboundaries Lithology provided in Figure 5 show a clear correlation with alluvial sediments and high deformation rates that exist in the southern and western part of the city. The LOS cumulative As previously stated, Bolvadin is located in the central part of the Afyon-Ak¸sehirGraben system, displacement time series of points P1 and P2 are selected from these regions and results are shown in and mainly characterized by the existence of alluvial sediments. The lithological units in the north Figure 6. Slope debris/and conglomerate are made of coarse grains and blocks and thus are less and the north-east area of Bolvadin are composed of slope debris/talus and conglomerates. The deformable sediments where a decrease in deformation rate could be observed as in the case of the outer regions of the town also contain limestones and schists. Mean deformation velocity maps and point P3. Displacement time series along the profile line between points P2 and P3 (Figure 6 c,d) geological boundaries provided in Figure5 show a clear correlation with alluvial sediments and shows a clear distinction in which the rate of deformation changes between alluvial sediments and high deformation rates that exist in the southern and western part of the city. The LOS cumulative slope debris/talus. The most stable area with almost zero deformation velocity is in the displacement time series of points P1 and P2 are selected from these regions and results are shown north-western part of the region where schist formation is observed. The reference point is selected in Figure6. Slope debris /and conglomerate are made of coarse grains and blocks and thus are less from that stable region and all other measurement points are updated accordingly. Therefore, the deformable sediments where a decrease in deformation rate could be observed as in the case of the velocity fields are relative to the reference point in this region. point P3. Displacement time series along the profile line between points P2 and P3 (Figure6c,d) shows The correlation between geological units and vertical and horizontal (E-W) deformation can be a clear distinction in which the rate of deformation changes between alluvial sediments and slope seen in Figure 7. The mean, standard deviation, and minimum and maximum deformation rates of debriseach / geologicaltalus. The most unit stableare calculated. area with As almost expected, zero deformation while most velocity of the is deformation in the north-western occurs in part soft of thealluvium region deposits, where schist the formationold geological is observed. formations The are reference much less point affected. is selected The maximum from that stablesubsidence region rate is observed to be around 35 mm/year in the alluvial deposits in the southern part of Bolvadin.

Remote Sens. 2019, 11, 1069 10 of 17 and all other measurement points are updated accordingly. Therefore, the velocity ﬁelds are relative to the reference point in this region. The correlation between geological units and vertical and horizontal (E-W) deformation can be seen in Figure7. The mean, standard deviation, and minimum and maximum deformation rates of each geological unit are calculated. As expected, while most of the deformation occurs in soft alluvium deposits, the old geological formations are much less aﬀected. The maximum subsidence rate is observed to be around 35 mm/year in the alluvial deposits in the southern part of Bolvadin. While average subsidence rate is about 15 mm/year in alluvium deposits, other geological units exhibit very little or no deformation. Although mean subsidence rate of slope debris/talus is about 7 mm/year, the maximum subsidence rate reaches up to 25 mm/year. On the other hand, mean horizontal (E-W) deformation velocity is almost zero for all geological units. However, especially in alluvial sediments, Remote Sens. 2019, 11, x FOR PEER REVIEW 10 of 17 both east and west movements are observed in some regions with the maximum rate about 12 mm/year.

(a) (b)

FigureFigure 7.7 Relationship between between geological geological units units in in the Bolvadin region and ( a) Vertical deformationdeformation velocity,velocity, ( b(b)) Horizontal Horizontal (E-W) (E-W) deformation deformation velocity. velocity The. The mean mean (red (red dots), dots), standard standard deviation deviation (blue (blue line), minimumline), minimum and maximum and maximum (bold gray) (bold values gray) of deformation values of deformation velocity are shown velocity for are each shown geological for eachunit. geological unit. 4.3. Comparison with Water Surface Area of Eber Lake While average subsidence rate is about 15 mm/year in alluvium deposits, other geological units Study [37] claims that the surface area of Lake Eber grew by 2.30% between 1990–2016. However, exhibit very little or no deformation. Although mean subsidence rate of slope debris/talus is about 7 it has recently been reported by several newspapers that the lake has almost dried up as of October mm/year, the maximum subsidence rate reaches up to 25 mm/year. On the other hand, mean 2018 [47,48]. Understanding groundwater–surface water relation is a difficult problem and has received horizontal (E-W) deformation velocity is almost zero for all geological units. However, especially in much attention in the literature. Several recent studies are focused on this issue. Researchers of [49] alluvial sediments, both east and west movements are observed in some regions with the maximum studied interactions between lakes/reservoirs/wetlands with groundwater while [50] presented a rate about 12 mm/year. comprehensive survey of groundwater and surface-water relation, [51] reported the detailed case study about the correlation between groundwater and the surface-water level of Lake Nasser. Nevertheless, 4.3. Comparison with Water Surface Area of Eber Lake it is claimed by [52] that Eber Lake does not directly interact with groundwater and groundwater level changesStudy do [ not37] aclaimsffect the that water thelevel. surface The area lake of is thoughtLake Eber by grew [35] to by be 2.30% mainly between fed by rainfall 1990–2016. and springHowever, water it has coming recently from been the reported Akarçay by and several Sultan newspapers Mountains. that However, the lake it has is suggested almost dried by [up50] as that of precipitationOctober 2018 is [47 also,48 one]. Understanding of the main sources groundwater of groundwater–surface level.waterThus, relation both is Ebera difficult Lake waterproblem surface and andhas groundwater received much level attention are affected in the by climateliterature and. Several precipitation. recent studies are focused on this issue. ResearchersThe water of surface[49] studied areaof interactions Eber Lake variesbetween over lakes/reservoirs/wetlands the years due to different with amounts ground of precipitation,water while evaporation,[50] presented irrigation, a comprehensiv pumpinge survey and surface of ground flowwater [35–37 and]. All su meteorologicalrface-water relation information, [51] reported in Turkey, the includingdetailed case rainfall study data, about is collected the correlation by The between Turkish Stategroundwater Meteorological and the Service surface (TSMS).-water level The monthly of Lake averageNasser. precipitationNevertheless, data it for is the claimed province by of[52 Afyon,] thatwhich Eber wasLake calculated does not using directly the data interact obtained with in periodgroundwater 1929–2018, and groundwater was published level on thechanges TSMS do website not affect [53]. the On water the other level. hand, The thelake Lake is thought of Eber by is one[35] ofto thebe mainly 5000 lakes fed by on rainfall the Earth and which spring is constantlywater coming monitored from the by Akarçay the Bluedot and Sultan Observatory. Mountains. The waterHowever, surface it is area suggested of the lake by is[50 analyzed] that precipitation using multispectral is also one Sentinel-2 of the main optical sources data between of ground Julywater 2015 andlevel. November Thus, both 2018 Eber [54 ].L Theake percentagewater surface of water and surfacegroundwater area relative level are to normal affected lake by boundaries climate and is precipitation. The water surface area of Eber Lake varies over the years due to different amounts of precipitation, evaporation, irrigation, pumping and surface flow [35–37]. All meteorological information in Turkey, including rainfall data, is collected by The Turkish State Meteorological Service (TSMS). The monthly average precipitation data for the province of Afyon, which was calculated using the data obtained in period 1929–2018, was published on the TSMS website [53]. On the other hand, the Lake of Eber is one of the 5000 lakes on the Earth which is constantly monitored by the Bluedot Observatory. The water surface area of the lake is analyzed using multispectral Sentinel-2 optical data between July 2015 and November 2018 [54]. The percentage of water surface area relative to normal lake boundaries is calculated for each Sentinel-2 data. The monthly average rainfall data (mm) and monthly average water surface area (%) of Eber Lake are shown together in Figure 8a. There is quite a good correlation between monthly rainfall data and water surface area of the lake, which shows that precipitation is one of the main sources of Eber Lake. Both rainfall data and water surface area reach to a maximum value in spring and decreases to the minimum level at

Remote Sens. 2019, 11, 1069 11 of 17 calculated for each Sentinel-2 data. The monthly average rainfall data (mm) and monthly average water surface area (%) of Eber Lake are shown together in Figure8a. There is quite a good correlation between monthly rainfall data and water surface area of the lake, which shows that precipitation is one of the main sources of Eber Lake. Both rainfall data and water surface area reach to a maximum value in spring and decreases to the minimum level at the end of summer period. Precipitation is one of the main reasons aﬀecting the water surface area of Eber Lake, but another possible reason might be irrigation. One of the water sources of agricultural areas is deﬁnitely Eber Lake and surrounding rivers that directly feed Eber Lake. Therefore, extensive use of water in agriculture during the summer periodRemote isSens. another 2019, 11, important x FOR PEER factorREVIEW in reducing the water surface area to a minimum level. 11 of 17

(a)

(b) (c)

FigureFigure 8. 8( (aa)) MonthlyMonthly averageaverage rainfall (mm) in in Afyon Afyon city city vs. vs. monthly monthly average average water water surface surface area area of of EberEber Lake Lake and and groundwater groundwater levellevel changeschanges after the long long-term-term decline decline trend trend line line is is removed removed.. After After removalremoval of of long-term long-term subsidencesubsidence trend,trend, P1 LOS displacement is is shown shown tog togetherether with with areal areal water water surfacesurface coverage coverage (%) (%) of Eber of Eber Lake Lake for ( bfor) Asc.-160. (b) Asc. (c-160.) Des.-65 (c) Des. tracks.-65 Bothtracks original. Both data original and polynomial data and approximationpolynomial approximation are shown in are the shown plot. in the plot. the end of summer period. Precipitation is one of the main reasons affecting the water surface area of InSAR LOS displacement time series measurements of the point P1 for both datasets are compared Eber Lake, but another possible reason might be irrigation. One of the water sources of agricultural withareas water is definitely surface area Eber of LakeLake Eber and obtained surrounding from Bluedotrivers that Observatory directly feed in Figure Eber8 b,c.Lake. The Therefore, long-term linearextensive subsidence use of trend water is in subtracted agriculture from during the theLOS summer time series period and is then anot aher polynomial important (15th factor order) in modelreducing is fit the to water each time surface series. area There to a minimum are deficiencies level. in surface water surface area data, especially in winterInSAR seasons LOS due displacement to lack of cloudless time series optical measurements data. Therefore, of the missing point parts P1 for are both filled da againtasets with are a polynomialcompared with approximation. water surface It can area be of observed Lake Eber that obtained the water from surface Bluedot area Observatory of the lake fallsin Figure to its 8 lowestb–c. levelThe atlong the-term end oflinear the subsidence summer period trend and is subtracted reaches its from highest the LOS level time in spring. series and When then the a waterpolynomial level is minimal,(15th order the) model surface is areafit to ofeach the time lake series. falls below There 10%are deficiencies of the natural in surface lake boundaries water surface and area even data, close toespecially zero in 2016. in winter The waterseason surfaces due to arealack startsof cloudless increasing optical in data. autumn Therefore and reaches, missing its parts highest are level filled in spring.again with However, a polynomial it should approximation. also be noted that It can there be isobserved a long term that decreasethe water at surface the highest area of level the of lake lake surfacefalls to area.its lowest As for level the at comparison the end of the with summer LOS displacements, period and reaches there its is highest a clear indicationlevel in spring. of temporal When patternsthe water aff levelected is by minimal seasonal, the water surface surface area areaof the changes lake falls in below the descending 10% of the data natural set. lake Compared boundaries to the descendingand even clos datae to set, zero the in ascending 2016. The dataset water showssurface relatively area starts less increasing correlation in withautumn seasonal and reaches patterns. its highest level in spring. However, it should also be noted that there is a long term decrease at the highest level of lake surface area. As for the comparison with LOS displacements, there is a clear indication of temporal patterns affected by seasonal water surface area changes in the descending data set. Compared to the descending data set, the ascending dataset shows relatively less correlation with seasonal patterns.

4.4. Comparison with Groundwater Level Variations in the groundwater level may trigger ground surface deformation for multiple reasons [17]. One of the most effective reasons is that the decrease in groundwater level causes changes in the stress level on deformable soils that result in subsidence. Therefore, the groundwater level in the region is an important factor that may trigger ground surface deformation. Groundwater level data are obtained from the Directorate of Turkish State Hydraulic Work. Location of the well no 28372 is shown in Figure 4c,d. Cumulative ground surface deformation measured in LOS is

Remote Sens. 2019, 11, 1069 12 of 17

4.4. Comparison with Groundwater Level Variations in the groundwater level may trigger ground surface deformation for multiple reasons [17]. One of the most eﬀective reasons is that the decrease in groundwater level causes changes in the stress level on deformable soils that result in subsidence. Therefore, the groundwater level in the region is an important factor that may trigger ground surface deformation. Groundwater level data are obtained from the Directorate of Turkish State Hydraulic Work. Location of the well no 28372 is shown in Figure4c,d. Cumulative ground surface deformation measured in LOS is compared with groundwater level in Figure9 for ascending and descending datasets, respectively. There is quite a good correlation between groundwater level changes and LOS deformation trends both in long term and short term. In the long term, both data show a generic decline trend due to overexploitation of Remote Sens. 2019, 11, x FOR PEER REVIEW 12 of 17 groundwater. In the short term, with the eﬀect of rainfall and irrigation, seasonal changes are observed.

(a) (b)

FigureFigure 9. 9 LOSLOS displacementdisplacement time series of point P1 vs. temporal temporal evolution evolution of of groundwater groundwater level level at at wellwell no no 28372 28372 are are shown shown together together for for (a (a)) Asc.-160. Asc.-160 (.b (b)) Des.-65 Des.-65 tracks. tracks. Location Location of of the the well well is is shown shown in Figurein Figure4c,d. 4c-d. compared with groundwater level in Figure 9 for ascending and descending datasets, respectively. Irrigation and precipitation are two main reasons for groundwater variations. Groundwater hasThere been is thequite primary a good source correlation of water between for irrigation groundwater in the level region. changes Extensive and LOS use ofdeformation groundwater trends for agricultureboth in long results term inand the short decrease term. of In groundwater the long term level, both throughout data show the a years.generic It decline can be observedtrend due in to Figureoverexploitation9 that there isof a groundwa generic declineter. In trend the short in groundwater term, with the levels effect that of causes rainfall subsidence and irrigation phenomenon, seasonal inchanges the long are term. observed. Together with irrigation, another important factor of groundwater level variation is the rainfall.Irrigation The and long-term precipitation decline are trend two linemain is reasons removed for from groundwater the groundwater variations level. Groundwater and compared has withbeen monthly the primary average source rainfall of water in Figure for 8 irrigationa. As precipitation in the region. increases Extensive in spring, use of the groundwater groundwater for levelagriculture also increases. results in As the a resultdecrease of theof groundwater groundwater level level throughout increase, short the years. term uplift It can is be observed observed on in theFigure ground. 9 that Towards there the is end a generic of the summer, decline thetrend opposite in groundwater effect is observed levels withthat decreasingcauses subsidence rainfall. Hence,phenomenon subsidence in the starts long to term be eff. ectiveTogether in the with region. irrigation, The increase another in important irrigation infactor the summerof groundwater period alsolevel contributes variation tois thethe aforementionedrainfall. The long process.-term decline trend line is removed from the groundwater level and compared with monthly average rainfall in Figure 8a. As precipitation increases in spring, 5.the Discussion groundwater level also increases. As a result of the groundwater level increase, short term uplift is observed on the ground. Towards the end of the summer, the opposite effect is observed with Several studies have been carried out to investigate the causes of surface deformations in the decreasing rainfall. Hence, subsidence starts to be effective in the region. The increase in irrigation in Bolvadin region, [5,6,8]. There is no debate about the primary cause of surface deformation which is the summer period also contributes to the aforementioned process. thought to be hydrological changes. However, there are different approaches about the effects of the tectonics. While [5] do not accept the effect of tectonic movements on ground surface deformations, [6] 5. Discussion suggest that even though there was no destructive earthquake, intensive micro-seismic activity, and relatedSeveral reasons studies may have have aff beenected carried the formation out to investigate of surface deformations. the causes of Groundsurface leveldeformations measurements in the withBolvadin GPS andregion, leveling [5,6,8]. show There the is rateno debate of deformation about the primary on specific cause points of surface around deformation faults and fissures.which is Althoughthought to these be hydrological kinds of techniques changes provide. However precise, there measurements, are different they approaches are limited about to discrete the effects locations. of the tectonics.In this Whil study,e [ the5] do ground not accept deformation the effect analysis of tectonic of the movements entire Bolvadin on ground region issurface performed deformations, by using the[6] multi-temporalsuggest that even InSAR though technique there withwas no Sentinel-1 destructive TOPSAR earthquake, data. Initially, intensive LOS micro deformation-seismic velocityactivity, mapsand arerelated produced reasons for ascendingmay have and affected descending the formation tracks as shown of surface in Figure deformations.4. Later on, Ground vertical and level horizontalmeasurements (E-W) with displacements GPS and leveling are generated show the by rate decomposition of deformation of ascending on specific and points descending around faults LOS and fissures. Although these kinds of techniques provide precise measurements, they are limited to discrete locations. In this study, the ground deformation analysis of the entire Bolvadin region is performed by using the multi-temporal InSAR technique with Sentinel-1 TOPSAR data. Initially, LOS deformation velocity maps are produced for ascending and descending tracks as shown in Figure 4. Later on, vertical and horizontal (E-W) displacements are generated by decomposition of ascending and descending LOS velocities, Figure 5. It is quite clear that the southern and western parts of the city have been exposed to vertical deformation as a whole. Towards the south, the subsidence rate increases and reaches up to 35 mm/year, most probably due to increasing groundwater extraction in the agricultural fields and thickening of the alluvial deposits. Deformation rate highly correlates with the lithological units of the region as most of the vertical deformations is observed in soft alluvial sediments and approaches to zero as we move away from alluvial sediments. In order to evaluate the possible effects of the Bolvadin fault entering the city from the east, deformation rates are examined with profiles of vertical velocity field across the fault. Figure 5

Remote Sens. 2019, 11, 1069 13 of 17 velocities, Figure5. It is quite clear that the southern and western parts of the city have been exposed to vertical deformation as a whole. Towards the south, the subsidence rate increases and reaches up to 35 mm/year, most probably due to increasing groundwater extraction in the agricultural fields and thickening of the alluvial deposits. Deformation rate highly correlates with the lithological units of the region as most of the vertical deformations is observed in soft alluvial sediments and approaches to zero as we move away from alluvial sediments. In order to evaluate the possible effects of the Bolvadin fault entering the city from the east, deformation rates are examined with profiles of vertical velocity field across the fault. Figure5 shows four profiles that perpendicularly cross the fault strike and its probable southwest continuation beneath the alluvium. Vertical and horizontal (E-W) mean deformation velocities over profiles are shown in Figure 10. There is no obvious or consistent vertical or horizontal deformation velocity change across the Bolvadin fault that might be indicative of tectonics origins. However, especially AA’ and BB’ profiles passing through soft alluvium sediments show significant variation across the Bolvadin fault. These asymmetric deformation patterns must be mainly due to the different amount of materials types within alluvial sediments because aseismic slip on the Bolvadin fault is expected to produce subsidence not on the northern side, but on the southern side (i.e., hangingwall) of the fault. The deformation rates of the CC’ and DD’ profiles are more stable in short time periods as they are situated over more compacted lithological units. When the profiles are examined, vertical velocities in the far field away from the Bolvadin fault are in general roughly the same, that is, there is no differential motion (i.e., downthrown of hangingwall southern block) across the fault that might have been caused by creep on a normal fault that dips to the south, implying that the fault has no apparent contribution Remote Sens. 2019, 11, x FOR PEER REVIEW 13 of 17 to the observed subsidence.

(a) (b)

FigureFigure 10. 10 MeanMean velocities along the profiles profiles AA’ AA’-DD’-DD’ shown shown in in Figure Figure 55..( (a)a )Vertical Vertical mean mean velocity. velocity. (b(b)) Horizontal Horizontal (E-W) (E-W) mean mean velocity. velocity. shows four profiles that perpendicularly cross the fault strike and its probable southwest We compare water surface area of Eber Lake and LOS displacement time series for ascending andcontinuation descending beneath orbits. the As alluvium shown in. Vertical Figure8 b,c,and a horizontal clear correlation (E-W) mean between deformation the seasonal velocities variation over of waterprofiles surface are shown area of in Eber Figure Lake 10. andThere displacement is no obvious time or consistent series is revealed vertical or when horizontal the long-term deformation linear subsidencevelocity change trend isacross removed the fromBolvadin theLOS fault displacements. that might be There indicative is a lag of time tectonics between origins. the peak However, level of theespecially lake water AA surface' and BB' area profiles and the passing displacement through timesoft series.alluvium It can sediments be observed show from significant the time variation series of individualacross the points Bolvadin that correlationfault. These of asymmetric LOS displacements deformation with seasonal patterns variation must be of mainly water duesurface to areathe increasesdifferent towardsamount of the materials south of types the region. within The alluvial effect sediments of seasonal because change aseismic on different slip on regions the Bolvadin can also befault examined is expected by looking to produce at the subsidence local standard not deviation on the northern of the mean side, deformation but on the southern velocity. side Statistics (i.e. hangingwall) of the fault. The deformation rates of the CC' and DD' profiles are more stable in short provided in Figure7 include the whole area for each geological unit. In order to observe how seasonal time periods as they are situated over more compacted lithological units. When the profiles are variation affects different regions, additionally, the standard deviations for local regions alone are examined, vertical velocities in the far field away from the Bolvadin fault are in general roughly the also calculated. The average standard deviation for the deformation rate increases for both dataset same, that is, there is no differential motion (i.e., downthrown of hangingwall southern block) across from 0.3 mm/year to 1.7 mm/year as from north to the south. The increase of standard deviation is the fault that might have been caused by creep on a normal fault that dips to the south, implying that a direct effect of seasonal variation. Although the water source of Eber Lake has no direct relation the fault has no apparent contribution to the observed subsidence. with groundwater, both Eber Lake water surface and ground water level are affected by climate and We compare water surface area of Eber Lake and LOS displacement time series for ascending fed by precipitation. Moreover, LOS displacement time series for both datasets are compared with and descending orbits. As shown in Figure 8b,c, a clear correlation between the seasonal variation of groundwater level changes. There is quite a good correlation between groundwater level changes and water surface area of Eber Lake and displacement time series is revealed when the long-term linear subsidence trend is removed from the LOS displacements. There is a lag time between the peak level of the lake water surface area and the displacement time series. It can be observed from the time series of individual points that correlation of LOS displacements with seasonal variation of water surface area increases towards the south of the region. The effect of seasonal change on different regions can also be examined by looking at the local standard deviation of the mean deformation velocity. Statistics provided in Figure 7 include the whole area for each geological unit. In order to observe how seasonal variation affects different regions, additionally, the standard deviations for local regions alone are also calculated. The average standard deviation for the deformation rate increases for both dataset from 0.3 mm/year to 1.7 mm/year as from north to the south. The increase of standard deviation is a direct effect of seasonal variation. Although the water source of Eber Lake has no direct relation with groundwater, both Eber Lake water surface and ground water level are affected by climate and fed by precipitation. Moreover, LOS displacement time series for both datasets are compared with groundwater level changes. There is quite a good correlation between groundwater level changes and LOS deformation trends both in long term and short term as shown in Figure 9a–b. As a result, the relation of groundwater level and lake water surface area with displacement time series show effects of hydrological changes to deformation. Correlation of the high deformation rate with soft lithological units of the Bolvadin region, the similarity between deformation rate and groundwater level changes and the high correlation of deformation rate change with seasonal variations in water surface of Eber Lake suggest that subsidence phenomenon might be a direct result of hydrological effects in the region. Besides, mean deformation velocities over the profile lines AA’-DD’ suggest that there is no clear spatial evidence about the effects of the Bolvadin fault on deformation. Furthermore, the surface deformations in the a, c regions such as fracturing and fissuring are almost in the south-west north-east direction which

Remote Sens. 2019, 11, 1069 14 of 17

LOS deformation trends both in long term and short term as shown in Figure9a,b. As a result, the relation of groundwater level and lake water surface area with displacement time series show effects of hydrological changes to deformation. Correlation of the high deformation rate with soft lithological units of the Bolvadin region, the similarity between deformation rate and groundwater level changes and the high correlation of deformation rate change with seasonal variations in water surface of Eber Lake suggest that subsidence phenomenon might be a direct result of hydrological effects in the region. Besides, mean deformation velocities over the profile lines AA’-DD’ suggest that there is no clear spatial evidence about the effects of the Bolvadin fault on deformation. Furthermore, the surface deformations in the a, c regions such as fracturing and fissuring are almost in the south-west north-east direction which is not parallel to the Bolvadin fault. As a result, it is very likely that those surface deformations observed in the region are the consequences of generic deformation trend affecting the whole graben system which is mainly caused by overexploitation of ground water and hydrological changes in the region.

6. Conclusions We have mapped the ground surface deformation in Bolvadin using Sentinel-1 SAR images on both ascending and descending tracks within the time period of October 2014 and October 2018. As a result of multi-temporal InSAR analysis, two main products are produced: (i) Mean deformation velocity maps, (ii) Time-series of displacements for measurement points. The mean velocity maps show that the town of Bolvadin almost entirely subsides at different rates reaching 35 mm/year. The subsidence is taking place not only in Bolvadin, but also almost in the entire Afyon-Ak¸sehirGraben. We observe a strong spatial correlation between subsidence and lithology, a high similarity between deformation rate and groundwater level changes, and a temporal correlation between subsidence rate and Eber Lake water surface changes. There is no specific deformation pattern observed around the Bolvadin fault that might continue westward into the city center. As a result, we conclude that the primary cause of surface deformations is the overexploitation of ground waters and hydrological changes that affect the entire basin. This study is the first step of the project related to the continuous monitoring of land deformation in the country scale using SAR data. The main purpose of this study is to investigate deformation phenomenon in a region-scale and extend our understanding of it to all subsidence regions across the country. Continuous monitoring of deformations throughout the country is crucial for sustainable planning and risk management. Sentinel-1 data show a high potential in the field of near real-time land deformation monitoring with wide coverage and regular data acquisition. This study suggests that the temporal and spatial behaviors of subsidence basins in Turkey can be investigated using Sentinel-1 data and PSI technique. To achieve this, all the processing steps used in this study will be updated to be capable of working on multiple processor stations and computer clusters to handle the enormous workload.

Author Contributions: M.I. carried out SAR data processing, data analysis, manuscript writing and revision. F.K. and Z.C. contributed to data interpretation, manuscript writing and revision. F.B.S. contributed to manuscript revision. Funding: This research was funded by Republic of Turkey Ministry of Interior Disaster and Emergency Management Presidency (AFAD) and TUBITAK BILGEM, project AYDES UZAL and project No. B740-100289. Acknowledgments: This is a part of the Ph.D. dissertation of Mümin Imamoglu. We thank the Republic of Turkey Ministry of Interior Disaster and Emergency Management Presidency (AFAD) for providing the active faults and geological settings maps of the study region. We thank the Turkish State Hydraulic Works for providing groundwater level data. Conﬂicts of Interest: The authors declare no conﬂict of interest. Remote Sens. 2019, 11, 1069 15 of 17

References

1. Strozzi, T.; Wegmüller, U.; Tosi, L.; Bitelli, G.; Spreckels, V. Land subsidence monitoring with differential SAR interferometry. Photogramm. Eng. Remote Sens. 2001, 67, 1261–1270. 2. Galloway, D.; Jones, D.R.; Ingebritsen, S.E. Land Subsidence in the United States; U.S. Department of the Interior, U.S. Geological Survey: Reston, VA, USA, 1999; Volume 1182. 3. Holzer, T.L.; Galloway, D.L. Impacts of land subsidence caused by withdrawal of underground fluids in the united states. Rev. Eng. Geol. 2005, 16, 87–99. 4. Tomás, R.; Li, Z. Earth observations for geohazards: Present and future challenges. Remote Sens. 2017, 9, 194. [CrossRef] 5. Demirta¸s,R.; Tepeu˘gur, E. Bolvadin (Afyon) Merkez yerle¸simalaninda son 15 yillik dönemde olu¸smu¸s yüzey kiriklarinin olu¸summekanizmasi. In Proceedings of the ATAG-21: Aktif Tektonik Ara¸stırmaGrubu Çalı¸stayı,Afyon, Turkey, 26–28 October 2017. 6. Özkaymak, Ç.; Sözbilir, H.; Tiryakio˘glu, I.;˙ Baybura, T. Bolvadin’de (Afyon-Ak¸sehirGrabeni, Afyon) Gözlenen Yüzey Deformasyonlarının Jeolojik, Jeomorfolojik ve Jeodezik Analizi. Türk. Jeol. Bül. Geolog. Bull. Turk. 2017, 60, 169–189. [CrossRef] 7. Emre, Ö.; Duman, T.Y.; Özalp, S.; Olgun, S.; Elmaci, H. 1:250,000 Scale active fault map series of Turkey Afyon (NJ 36-5) quadrangle. In General Directorate of Mineral Research and Exploration; (Serial number: 16); General Directorate of Mineral Research and Expansion (MTA): Ankara, Turkey, 2011. 8. Tiryakio˘glu, I.;˙ U˘gur, M.A.; Yalçın, M.; Baybura, T.; Yilmaz, M.; Özkaymak, Ç.; Sözbilir, H. Ak¸sehirSimav Fay Sistemindeki Güncel Tektonik Hareketlerinin Izlenmesi: Nivelman Çali¸smalari.4. In Proceedings of the Uluslararası Deprem Mühendisli˘give Sismoloji Konferansı, Eski¸sehir, Turkey, 11–13 October 2017. 9. Qu, F.; Lu, Z.; Zhang, Q.; Bawden, G.W.; Kim, J.-W.; Zhao, C.; Qu, W. Mapping ground deformation over Houston-Galveston, Texas using multi-temporal InSAR. Remote Sens. Environ. 2015, 169, 290–306. [CrossRef] 10. Gabriel, A.K.; Goldstein, R.M.; Zebker, H.A. Mapping small elevation changes over large areas: differential radar interferometry. J. Geophys. Res. Solid Earth 1989, 94, 9183–9191. [CrossRef] 11. Massonnet, D.; Feigl, K. Radar interferometry and its application to changes in the Earth’s surface. Rev. Geophys. 1998, 36, 441–500. [CrossRef] 12. Tomás, R.; Romero, R.; Mulas, J.; Marturià, J.J.; Mallorquí, J.J.; Lopez-Sanchez, J.M.; Duque, S. Radar interferometry techniques for the study of ground subsidence phenomena: a review of practical issues through cases in Spain. Environ. Earth Sci 2014, 71, 163–181. [CrossRef] 13. Hooper, A.; Bekaert, D.; Spaans, K.; Arıkan, M. Recent advances in SAR interferometry time series analysis for measuring crustal deformation. Tectonophysics 2011, 514, 1–13. [CrossRef] 14. Osmano˘glu,B.; Sunar, F.; Wdowinski, S.; Cabral-Cano, E. Time series analysis of InSAR data: Methods and trends. ISPRS J. Photogramm. Remote Sens. 2015, 115, 90–102. [CrossRef] 15. 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] 16. Bell, J.W.; Amelung, F.; Ferretti, A.; Bianchi, M.; Novali, F. Permanent scatterer InSAR reveals seasonal and long-term aquifer-system response to groundwater pumping and artificial recharge. Water Resour. Res. 2008, 44, 1–18. [CrossRef] 17. Bonì, R.; Herrera, G.; Meisina, C.; Notti, D.; Béjar-Pizarro, M.; Zucca, F.; Gonzàlez, P.J.; Palano, M.; Tomàs, R.; Fernàndez, J.; et al. Twenty-year advanced DIn-SAR analysis of severe land subsidence: The Alto Guadalentín Basin (Spain) case study. Eng. Geol. 2015, 198, 40–52. [CrossRef] 18. Chen, M.; Tomás, R.; Li, Z.; Motagh, M.; Li, T.; Hu, L.; Gong, H.; Li, X.; Yu, J.; Gong, X. Imaging land subsidence induced by groundwater extraction in Beijing (China) using satellite radar interferometry. Remote Sens. 2016, 8, 468. [CrossRef] 19. Chaussard, E.; Milillo, P.; Bürgmann, R.; Perissin, D.; Fielding, E.J.; Baker, B. Remote sensing of ground deformation for monitoring groundwater management practices: Application to the Santa Clara Valley During the 2012–2015 California drought. J. Geophys. Res. Solid Earth 2017, 122, 8566–8582. [CrossRef] 20. Haghighi, M.H.; Motagh, M. Ground surface response to continuous compaction of aquifer system in Tehran, Iran: Results from a long-term multi-sensor InSAR analysis. Remote Sens. Environ. 2019, 221, 534–550. [CrossRef] Remote Sens. 2019, 11, 1069 16 of 17

21. Imamoglu, M.; Kahraman, F.; Abdikan, S. Preliminary results of temporal deformation analysis in Istanbul using multi-temporal InSAR with Sentinel-1 SAR data. In Proceedings of the 2018 IEEE International Geoscience and Remote Sensing Symposium (IGARSS), Valencia, Spain, 22–27 July 2018. 22. Aslan, G.; Cakır, Z.; Ergintav, S.; Lasserre, C.; Renard, F. Analysis of secular ground motions in Istanbul from a long-term InSAR time-series (1992–2017). Remote Sens. 2018, 10, 408. [CrossRef] 23. Bonì, R.; Bosino, A.; Meisina, C.; Novellino, A.; Bateson, L.; McCormack, H. A methodology to detect and characterize uplift phenomena in urban areas using Sentinel-1 data. Remote Sens. 2018, 10, 607. [CrossRef] 24. Erten, E.; Rossi, C. The worsening impacts of land reclamation assessed with Sentinel-1: The Rize (Turkey) test case. Int. J. Appl. Earth Obs. Geoinf. 2019, 74, 57–64. [CrossRef] 25. Abdikan, S.; Arıkan, M.; ¸Sanlı,F.B.; Çakir, Z. Monitoring of coal mining subsidence in peri-urban area of Zongundak city (NW Turkey) with persistent scatterer interferometry using ALOS-PALSAR. Environ. Earth Sci. 2014, 71, 4081–4089. [CrossRef] 26. Czikhardt, R.; Papco, J.; Bakon, M.; Liscak, P.; Ondrejka, P.; Zlocha, M. Ground stability monitoring of undermined and landslide prone areas by means of sentinel-1 multi-temporal InSAR, case study from Slovakia. Geosciences 2017, 7, 87. [CrossRef] 27. Carlà, T.; Farina, P.; Intrieri, E.; Ketizmen, H.; Casagli, N. Integration of ground-based radar and satellite InSAR data for the analysis of an unexpected slope failure in an open-pit mine. Eng. Geol. 2018, 235, 39–52. [CrossRef] 28. Béjar-Pizarro, M.; Notti, D.; Mateos, R.M.; Ezquerro, P.; Centolanza, G.; Herrera, G.; Bru, G.; Sanabria, M.; Solari, L.; Duro, J.; et al. Mapping vulnerable urban areas aﬀected by slow-moving landslides using sentinel-1 InSAR data. Remote Sens. 2017, 9, 876. [CrossRef] 29. Intrieri, E.; Raspini, F.; Fumagalli, A.; Lu, P.; Del Conte, S.; Farina, P.; Allievi, J.; Ferretti, A.; Casagli, N. The maoxian landslide as seen from space: Detecting precursors of failure with sentinel-1 data. Landslides 2018, 15, 123–133. [CrossRef] 30. Bayer, B.; Simoni, A.; Mulas, M.; Corsini, A.; Schmidt, D. Deformation responses of slow moving landslides to seasonal rainfall in the Northern Apennines, measured by InSAR. Geomorphology 2018, 308, 293–306. [CrossRef] 31. Kim, J.W.; Lu, Z.; Degrandpre, K. Ongoing deformation of sinkholes in Wink, Texas, observed by time-series Sentinel-1a SAR interferometry (preliminary results). Remote Sens. 2016, 8, 313. [CrossRef] 32. Caló, F.; Notti, D.; Galve, J.P.; Abdikan, S.; Görüm, T.; Pepe, A.; Balik ¸Sanli,F. DInSAR-based detection of land subsidence and correlation with groundwater depletion in Konya Plain, Turkey. Remote Sens. 2017, 9, 83. [CrossRef] 33. Hussain, E.; Wright, T.J.; Walters, R.J.; Bekaert, D.P.; Lloyd, R.; Hooper, A. Constant strain accumulation rate between major earthquakes on the North Anatolian Fault. Nat. Commun. 2018, 9, 1392. [CrossRef] 34. Liu, C.; Ji, L.; Zhu, L.; Zhao, C. InSAR-constrained interseismic deformation and potential seismogenic

asperities on the Altyn Tagh Fault at 91.5–95◦ E, Northern Tibetan Plateau. Remote Sens. 2018, 10, 943. [CrossRef] 35. Sener, E.; Davraz, A.; Sener, S. Investigation of Aksehir and Eber Lakes (SW Turkey) coastline change with multitemporal satellite images. Water Resour. Manag. 2010, 24, 727–745. [CrossRef] 36. Yıldırım, Ü.; Erdo˘gan,S.; Uysal, M. Changes in the coastline and water level of the Ak¸sehirand Eber Lakes between 1975 and 2009. Water Resour. Manag. 2011, 25, 941–962. [CrossRef] 37. Köle, M.M.; Ataol, M.; Erkal, T. Eber ve Ak¸sehirGölleri’nde 1990–2016 Yılları arasinda gerçekle¸senalansal de˘gi¸simler. In Proceedings of the TUCAUM 2016 International Geography Symposium Book of Proceedings, Ankara, Turkey, 13–14 October 2016. 38. Sentinel Application Platform (SNAP). Available online: http://step.esa.int/main/toolboxes/snap (accessed on 20 December 2018). 39. Miranda, N. Sentinel-1 Instrument Processing Facility: Impact of the Elevation Antenna Pattern Phase Compensation on the Interferometric Phase Preservation; Technical Report, ESA-EOPG-CSCOP-TN-0004; European Space Agency: Paris, France, 2015. 40. Sentinel-1 Quality Control. Available online: https://qc.sentinel1.eo.esa.int (accessed on 20 December 2018). 41. Farr, T.G.; Rosen, P.A.; Caro, E.; Crippen, R.; Duren, R.; Hensley, S.; Alsdorf, D. The shuttle radar topography mission. Rev. Geophys. 2007, 45.[CrossRef] Remote Sens. 2019, 11, 1069 17 of 17

42. Hooper, A.; Zebker, H.; Segall, P.; Kampes, B. A new method for measuring deformation on volcanoes and other natural terrains using InSAR persistent scatterers. Geophys. Res. Lett. 2004, 31, 5. [CrossRef] 43. Ferretti, A.; Prati, C.; Rocca, F. Permanent scatterers in SAR interferometry. IEEE Transact. on Geoscie. Remote Sens. 2001, 39, 8–20. [CrossRef] 44. Bekaert, D.P.S.; Walters, R.J.; Wright, T.J.; Hooper, A.J.; Parker, D.J. Statistical comparison of InSAR tropospheric correction techniques. Remote Sens. Environ. 2015, 170, 40–47. [CrossRef] 45. Wright, T.J.; Parsons, B.E.; Zhong, L. Toward mapping surface deformation in three dimensions using InSAR. Geophys. Res. Lett. 2004, 31, 169–178. [CrossRef] 46. Hu, J.; Li, Z.W.; Ding, X.L.; Zhu, J.J.; Zhang, L.; Sun, Q. Resolving three-dimensional surface displacements from InSAR measurements: A review. Earth Sci. Rev. 2014, 133, 1–17. [CrossRef] 47. Hürriyet Gazetesi. Available online: http://www.hurriyet.com.tr/gundem/eber-golu-tamamen-kurudu- 40991682 (accessed on 2 January 2019). 48. T24 Ba˘gımsız Internet˙ Gazetesi. Available online: http://t24.com.tr/haber/afyonkarahisardaki-eber-golu- tamamen-kurudu,727012 (accessed on 2 January 2019). 49. Winter, T.C.; Harvey, J.W.; Franke, O.L.; Alley, W.M. Ground Water and Surface Water—A Single Resource; Circular 1139; US Geological Survey: Reston, VA, USA, 1998. 50. Sophocleous, M. Interactions between groundwater and surface water: the state of the science. Hydrogeol. J. 2002, 10, 52–67. [CrossRef] 51. Ghoubachi, S.Y. Impact of lake Nasser on the groundwater of the Nubia sandstone aquifer system in Tushka area, South Western Desert, Egypt. J. King Saud Univ. Sci. 2012, 24, 101–109. [CrossRef] 52. Tezcan, L.; Meriç, B.T.; Do˘gdu,N.; Akan, B.; Atilla, A.Ö.; Kurtta¸s,T. Akarçay havzası hidrojeolojisi ve yeraltısuyu akım modeli; Project Final Report; Hacettepe University International Karst Water Sources Research and Application Center (UKAM): Valencia, CA, USA, 2002; p. 13. 53. The Turkish State Meteorological Service Web Site. Available online: https://www.mgm.gov.tr/ veridegerlendirme/il-ve-ilceler-istatistik.aspx?k=A&m=AFYONKARAHISAR (accessed on 7 April 2019). 54. Bluedot Water Observatory. Available online: https://water.blue-dot-observatory.com/2361 (accessed on 19 November 2018).

© 2019 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/).