Remote Sensing

remote sensing

Article InSAR Campaign Reveals Ongoing Displacement Trends at High Impact Sites of Thessaloniki and Chalkidiki, Greece

Nikos Svigkas 1,* , Constantinos Loupasakis 2 , Ioannis Papoutsis 3 , Charalampos (Haris) Kontoes 3, Stavroula Alatza 3 , Ploutarchos Tzampoglou 4, Cristiano Tolomei 1 and Thomas Spachos 5

1 Osservatorio Nazionale Terremoti, Istituto Nazionale di Geoﬁsica e Vulcanologia, 00143 Rome, Italy; [email protected] 2 Laboratory of Engineering Geology and Hydrogeology, Department of Geological Sciences, School of Mining and Metallurgical Engineering, National Technical University of Athens, 15780 Athens, Greece; [email protected] 3 Institute of Space Applications and Remote Sensing, National Observatory of Athens, 15236 Athens, Greece; [email protected] (I.P.); [email protected] (C.H.K.); [email protected] (S.A.) 4 Department of Civil and Environmental Engineering, School of Engineering, University of Cyprus, 1678 Nicosia, Cyprus; [email protected] 5 Water Supply and Sewerage Company of Thessaloniki, 54635 Thessaloniki, Greece; [email protected] * Correspondence: [email protected]

Received: 19 June 2020; Accepted: 24 July 2020; Published: 26 July 2020

Abstract: We studied the broader area of Thessaloniki in northern Greece and Chalkidiki and performed an InSAR campaign to study the surface deformation phenomena that have been known to exist for at least two decades. Sentinel-1 data (2015–2019) together with drill measurements were exploited to focus on specific sites of interest. Our results indicate an ongoing displacement field. At the region of Kalochori and Sindos—where intense subsidence in the 1990s was previously found to have had a natural surface rebound in the 2000s—a new period of subsidence, caused by the enlivenment of the groundwater overexploitation, was reported. The uplifting trend of Oreokastro is still active and subsidence in Anthemountas graben is ongoing; special focus was set on the Makedonia Airport, where significant displacement is occurring. The study also reveals a new area at Nea Moudania, that was not known previously to deform; another case corresponding to anthropogenic-induced surface displacement. Thessaloniki is surrounded by different persistent displacement phenomena, whose main driving mechanisms are anthropogenic. The sensitivity of the surface displacements to the water trends is highlighted in parts of the study area. Results highlight the plan of a water resources management as a high priority for the area.

Keywords: InSAR; PSI time-series; deformation; Northern Greece; Kalochori; Oreokastro; Anthemountas; thessaloniki airport; Nea Moudania

1. Introduction Thessaloniki is located in Northern Greece and is the second largest city of the country (Figure1). Past studies of the broader area have highlighted several subsiding deformations occurring in the wider area (e.g., [1,2]). Synthetic Aperture Radar Interferometry (InSAR) techniques have been successfully applied using data from radar satellites for the monitoring of surface displacement in several cases globally (e.g., [3,4]) and also in Greece (e.g., [5–8]). The advent of technological advancements of radar satellites has provided signiﬁcant input data and has given rise to a series of InSAR studies at the

Remote Sens. 2020, 12, 2396; doi:10.3390/rs12152396 www.mdpi.com/journal/remotesensing Remote Sens. 2020, 12, 2396 2 of 26 wider region of Thessaloniki. Some of these studies have detected and measured the ground surface displacement (e.g., [9–11]). Other InSAR studies used in-situ data and ground campaigns to interpret Remote Sens. 2020, 12, x FOR PEER REVIEW 2 of 27 and ﬁnd the cause of the reported displacement signals [12–17]. It is widely accepted that underground wateraims activityto study can the cause surface surface deformati displacementon occurring [18]. in This the papervicinity aims of toThessaloniki study the surface(Figure deformation1) and to occurringconsider inthe the role vicinity of anthropogenic of Thessaloniki activities (Figure in1) andthe evolution to consider of the these role deformation of anthropogenic phenomena. activities inAquifer the evolution overpumping of these had deformation been identified phenomena. as the Aquifer main cause overpumping of deformation had been for identiﬁed the areas asof the mainKalochori–Sindos, cause of deformation for the broader for the areas area ofof Kalochori–Sindos,Oreokastro, as well for as thefor broaderthe Anthemountas area of Oreokastro, basin (Figure as well as1) for [12–17]. the Anthemountas basin (Figure1)[12–17].

Figure 1. Areas of interest in Northern Greece located in the wider Thessaloniki and Chalkidiki regions. OrangeFigure rectangle1. Areas inof theinterest upper in inset Northern map indicatesGreece locate the broaderd in the study wider site. Thessaloniki The areas haveand Chalkidiki been studied inregions. the framework Orange ofrectangle the Greece in the InSAR upper project inset runmap by indicates NOA/BEYOND the broader (www.beyond-eocenter.eu study site. The areas have) using Sentinel-1been studied data ofin the the period framework 2015-2019. of the Greece InSAR project run by NOA/BEYOND (www.beyond-eocenter.eu) using Sentinel-1 data of the period 2015-2019. In this study, we exploit the Sentinel-1 satellite, operated by the European Space Agency (ESA), to answerIn this the study, question we exploit of whether the Sentinel-1 there is a satellit considerablee, operated continuing, by the European persisting Space hazard Agency at the (ESA), sites of interestto answer or whether the question it has of diminished whether there or beenis a consider replacedable by continuing, other types persisting of geo-hazards. hazard Theat the geo-hazard sites of interpretationinterest or whether is strongly it supportedhas diminished by using or databeen from replaced in-situ conductedby other types campaigns of geo-hazards. in the specific The sites ofgeo-hazard interest. Focus interpretation is set on theis strongly areas of supported Kalochori–Sindos, by using Oreokastro,data from in-situ and Anthemountas, conducted campaigns which werein the specific sites of interest. Focus is set on the areas of Kalochori–Sindos, Oreokastro, and highly impacted by surface displacement, the latter resulting occasionally in numerous damages of Anthemountas, which were highly impacted by surface displacement, the latter resulting houses, roads, and industrial buildings. Moreover, new deformation evidence has emerged at the area occasionally in numerous damages of houses, roads, and industrial buildings. Moreover, new of Nea Moudania (Figure1). No previous study has reported any type of deformation occurring in this deformation evidence has emerged at the area of Nea Moudania (Figure 1). No previous study has area. Apart from the scientific curiosity that is the dominant driver for the study, significant motivation reported any type of deformation occurring in this area. Apart from the scientific curiosity that is the has also been the fact that the broader Thessaloniki metropolitan area consists a hub for the Balkans dominant driver for the study, significant motivation has also been the fact that the broader andThessaloniki south-eastern metropolitan Europe; here area lay consists important a hub economic for the Balkans activities and and south-eastern critical infrastructure, Europe; here and lay many actorsimportant are placed economic belonging activities to theand primary critical infrastructure, and touristic sectors and many as well. actors are placed belonging to the primary and touristic sectors as well.

Remote Sens. 2020, 12, 2396 3 of 26

2. Methods and Data Used We used 91 Sentinel-1 satellite images, descending pass, orbit no. 7, and frame 457, spanning the period 12 December 2015–15 December 2015 (Table1). The detection and measurement of surface displacement was achieved by performing Multi Temporal InSAR (MT-InSAR). Our processing chain was Parallelized Persistent Scatterer Interferometry (P-PSI; Papoutsis et al. [19]) which is a novel, distributed processor for the fully automated and computationally eﬃcient assessment of line-of-sight ground velocities through PSI, tailored to scale to the large multi-temporal archive of Sentinel-1 data.

Table 1. Radar data used. Normal Normal Normal Date Date Date Baseline Baseline Baseline 12 December 15 88 28 February 17 37 26 November 18 15 − 24 December 15 6 12 March 17 15 08 December 18 38 − − 05 January 16 48 24 March 17 84 20 December 18 47 17 January 16 94 05 April 17 77 01 January 19 45 29 January 16 89 17 April 17 75 13 January 19 75 10 February 16 7 29 April 17 49 06 February 19 19 05 March 16 33 11 May 17 26 02 March 19 19 − − 17 March 16 112 20 October 17 0 14 March 19 117 29 March 16 65 01 November 17 14 26 March 19 58 − 10 April 16 35 13 November 17 53 07 April 19 5 − 22 April 16 5 25 November 17 45 19 April 19 40 − 16 May 16 73 07 December 17 37 01 May 19 40 − 28 May 16 78 19 December 17 51 13 May 19 98 − 09 June 16 73 31 December 17 31 06 June 19 61 − 15 July 16 95 05 February 18 30 18 June 19 58 27 July 16 103 01 March 18 15 30 June 19 16 − 08 August 16 59 11 June 18 18 12 July 19 86 20 August 16 12 23 June 18 62 24 July 19 98 − 01 September 16 81 05 July 18 163 05 August 19 82 13 September 16 61 17 July 18 85 17 August 19 25 − 25 September 16 23 29 July 18 77 29 August 19 27 07 October 16 33 10 August 18 24 10 September 19 81 19 October 16 66 22 August 18 116 22 September 19 48 31 October 16 36 03 September 18 44 04 October 19 17 − 12 November 16 36 15 September 18 71 16 October 19 46 24 November 16 40 27 September 18 23 28 October 19 31 06 December 16 22 09 October 18 19 09 November 19 36 11 January 17 30 21 October 18 34 21 November 19 68 − 23 January 17 68 02 November 18 49 03 December 19 44 04 February 17 68 14 November 18 15 15 December 19 29 − 16 February 17 86

Our P-PSI implementation invokes two main software packages that have been modified to parallelize the execution of the different processing tasks. We use the InSAR Scientific Computing Environment (ISCE) with the topsStack processor [20] for creating a stack of coregistered Single Look Complex (SLC) datasets. Several processing steps with ISCE software were parallelized, including the generation of overlap interferograms and the computation of geometrical offsets among all slave SLCs and the stack master, for precise coregistration. The SLC stack was then used for PSI analysis with a parallelized implementation of the Stanford Method for Persistent Scatterers (StaMPS; Hooper et al. [21,22]), a process that produces ground velocities and deformation histories for the Persistent Scatterers (PS) that have been detected. The full Sentinel-1 frame was divided in sub-regions, i.e., “patches”. Our processing chain exploited multi-core programming techniques to execute simultaneously the processing steps of the Stanford method for Persistent Scatterers in multiple Remote Sens. 2020, 12, x FOR PEER REVIEW 4 of 27 for the Persistent Scatterers (PS) that have been detected. The full Sentinel-1 frame was divided in sub-regions, i.e., “patches”. Our processing chain exploited multi-core programming techniques to execute simultaneously the processing steps of the Stanford method for Persistent Scatterers in Remote Sens. 2020, 12, 2396 4 of 26 multiple patches. StaMPS relies on the spatial correlation of the deformation, rather than on a temporal linear deformation model. It is therefore suitable to also detect non-linear deformation. patches.To estimate StaMPS the relies topography on the spatial effect, correlation the Shuttl of thee Radar deformation, Topography rather thanMission on a temporalDigital Elevation linear Modeldeformation (SRTM DEM)[23] model. It iswith therefore a resolution suitable to of also 1 detectarc-second, non-linear was deformation. employed. In order to have a To estimate the topography effect, the Shuttle Radar Topography Mission Digital Elevation Model reasonable trade-off between a larger pixel spacing and pixels with random phase, we opted for a (SRTM DEM) [23] with a resolution of 1 arc-second, was employed. In order to have a reasonable threshold of 20 m for the maximum topographic error. To exclude pixels considered as “noisy” or trade-off between a larger pixel spacing and pixels with random phase, we opted for a threshold of pixels20 selected m for the as maximum PSI, due topographic to signal error.contribution To exclude from pixels neighboring considered asground “noisy” resolution or pixels selected elements, as a thresholdPSI, due of to1 signalwas selected contribution for the from phase neighboring noise standard ground resolution deviation elements, for all apixel threshold pairs. of Therefore, 1 was pixelsselected exceeding for the this phase threshold noise standardwere exclud deviationed from for further all pixel steps pairs. of Therefore, PS processing. pixels exceeding this Thethreshold open-source were excluded Toolbox from for further Reducing steps Atmosphe of PS processing.ric InSAR Noise (TRAIN) [24] was employed, by implementingThe open-source a phase-based Toolbox for linear Reducing correction Atmospheric on interferograms InSAR Noise (TRAIN) produced [24] was for employed, PS processing. by Sinceimplementing the areas of a phase-basedinterest investigated linear correction are oncharacterized interferograms by produced divergent for PS topography, processing. Since the thelinear correctionareas ofapplied, interest investigatedwhich compensates are characterized for the by divergenttropospheric topography, delay therelative linear correctionto the topography, applied, provideswhich a compensatessatisfying accuracy for the tropospheric in the tropospheric delay relative delay to the correction. topography, provides a satisfying accuracy in the tropospheric delay correction. The outcome of P-PSI is a collection of pixels (PS) with stable phase over time. These points The outcome of P-PSI is a collection of pixels (PS) with stable phase over time. These points provide estimates of annual velocities and a quality measure named velocity standard deviation provide estimates of annual velocities and a quality measure named velocity standard deviation (Figure(Figure S1). S1).High High values values for forthe the latter latter indicate indicate eith eitherer a noisynoisy estimateestimate or or provide provide an an indication indication of of non-linearnon-linear deformation. deformation. In addition, In addition, for for each each PS, PS, a a deformation time-series time-series in in the the period period 2015–2019 2015–2019 was providedwas provided that that highlights highlights various various deformation deformation trends.

3. Results3. Results The Theinterferometric interferometric pairs pairs that that were were formed formed are are sc schematicallyhematically shown shown in in Figure Figure2 with 2 with respect respect to to their temporal and perpendicular baselines. We processed the entire set of available Sentinel-1 images their temporal and perpendicular baselines. We processed the entire set of available Sentinel-1 for the area and period of interest and identiﬁed a dense and spatially well distributed network of images for the area and period of interest and identified a dense and spatially well distributed 1,551,817 stable-phase pixels, for which we estimated ground displacements (Figure3). network of 1,551,817 stable-phase pixels, for which we estimated ground displacements (Figure 3).

Figure 2. Sentinel-1 SAR imagery PSI connection graph. The date of the master SAR scene used for PSI Figure 2. Sentinel-1 SAR imagery PSI connection graph. The date of the master SAR scene used for analysis was 20 October 2017. Perpendicular baselines are conﬁned within 150 m only, given the ± PSI analysisshort Sentinel-1 was 20 orbital October tube. 2017. Perpendicular baselines are confined within ± 150 m only, given the short Sentinel-1 orbital tube.

Remote Sens. 2020, 12, 2396 5 of 26 Remote Sens. 2020, 12, x FOR PEER REVIEW 5 of 27

Figure 3. Sentinel-1 PS LoS velocity results in Northern Greece, for the period 2015–2019. Blue rectangles Figure 3. Sentinel-1 PS LoS velocity results in Northern Greece, for the period 2015–2019. Blue indicate the areas of study (see also Figure1). The magenta dot indicates the VRAS c-GPS stations used rectangles indicate the areas of study (see also Figure 1). The magenta dot indicates the VRAS c-GPS as a reference point for the processing of the speciﬁc Sentinel-1 frame. stations used as a reference point for the processing of the specific Sentinel-1 frame. Since PSI estimates are a relative measurement on both the spatial and temporal domain, we used in ourSince PSI thePSI SAR estimates acquisition are a ofrelative 20 October measurement 2017 as the on master both the scene spatial for temporal and temporal reference, domain, and thewe VRASused in continuous our PSI the GPS SAR station acquisition (Figure 3of) as20 aOctober spatial reference2017 as the point. master VRAS scene has for data temporal for four reference, years in theand period the VRAS 2013–2016, continuous is 80 kmGPS away station from (Figure Thessaloniki, 3) as a spatial and is reference located in point. an area VRAS for which has data there for is four no prioryears evidence in the period that it2013–2016, has been deforming.is 80 km away VRAS from data Thessaloniki, shows that and movement is located in in the an Up area direction for which is negligiblethere is no (~0.1 prior mm evidence/yr) with that some it has seasonal been movement,deforming. asVRAS expected. data shows The stability that movement of the VRAS in stationthe Up ensuresdirection us is that negligible we do not(~0.1 need mm/yr) to take with account some ofseasonal a constant movement, bias in our as velocitiesexpected. estimations.The stability of the VRASThe station results ensures indicate us that that there we isdo surface not need deformation to take account at the areas of a of constant interest, bias enclosed in our by velocities the blue rectanglesestimations. shown in Figure3. More specifically in the area of Kalochori–Sindos there is a subsidence signal, OreokastroThe results has an indicate uplifting that trend there while is a surface subsidence deforma trendtion appears at the to existareas at of the interest, basin of enclosed Anthemountas. by the Ablue new rectangles area not previously shown in known Figure to 3. deform More isspecific the Neaally Moudania in the area plain of (Figures Kalochori–Sindos1 and3). there is a subsidenceThe following signal, sectionsOreokastro focus has on an each uplifting of these trend areas, while analyze a subsidence the ﬁndings trend from appears InSAR to processingexist at the andbasin ground of Anthemountas. truth data when A new available, area not and previous attempt toly provideknown to interpretations deform is the on Nea the reportedMoudania signals. plain (Figures 1 and 3). 3.1. Kalochori–SindosThe following sections Region focus on each of these areas, analyze the findings from InSAR processing and ground truth data when available, and attempt to provide interpretations on the reported The villages of Kalochori and Sindos are located at the west of Thessaloniki (Figure1). The area is signals. known for the intense industrial activity. It is located on Quaternary deposits (Figure4) with the sediments consisting of sands, clays, and silty clays [25,26]. The upper 5 to 15 m of the Quaternary deposits are 3.1. Kalochori–Sindos Region occupied by fine-to-medium grained sand intercalated by silty sand horizons. Underneath, a 25 to 35 m thickThe black-grey villages silty of Kalochori sand horizon and with Sindos organic are located materials at the extends, west of forming Thessaloni an impermeableki (Figure 1). boundary, The area determinantis known for for the the intense hydrogeological industrial activity. conditions It is oflocated the site. on TheQuaternary lower strata, deposits down (Figure to the 4) Neogene with the bedrock,sediments consist consisting of alternating of sands, brown clays, sandand silty and black-greyclays [25,26]. silty The sand upper horizons 5 to [1526 m,27 of]. the Quaternary deposits are occupied by fine-to-medium grained sand intercalated by silty sand horizons. Underneath, a 25 to 35 m thick black-grey silty sand horizon with organic materials extends, forming an impermeable boundary, determinant for the hydrogeological conditions of the site. The lower

Remote Sens. 2020, 12, x FOR PEER REVIEW 6 of 27 strata,Remote Sens.down2020 to, 12the, 2396 Neogene bedrock, consist of alternating brown sand and black-grey silty sand6 of 26 horizons [26,27].

Figure 4. Geological map of the broader area of Kalochori–Sindos, source: [28] and [13] Figure(modified 4. Geological from [14]). map of the broader area of Kalochori–Sindos, source: [28] and [13] (modified from [14]). From a hydrogeological point of view, two aquifers can be identified: a shallow aquifer unconfined withinFrom the a sandy hydrogeological layer of the upperpoint of strata, view, and two a deeper aquifers confined can be aquifer identified: system a withinshallow the aquifer brown unconfinedsand of the within lower strata.the sandy The layer quality of ofthe the upper water stra ofta, the and shallow a deeper aquifer confined is poor, aquifer in contrast system to thewithin high thequality brown water sand of of the the second lower strata. confined The aquifer quality system, of the water extending of the below shallow the aquifer impermeable is poor, organicin contrast silty tosand the high horizon. quality The water overexploitation of the second of theconfined confined aquifer aquifer system, has been extending proven below as the the main impermeable cause of the organicland subsidence silty sand phenomenonhorizon. The over occurringexploitation at the site.of the confined aquifer has been proven as the main cause Groundof the land truth subsidence techniques phen hadomenon been used occurring to study at the the subsiding site. pattern that was known since the 1960sGround [25,29 –truth32]. Thetechniques 1990s InSAR had been displacement used to study was the first subsiding identified pattern by Raucoules that was et known al. [10] since using theacquisitions 1960s [25,29–32]. from the The European 1990s InSAR Remote displacement sensing Satellite was first (ERS) identified satellites. by Raucoules A detailed et al. investigation [10] using acquisitionsby Raspini from et al. [the13] European used the ERSRemote satellite sensing results Satellite (1990s (ERS) dataset), satellites. together A detailed with in-situ investigation data for by the Raspiniinterpretation et al. [13] of the used detected the ERS subsidence. satellite results However, (1990s in a laterdataset), study, together an uplifting with patternin-situ wasdata identified for the interpretationin Svigkas et of al. the [11 ]detected corresponding subsidence. to the Howeve post-2000r, period,in a later covering study, thean Environmentaluplifting pattern Satellite was identified(ENVISAT) in Svigkas era. This et change al. [11] in corresponding the deformation to the trend post-2000 was attributed period, tocovering the natural the Environmental recharge of the Satelliteaquifers, (ENVISAT) due to the satellite cessation era. of This groundwater change in pumpingthe deformation [14]. A trend Global was Navigation attributed Satellite to the natural System recharge(GNSS) studyof the covering aquifers, the due period to the 2013–2015 cessation showed of groundwater a mixture pumping of subsidence [14]. and A Global uplift displacementNavigation Satellitepattern System [33]. (GNSS) study covering the period 2013–2015 showed a mixture of subsidence and uplift Thedisplacement Sentinel-1 pattern interferometric [33]. results presented in Figure5 indicate that the uplifting trend of theThe 2000s Sentinel-1 has been interferometric reversed again results and now presented a new in period Figure of 5 subsidenceindicate that is the occurring uplifting in trend the area. of theMaximum 2000s has displacement been reversed values again in theand vicinity now a of new Kalochori period are of ofsubsidence the order ofis occurring10 mm/yr, in while the area. in the − Maximumbroader area, displacement maximum values values in reach the avici ratenity of of14 Kalochori mm/yr. are of the order of −10 mm/yr, while in the broader area, maximum values reach a rate −of −14 mm/yr. Hydrogeological data of the region enabled the validation and interpretation of the PSI Sentinel-1 based results. The drill data of the same period indicate a decrease in the underground

Remote Sens. 2020, 12, x FOR PEER REVIEW 7 of 27 water levels (Figure 6), which in most drills appears to have been initiated between 2016 and 2017.

TheRemote amount Sens. 2020 of ,dropdown12, 2396 from November 2015 to November 2019 agrees spatially with the areas7 ofof 26 significant displacement (Figure 7) depicted by the PSI processing.

FigureFigure 5. 5. PSIPSI ground velocityvelocity results results for for Kalochori–Sindos Kalochori–Sindos as derivedas derived from from the Sentinel-1 the Sentinel-1 PSI analysis. PSI analysis.Blue rectangle Blue rectangle is the area is shownthe area also shown in Figure also6 in (inset Figures map) 6 and(inset Figure map)7. and The 7. histogram The histogram at the bottom at the bottomrow highlights row highlights the prevailing the prevailing subsidence subsidence trend intrend the region.in the region.

Hydrogeological data of the region enabled the validation and interpretation of the PSI Sentinel-1 based results. The drill data of the same period indicate a decrease in the underground water levels (Figure6), which in most drills appears to have been initiated between 2016 and 2017. The amount of dropdown from November 2015 to November 2019 agrees spatially with the areas of signiﬁcant displacement (Figure7) depicted by the PSI processing.

Remote Sens. 2020, 12, x FOR PEER REVIEW 8 of 27

Remote Sens. 2020,, 12,, 2396x FOR PEER REVIEW 8 of 27 26

Figure 6. In-situ underground water level measurements conducted during the Sentinel-1 PSI

monitoring period. The green rectangles in the inset map are the locations of the drills; the area of the insetFigure map 6. isIn-situ shown underground in the blue rectangle water level of measurements Figure 5. conducted during the Sentinel-1 PSI monitoring Figure 6. In-situ underground water level measurements conducted during the Sentinel-1 PSI period. The green rectangles in the inset map are the locations of the drills; the area of the inset map is monitoring period. The green rectangles in the inset map are the locations of the drills; the area of the shown in the blue rectangle of Figure5. inset map is shown in the blue rectangle of Figure 5.

Figure 7. Underground water level changes of the deep conﬁned aquifer between Novemeber Figure2019–November 7. Underground 2015 atwater the broader level changes area of of Kalochori. the deep Thisconfined a zoom aquifer of the between area; its Novemeber extent is shown 2019 with – Novemberthe blue rectangle 2015 at the in broader Figure5. area of Kalochori. This a zoom of the area; its extent is shown with the blue rectangle in Figure 5. Figure 7. Underground water level changes of the deep confined aquifer between Novemeber 2019 – November 2015 at the broader area of Kalochori. This a zoom of the area; its extent is shown with the blue rectangle in Figure 5.

Remote Sens. 2020, 12, 2396x FOR PEER REVIEW 9 of 27 26

In order to have a holistic and temporally complete view of the phenomenon that has been occurringIn order at Kalochori to have a over holistic the andlast decades, temporally a gr completeaph presenting view of the the groundwater phenomenon level that of has a drill, been togetheroccurring with at Kalochori the surface over deformation the last decades, estimated a graphat the exact presenting neighborhood the groundwater of that drill, level is presented of a drill, intogether Figure with 8. The the plot surface shows deformation that the subsidence estimated ta atking the exactplace neighborhood during the 1990s of that was drill, followed is presented by a rise in ofFigure the 8hydraulic. The plot head—due shows that theto natural subsidence rebound—which taking place during in turn the resulted 1990s was in followeda surface by uplift a rise the of followingthe hydraulic years head—due [14]. The to current natural analysis rebound—which of satellite in turnand resultedin-situ hydrogeology in a surface uplift data, the indicates following a potentialyears [14]. onset The currentof a new analysis cycle of of subsidence, satellite and deriv in-situed from hydrogeology the recent data,downward indicates trend a potential of the aquifer onset waterof a new level cycle (Figure of subsidence, 8). The subsidence derived trend from of the Sent recentinel-1 downward SAR results trend appeared of the with aquifer a time-lag water of level ~2 (Figureyears with8). Therespect subsidence to the initiation trend of of Sentinel-1 the lowering SAR of results the hydraulic appeared head. with Similarly a time-lag a of~2-year ~2 years time-lag with hadrespect been to also the initiationidentified of at the all lowering former changes of the hydraulic of the trend head. of Similarlythe curves; a ~2-yearnamely time-lagwhen the had aquifer been startedalso identiﬁed to recover at alland former when changesthe recovery of the trend trend was of reduced the curves; during namely 2006 when (Figure the 8). aquifer started to recover and when the recovery trend was reduced during 2006 (Figure8).

Figure 8. PSI based surface displacement vs. in-situ underground water level. Older data, corresponding Figureto the period 8. PSI 1990–2014, based surface were derived displacement from the interefrometricvs. in-situ underground processing of water ERS and level. ENVISAT Older data data, as wellcorresponding as in-situ waterto the levelperiod measurements 1990–2014, were as reported derived infrom [14 ].the Whenever interefrometric there is processing a trend change of ERS in and the ENVISATunderground data water as well level, as in-situ surface water deformation level measur followsements this changeas reported with in an [14]. approximate Whenever time-lag there is ofa trendtwo years. change in the underground water level, surface deformation follows this change with an approximate time-lag of two years. 3.2. Oreokastro Industrial Zone 3.2. OreokastroOreokastro Industrial is located Zone at the exact proximity of Thessaloniki (Figure1) and is an area strongly relatedOreokastro with the industrialis located activitiesat the exact of theproximity city. The of town Thessaloniki was founded (Figure on a1) Pliocene and is an lake area and strongly ﬂuvial relateddeposits. with The the industrial industrial zone activities of Oreokastro of the city. is The on Pliocenetown was clays founded intercalated on a Pliocene by sand, lake lying and fluvial above deposits.the Pre-Alpine The industri bedrockal (Figureszone of Oreokastro9 and 10). The is on main Pliocene tectonic clays features intercalated are the by Efkarpia sand, lying fault above and thethe AsvestochoriPre-Alpine bedrock fault, both (Figures considered 9 and as 10). potentially The main active tectonic [34]. Nofeatures major are seismic the Efkarpia event is knownfault and to exist the Asvestochoriand the area is fault, mainly both related considered with microseismicity as potentially active [35–37 ].[34]. No major seismic event is known to exist InSARand the techniques area is mainly had previouslyrelated with revealed microseismicity that the area [35–37]. has been deforming [10]. The subsidence of the 1990s was followed by an uplifting trend after 2000. In Svigkas et al. [15], by combining remote sensing analysis with seismological, geological, and hydrogeological data of the area, it was shown that the observed displacement—which also caused building failures—was related with the activity of aquifers. Also, the displacement was found to be a fault-controlled pattern, in a sense that the faults at the area were most likely acting like water-ﬂow barriers. Sentinel-1 PSI results indicate a continuation of the uplifting pattern (Figure 11), previously detected to occur during the 2002–2010 ENVISAT based monitoring period. Specifically, the maximum uplifting rate was found to be as high as 8 mm/yr, which is essentially the same order of value as the one observed during the ENVISAT era (9 mm/yr).

Remote Sens. 2020, 12, x FOR PEER REVIEW 10 of 27 Remote Sens. 2020, 12, 2396 10 of 26

FigureFigure 9.9.The The geology geology of of the the broader broader area area of of Oreokastro Oreokastro (blue (blue rectangle). rectangle). Main Main tectonic tectonic features features are theare Asvestochorithe Asvestochori fault fault (F-As) (F-As) and theandEfkarpia the Efkarpia fault faul (F-E).t (F-E). Drill Drill data data are shown are shown in Figure in Figure 10, source: 10, source: [38] (modiﬁed[38] (modified from from [15]). [15]).

To analyze with higher detail the displacement pattern, velocity profiles were created (Figure 12). It can be seen that in all the profiles under study the Asvestochori fault (Figure9) is a ffecting the displacement trend.

RemoteRemote Sens. Sens.2020 2020, 12, ,12 2396, x FOR PEER REVIEW 1111 of of 26 27

Figure 10. Stratigraphy of the broader area of Oreokastro (Figure9). Borehole data are archives of the Figure 10. Stratigraphy of the broader area of Oreokastro (Figure 9). Borehole data are archives of the Directory of Environment, Thessaloniki Prefecture (modified from [15]). Directory of Environment, Thessaloniki Prefecture (modified from [15]). Based on the ENVISAT results, Svigkas et al. [15] proposed that at the area there is an unknown featureInSAR highlighted techniques by abrupt had previously changes in revealed the deformation that the trends. area has The been interferometric deforming analysis[10]. The conductedsubsidence showed of the that1990s between was followed the uplifting by an pattern uplifting and trend the Asvestochoriafter 2000. In fault, Svigkas there et was al. [15], a thin by subsidingcombining zone remote just north sensing of the analysis fault and with before seismological, the expression geological, of the uplift. and hydrogeological The same pattern data appears of the alsoarea, in theit was present shown study that (Figure the observed 11) of the displace Sentinel-1ment—which based interefrometric also caused processing.building failures—was However thisrelated time thewith deformation the activity pattern of aquifers. presents Also, a slightly the displacement different geometry was found than to the be one a fault-controlled of the period 2003–2010,pattern, in a a phenomenon sense that the that faults might at the be relatedarea were to themost di fflikelyerence acting of underground like water-flow water barriers. level activity, duringSentinel-1 the two monitoring PSI results periods.indicate Ana continuation explanation of for the this uplifting could be pattern the diff (Figureerent amounts 11), previously of the undergrounddetected to wateroccur level during uprising the that2002–2010 might cause ENVISAT a different based surface monitoring displacement period. pattern, Specifically, depending the alsomaximum on the various uplifting permeabilities rate was found of the to earthbe as stratahigh as that 8 mm/yr, the water which flow is meets. essentially the same order of value as the one observed during the ENVISAT era (9 mm/yr).

Remote Sens. 2020, 12, 2396 12 of 26 Remote Sens. 2020, 12, x FOR PEER REVIEW 12 of 27

Figure 11. PSPS velocities of the broaderbroader areaarea ofof Oreokastro.Oreokastro. Green Green lines lines are are velocity velocity proﬁles profiles presented presented in inFigure Figure 12. 12. AA’, AA’, BB’, andBB’, CC’and and CC’ numbers and numbers 1–4 are velocity1–4 are proﬁlesvelocity and profiles points and of interest points respectively,of interest respectively,noted also in noted Figure also 12. Faultsin Figure are 12. from Faults the Nationalare from Observatorythe National ofObservatory Athens (NOA) of Athens fault database (NOA) fault [39]. database [39].

Remote Sens. 2020, 12, x FOR PEER REVIEW 13 of 27

To analyze with higher detail the displacement pattern, velocity profiles were created (Figure Remote Sens. 2020, 12, 2396 13 of 26 12). It can be seen that in all the profiles under study that the Asvestochori fault (Figure 9) is affecting the displacement trend.

Figure 12. Velocity proﬁles in Oreokastro. Vertical black lines 1 and 2 in AA’ and BB’ respectively indicate locations attributed to the Asvestochori fault. Vertical Lines 3 and 4 in proﬁle BB’ and CC’ respectively mark terminations of the Oreokastro uplifting displacement pattern. Remote Sens. 2020, 12, 2396 14 of 26

3.3. Anthemountas Basin Anthemountas basin is located west of the city of Thessaloniki (Figure1). In the area the development trends had already started before the 1990s [40]. Here lay the longest tectonic structures close to the city of Thessaloniki; the northern branch of Anthemountas fault (F1 in Figure 13) that has a length of 21 km, considered inactive [34], and the south branch of Anthemountas fault (F-2). Morphotectonic studies indicate it as being an active fault structure [34,41]. Regarding the geological structure of the site, the top Quaternary formations consist of alterations of clasticRemote and fineSens. 2020 sediment, 12, x FOR PEER [34, 42REVIEW,43]. At the Quaternary deposits (Figures 13 and 1414), of three 27 main horizons can beFigure identified: 12. Velocity a) profiles coarse in toOreokastro. fine sand Vertical b) clay black to lines silty 1 and clay 2 in horizon, AA’ and BB’ and respectively c) Neogene deposits characterizedindicate by coarse locations to fine attributed sands. to the The Asvestochori Neogene fa depositsult. Vertical consist Lines 3 and of a4 topin profile sand BB’ and and gravelCC’ layer and a bottom clay-marlrespectively layer mark [42 terminations,44]. The areaof the hasOreokastro a Mesozoic uplifting bedrockdisplacement where pattern. phillites, gneiss, and granites lie. The thickness of the sediments decreases when moving towards the east. Based on the ENVISAT results, Svigkas et al. [15] proposed that at the area there is an unknown In thefeature upper highlighted strata of theby Quaternaryabrupt changes deposits, in the de thereformation is a shallowtrends. The aquifer interferometric system interacting analysis with a deeper semi-confinedconducted showed aquifer that extendingbetween the downuplifting to pattern the Neogene and the Asvestochori formations [fault,45]. there The waterwas a thin demands of the area havesubsiding caused zone a just decrease north of of the the fault groundwater and before level.the expression of the uplift. The same pattern Raucoulesappears etalso al. in [10the] present used acquisitionsstudy (Figure 11) of ERSof the satellites Sentinel-1 andbased by interefrometric applying InSARprocessing. time-series However this time the deformation patter presents a slightly different geometry than the one of the detectedperiod subsidence 2003–2010, during a phenomenon the 1990s. that Itmight was be found related thatto the the difference displacements of underground were water extending level from the coastalactivity, zone during of the the graben two monitoring throughout periods. the An entire explanation structure, for this up could to thebe the far different East of amounts the basin [16]. Raspini etof al. the [12 underground] attributed water the subsidencelevel uprising phenomena that might cause of the a different 1990s–that surface appeared displacement to have pattern, also caused building failures–todepending also aquifer on the overpumping.various permeabilities The subsidenceof the earth strata in Anthemountas that the water flow continued meets. after 2000 and had been3.3. occurring Anthemountas continuously Basin and increasingly for at least 20 years [17]. The observations based on PSI processing of Sentinel-1 images are presented in this study and Anthemountas basin is located west of the city of Thessaloniki (Figure 1). In the area the indicate thatdevelopment the subsidence trends had phenomenon already started is still before active the (Figure 1990s [40].15). Here The maximumlay the longest subsidence tectonic value at the site appearsstructures to close be to18 the mm city/ yr,of Thessaloniki; the same as the the northern maximum branch valuesof Anthemountas previously fault detected (F1 in Figure during the − ENVISAT13) imagery that has eraa length [17]. of Detailed21 km, considered velocity inactive profiles [34], are and shown the south in branch Figure of 16 Anthemountas. fault (F-2). Morphotectonic studies indicate it as being an active fault structure [34,41].

Figure 13. Geological map of Anthemountas graben, source: [46] (modiﬁed from [17]). Figure 13. Geological map of Anthemountas graben, source: [46] (modified from [17]).

Regarding the geological structure of the site, the top Quaternary formations consist of alterations of clastic and fine sediment [34,42,43]. At the Quaternary deposits (Figures 13 and 14), three main horizons can be identified: a) coarse to fine sand b) clay to silty clay horizon, and c) Neogene deposits characterized by coarse to fine sands. The Neogene deposits consist of a top sand and gravel layer and a bottom clay-marl layer [42,44]. The area has a Mesozoic bedrock where

Remote Sens. 2020, 12, x FOR PEER REVIEW 15 of 27

Remotephillites, Sens. gneiss,2020, 12 ,and 2396 granites lie. The thickness of the sediments decreases when moving towards15 of the 26 east.

Remote Sens.Figure 2020, 14.12, xBorehole FOR PEER REVIEW data of Anthemountas basin, source: [12] (modiﬁed from [1716]). of 27 Figure 14. Borehole data of Anthemountas basin, source: [12] (modified from [17]).

In the upper strata of the Quaternary deposits, there is a shallow aquifer system interacting with a deeper semi-confined aquifer extending down to the Neogene formations [45]. The water demands of the area have caused a decrease of the groundwater level. Raucoules et al. [10] used acquisitions of ERS satellites and by applying InSAR time-series detected subsidence during the 1990s. It was found that the displacements were extending from the coastal zone of the graben throughout the entire structure, up to the far East of the basin [16]. Raspini et al. [12] attributed the subsidence phenomena of the 1990s–that appeared to have also caused building failures–to aquifer overpumping. The subsidence in Anthemountas continued after 2000 and had been occurring continuously and increasingly for at least 20 years [17]. The observations based on PSI processing of Sentinel-1 images are presented in this study and indicate that the subsidence phenomenon is still active (Figure 15). The maximum subsidence value at the site appears to be -18 mm/yr, the same as the maximum values previously detected during the ENVISAT imagery era [17]. Detailed velocity profiles are shown in Figure 16. An important infrastructure located at the Anthemountas graben is the International Airport of Thessaloniki (cyan rectangle Figure 15). It is shown here that during the monitoring period, the airport has a subsidence signal reaching −12 mm/yr (Figure 17). This is located at the intersection of the two airport runways. The current terminal of the airport was found to have been subsiding with a rate of −8 mm/yr, during 2015–2019.

Figure 15.FigureVelocity 15. Velocity results results for for the the Anthemountas Anthemountas graben. graben. Cyan Cyan rectangle rectangle indicates indicates the location the of location the of the MakedoniaMakedonia International International Airport Airport (Figure (Figure 1717).).

Remote Sens. 2020, 12, 2396 16 of 26 Remote Sens. 2020, 12, x FOR PEER REVIEW 17 of 27

Figure 16. Cont.

Remote Sens. 2020, 12, x FOR PEER REVIEW 18 of 27 Remote Sens. 2020, 12, 2396 17 of 26

FigureFigure 16. 16.Graphs Graphs are theare PSI the velocity PSI velocity profiles profiles (shown with(shown green with lines green at the toplines map) at ofthe the top Anthemountas map) of the graben.Anthemountas The vertical graben. white andThe blackvertical lines white in the and graphs black indicate lines in the the faults, graphs as they indicate appear the at faults, the top as map. they appear at the top map.

Remote Sens. 2020, 12, 2396 18 of 26 Remote Sens. 2020, 12, x FOR PEER REVIEW 19 of 27

FigureFigure 17. 17. PSPS Sentinel-1 Sentinel-1 results of the MakedoniaMakedonia InternationalInternational Airport. Airport. Graphs Graphs are are velocity velocity profiles profiles of ofthe the two two runways. runways. Blue Blue profile profile swath swath shown shown in the in map the ismap shown is shown at the topat the graph, top redgraph, profile red swath profile of swaththe map of the is shown map is at shown the bottom at the graph. bottom graph.

Remote Sens. 2020, 12, 2396 19 of 26

An important infrastructure located at the Anthemountas graben is the International Airport of Thessaloniki (cyan rectangle Figure 15). It is shown here that during the monitoring period, the airport has a subsidence signal reaching 12 mm/yr (Figure 17). This is located at the intersection of the two − airport runways. The current terminal of the airport was found to have been subsiding with a rate of 8 mm/yr, during 2015–2019. − 3.4. NeaRemote Moudania Sens. 2020, 12, x FOR PEER REVIEW 20 of 27

Nea3.4. Nea Moudania Moudania plain drew the attention of the researchers for the first time, due to recently occurringNea deformation Moudania pattern plain drew identified the attention at the site. of the The researchers wider region for the belongs first time, to the due Paeonian to recently subzone, part ofoccurring the Axios deformation geotectonic pattern zone. Theidentified Litho-stratigraphy at the site. The of thewider study region area belongs is divided to intothe Paeonian the following formationssubzone, [47 part,48]. of the Axios geotectonic zone. The Litho-stratigraphy of the study area is divided into Quaternarythe following formations formations, [47,48]. located mainly near the coastal areas, including alluvial deposits (sands, clay),Quaternary coastal depositsformations, (beach located ridges mainly etc.), near lagoon the deposits coastal (sands,areas, including clayey sand), alluvial and deposits alluvial fans (Figure(sands, 18). Furthermore,clay), coastal deposits along the (beach streams, ridges two etc), terrace lagoon systems deposits can(sands, beidentified. clayey sand), The and lower alluvial consists mainlyfans of (Figure sands, 18). schists, Furthermore, and limestone along pebblesthe streams, and two the terrace upper ofsystems metamorphic can be identified. rock pebbles. The lower Neogene formationsconsists cover mainly the of pre-Neogenesands, schists, bedrockand limestone formations. pebbles and They the can upper be distinguishedof metamorphic into rock the pebbles. lower red Neogene formations cover the pre-Neogene bedrock formations. They can be distinguished into the clays series, occupying most of the study area and the upper marls with sandstone series, outcropping lower red clays series, occupying most of the study area and the upper marls with sandstone series, at theoutcropping west of the at study the west area. of the The study red claysarea. The series red consists clays series mainly consists of redmainly silty of clays red silty with clays fine-grained with quartzfine-grained and mica, quartz intercalated and mica, by sand,intercalated marl, andby sand, conglomerate marl, and lances.conglomerate The marl lances. with The sandstone marl with series consistssandstone of altering series bedsconsists of sand,of altering clay, beds and of clay-marl. sand, clay, Regarding and clay-marl. the Regarding bedrock, itthe consists bedrock, of it party recrystallizedconsists of bluish party limestone,recrystallized greenish-brown bluish limeston twoe, greenish-brown mica gneiss, andtwo muscovitemica gneiss, gneiss. and muscovite gneiss.

FigureFigure 18. 18.Geological Geological map of of Nea Nea Moudania. Moudania.

Remote Sens. 2020, 12, 2396 20 of 26

Two aquifer systems can be identified at the study area: a shallow unconfined aquifer and a deep confined aquifer [49]. The shallow phreatic aquifer mainly occupies the alluvial Quaternary deposits and the deep confined aquifer the permeable strata intercalated into the Neogene formations, namely sand, Remote Sens. 2020, 12, x FOR PEER REVIEW 21 of 27 conglomerate, and sandstone layers. The confined aquifer system’s thickness exceeds 200 m in depth [50]. Thebedrock water formations, provided laterallyas well as by the the water karstic infiltrat aquifersed from of the the bedrock streams formations,and the precipitation, as well as are the the water infiltratedmain sources from of the water streams for the and aquifers. the precipitation, are the main sources of water for the aquifers. TheThe Sentinel-1Sentinel-1 based based PSI PSI results results indicate indicate a significant a signiﬁcant surface surface displacement displacement (Figure (Figure19). The 19). Thesubsidence subsidence rates rates have have a maxi a maximummum value value of -23 of mm/yr, -23 mm during/yr, during the time the span time of span 2015–2019. of 2015–2019.

Figure 19. PS velocity map in the broader area of Nea Moudania. Figure 19. PS velocity map in the broader area of Nea Moudania. The small extent of the alluvial deposits, in combination with the overexploitation of the shallow The small extent of the alluvial deposits, in combination with the overexploitation of the aquifer over the past decades has led to the draining of the phreatic aquifer [51]. Since the 1990s, shallow aquifer over the past decades has led to the draining of the phreatic aquifer [51]. Since the 1990s, numerous deep wells have been constructed in the area to meet the water needs [52]. The confined aquifer system of the Neogene formations is considered to be the main source of underground water.

Remote Sens. 2020, 12, 2396 21 of 26 numerous deep wells have been constructed in the area to meet the water needs [52]. The conﬁned aquiferRemote systemSens. 2020 of, 12 the, x FOR Neogene PEER REVIEW formations is considered to be the main source of underground22 of 27 water. According to data provided by Veranis and Xatzikirkou [50] during the late 2000s, the distribution of the piezometricAccording to lines data have provided been clearlyby Veranis indicating and Xa thetzikirkou initiation [50] of aduring sea water the intrusion,late 2000s, alongthe the coastaldistribution area close of tothe Sozopoly, piezometric as well lines as have a depression been clearly cone atindicating the plane the area initiation NW of Neaof a Moudaniasea water town, intrusion, along the coastal area close to Sozopoly, as well as a depression cone at the plane area NW both due to the overexploitation of the aquifers (Figure 20). The depression cone had been covering of Nea Moudania town, both due to the overexploitation of the aquifers (Figure 20). The depression the plane area extending up to the north close to Portaria and Zographou villages with a maximum cone had been covering the plane area extending up to the north close to Portaria and Zographou depressionvillages with of approximately a maximum depressio -15 m. Accordingn of approximately to the 2016 -15 data m. According provided byto Gavrihlidouthe 2016 data [provided53], since 2008 theby conditions Gavrihlidou radically [53], since changed. 2008 the The conditions zero level radically contour changed. line (blue The line zero in level Figure contour 20) moved line (blue half way towardsline in NeaFigure Triglia 20) moved village half and way the depressiontowards Nea cone Triglia extended village much and the more depression to the NW cone including extended clearly Portariamuch more and Zographouto the NW including villages (Figureclearly Portaria 18). and Zographou villages (Figure 18).

Figure 20. The confined aquifer piezometric contour lines of the broader area of Nea Moudania. Map at the bottom refers to May 2008 [50]. The magenta dashed rectangle is also shown enlarged in the Figure 20. The confined aquifer piezometric contour lines of the broader area of Nea Moudania. Map two inset maps, where the differences between May 2008 and May 2016—modified by the authors at the bottom refers to May 2008 [50]. The magenta dashed rectangle is also shown enlarged in the from [53]—contour patterns can be compared together also with the PS velocities. Blue line is the two inset maps, where the differences between May 2008 and May 2016—modified by the authors 0-levelfrom contour[53]—contour line. Waterpatterns flow can is be indicated compared by together Greenarrows. also with the PS velocities. Blue line is the 0-level contour line. Water flow is indicated by Green arrows.

Remote Sens. 2020, 12, 2396 22 of 26

Unluckily, the available groundwater level data referring to 2016, do not allow the production of piezometric curves covering the same area with the one covered by the 2008 data, but nevertheless an expansion of the depression cone is clearly evident (Figure 20). Also, the changes observed in the flow direction of the ground water prove the expansion of the depression cone. As shown from the comparisons of the two upper inset maps in Figure 20 (Contours May 2008 and May 2016), the flow direction heading to the sea during 2008 changed by turning to a SE flow, towards the center of the plane area. As the extent of the depression cone coincides with the spatial distribution of the PS indicating subsidence, it can be inferred that the deformations are connected with the consolidation of the formations, triggered by the overexploitation of the aquifer.

4. Discussion In all cases presented, the detected surface displacements are mainly related to the trends of the hydraulic head. During the recharge of the underground water level, there is an increase of the pore fluid pressure and a decrease of the effective stresses. This is the period where we can have a surface uplift, like for example the one identified in Oreokastro during the 2000s and also from ~2015 to 2019 or like in Kalochori–Sindos during the 2000s. On the contrary, a decrease of the groundwater level causes an increase of the effective stresses and as a result, subsidence. As identified by the InSAR studies this was occurring at Kalochori–Sindos during the 1990s and from ~2015 to 2019, in Oreokastro during the 1990s and in Athemountas during 1990s, 2000s, and during the period of 2015–2019. In several areas around the world artificial aquifer recharge strategies are followed aiming to overcome the subsidence phenomena. It has been shown that at least for the areas of Oreokastro and Kalochori–Sindos, the surface rebound and the return to a non-deforming state, can be achieved naturally. This could potentially be the result of the generation of rational water management. However, if the groundwater decline continues, the subsidence will continue and the failures will be ongoing in the future. Even though significant hazards have occurred at those areas, the detected natural recharge is a positive factor to be taken into account; at those areas, there is still a regime of rebound deformation. If at some point the underground water level decrease becomes severe enough to cause stresses that exceed the preconsolidation stresses, a rearrangement of the granural material of the aquitards could occur. Then, the area will start experiencing inelastic deformation (e.g., [54–58]). In this case, the results would be irreversible or only partially reversible, even in the case of an aquifer recharge [59].

5. Conclusions Based on Sentinel-1 and new in-situ underground water level data, we analyzed several areas in the broader area of the city of Thessaloniki, which were found to be deforming during the years of 2015–2019. For the area of Kalochori–Sindos, even though there was a natural rebound after 2000—pausing the subsidence rates of the past—this study showed that a clear subsidence trend has been initiated again. It also became clear that the surface of the area is very sensitive to the changes of the aquifer trends. Moreover, the study shows that the broader area of Oreokastro is still uplifting since the pause of its subsidence trend that occurred after the 1990s. The groundwater level activity enabled us to see discontinuities whose nature and cause could be related to tectonic or other types of geological structures. At this stage, we consider that a geophysical survey at Oreokastro might provide additional insights. The displacement of Anthemountas basin is still ongoing, with the same magnitude as the values detected during the 2000s. Another important fact presented here is that the displacement trend still continues throughout the entire basin and not only on the coastal part of Anthemountas. Significant subsidence occurs also at the area of the International Airport of Thessaloniki where as before, a continuous monitoring seems necessary. A new area that was found for the first time to be affected by subsidence is the Nea Moudania plain. Our analysis highlights that it is also attributed to an anthropogenically derived hazard, linked with water overpumping. Remote Sens. 2020, 12, 2396 23 of 26

Since most of the study areas presented here have been deforming, occasionally with diﬀerent trends, for at least ~25 years, it becomes clear that a water management plan should be a forefront priority in the agenda of the decision makers.

Supplementary Materials: The following are available online at http://www.mdpi.com/2072-4292/12/15/2396/s1, Figure S1: Standard deviation of the PS S-1 results. Author Contributions: Conceptualization, N.S., C.L, I.P., and C.H.K.; formal analysis, S.A., I.P., C.T., N.S.; data curation, P.T., T.S., S.A., I.P., C.T., N.S.; writing—original draft preparation, N.S., C.L; writing—review and editing, C.L, I.P., C.H.K., S.A., C.T.; T.S.; visualization, N.S., C.L., I.P., S.A.; funding acquisition, C.H.K. All authors have read and agreed to the published version of the manuscript. Funding: This research was supported by the Next Generation GEOSS for Innovation Business (NextGEOSS), European Commission, grant agreement ID 730329. Acknowledgments: European Space Agency and Copernicus EU are thanked for the provision of Sentinel-1 images. Hydrogeological data for Kalochori–Sindos is a courtesy of the Water Supply and Sewerage Company of Thessaloniki (EYATH). The faults are from Zervopoulou (2010) and the NOA fault database (Ganas et al. 2013). The authors would like to thank Tree-Company and Dionysos Satellite Obesrvatory of National Technical University of Athens for providing GPS velocities for VRAS station. Conﬂicts of Interest: The authors declare no conﬂict of interest.

References

1. Loupasakis, C.; Angelitsa, V.; Rozos, D.; Spanou, N. Mining geohazards—Land subsidence caused by the dewatering of opencast coal mines: The case study of the Amyntaio coal mine, Florina, Greece. Nat. Hazards 2013, 70, 675–691. [CrossRef] 2. Psimoulis, P.; Ghilardi, M.; Fouache, E.; Stiros, S. Subsidence and evolution of the Thessaloniki plain, Greece, based on historical leveling and GPS data. Eng. Geol. 2007, 90, 55–70. [CrossRef] 3. Galloway, D.L.; Hoffmann, J. The application of satellite differential SAR interferometry-derived ground displacements in hydrogeology. Hydrogeol. J. 2006, 15, 133–154. [CrossRef] 4. Lu, Z.; Danskin, W.R. InSAR analysis of natural recharge to define structure of a ground-water basin, San Bernardino, California. Geophys. Res. Lett. 2001, 28, 2661–2664. [CrossRef] 5. Alatza, S.; Papoutsis, I.; Paradissis, D.; Kontoes, C.; Papadopoulos, G.A. Multi-Temporal InSAR Analysis for Monitoring Ground Deformation in Amorgos Island, Greece. Sensors 2020, 20, 338. [CrossRef] 6. Triantafyllou, A.I.; Vergos, G.S.; Tziavos, I.N. Contribution of Sentinel-1 Dinsar to the Determination of Vertical Deformation and Height System Monitoring. In Proceedings of the 4th Joint International Symposium on Deformation Monitoring (JISDM), Athens, Greece, 15–17 May 2019. 7. Papoutsis, I.; Kontoes, C.; Paradissis, D. Multi-Stack Persistent Scatterer Interferometry Analysis in Wider Athens, Greece. Remote Sens. 2017, 9, 276. [CrossRef] 8. Foumelis, M.; Papageorgiou, E.; Stamatopoulos, C.A. Episodic ground deformation signals in Thessaly Plain (Greece) revealed by data mining of SAR interferometry time series. Int. J. Remote Sens. 2016, 37, 3696–3711. [CrossRef] 9. Mouratidis, A.; Briole, P.; Ilieva, M.; Astaras, T.; Rolandone, F.; Baccouche, M. Subsidence and deformation phenomena in the vicinity of Thessaloniki (N. Greece) monitored by Envisat/ASAR interferometry. In Proceedings of the Fringe 2009 Workshop: “Advances in the Science and Applications of SAR Interferometry”, Frascati, Italy, 30 November–4 December 2009. 10. Raucoules, D.; Parcharidis, I.; Feurer, D.; Novalli, F.; Ferretti, A.; Carnec, C.; Lagios, E.; Sakkas, V.; Le Mouélic, S.; Cooksley, G.; et al. Ground deformation detection of the greater area of Thessaloniki (Northern Greece) using radar interferometry techniques. Nat. Hazards Earth Syst. Sci. 2008, 8, 779–788. [CrossRef] 11. Svigkas, N.; Papoutsis, I.; Loupasakis, C.; Kontoes, C.; Kiratzi, A. Geo-hazard Monitoring in Northern Greece Using InSAR techniques: The case Study of Thessaloniki. In Proceedings of Fringe 2015: Advances in the Science and Applications of SAR Interferometry and Sentinel-1 InSAR Workshop; European Space Agency: Paris, France, 2015. 12. Raspini, F.; Loupasakis, C.; Rozos, D.; Moretti, S. Advanced interpretation of land subsidence by validating multi-interferometric SAR data: The case study of the Anthemountas basin (Northern Greece). Nat. Hazards Earth Syst. Sci. 2013, 13, 2425–2440. [CrossRef] Remote Sens. 2020, 12, 2396 24 of 26

13. Raspini, F.; Loupasakis, C.; Rozos, D.; Adam, N.; Moretti, S. Ground subsidence phenomena in the Delta municipality region (Northern Greece): Geotechnical modeling and validation with Persistent Scatterer Interferometry. Int. J. Appl. Earth Obs. Geoinf. 2014, 28, 78–89. [CrossRef] 14. Svigkas, N.; Papoutsis, I.; Loupasakis, C.; Tsangaratos, P.; Kiratzi, A.; Kontoes, C.H. Land subsidence rebound detected via multi-temporal InSAR and ground truth data in Kalochori and Sindos regions, Northern Greece. Eng. Geol. 2016, 209, 175–186. [CrossRef] 15. Svigkas, N.; Papoutsis, I.; Loupasakis, C.; Tsangaratos, P.; Kiratzi, A.; Kontoes, C.H. InSAR time-series monitoring of ground displacement trends in an industrial area (Oreokastro—Thessaloniki, Greece): Detection of natural surface rebound and new tectonic insights. Environ. Earth Sci. 2017, 76.[CrossRef] 16. Svigkas, N.; Papoutsis, I.; Loupasakis, C.; Tsangaratos, P.; Kiratzi, A.; Kontoes, C.H. Radar Space Measurements of the Deforming Trends at Northern Greece Resulting from Underground Water Activity. In Advances in Remote Sensing and Geo Informatics Applications; Springer International Publishing: Cham, Switzerland, 2019; pp. 309–313. 17. Svigkas, N.; Loupasakis, C.; Tsangaratos, P.; Papoutsis, I.; Kiratzi, A.; Kontoes, C.H. A deformation study of Anthemountas graben (northern Greece) based on in situ data and new InSAR results. Arab. J. Geosci. 2020, 13, 1–13. [CrossRef] 18. Galloway, D.L.; Hudnut, K.W.; Ingebritsen, S.E.; Phillips, S.P.; Peltzer, G.; Rogez, F.; Rosen, P.A. Detection of aquifer system compaction and land subsidence using interferometric synthetic aperture radar, Antelope Valley, Mojave Desert, California. Water Resour. Res. 1998, 34, 2573–2585. [CrossRef] 19. Papoutsis. InSAR Greece: A national ground motion service, based on big Copernicus Sentinel-1 data. Remote Sens. 2020. under review. 20. Fattahi, H.; Agram, P.; Simons, M. A Network-Based Enhanced Spectral Diversity Approach for TOPS Time-Series Analysis. IEEE Trans. Geosci. Remote Sens. 2016, 55, 777–786. [CrossRef] 21. 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, L23611. [CrossRef] 22. Hooper, A.; Segall, P.; Zebker, H. Persistent scatterer interferometric synthetic aperture radar for crustal deformation analysis, with application to Volcán Alcedo, Galápagos. J. Geophys. Res. Space Phys. 2007, 112, B07407. [CrossRef] 23. Farr, T.G.; Kobrick, M. Shuttle radar topography mission produces a wealth of data. Eos Trans. Am. Geophys. Union 2000, 81, 583–585. [CrossRef] 24. Bekaert, D.; Walters, R.; Wright, T.J.; Hooper, A.; Parker, D.J. Statistical comparison of InSAR tropospheric correction techniques. Remote Sens. Environ. 2015, 170, 40–47. [CrossRef] 25. Hatzinakos, I.; Rozos, D.; Apostolidis, E. Engineering geological mapping and related geotechnical problems in the wider industrial area of Thessaloniki, Greece. In International Congress International Association of Engineering Geology; Price, D., Ed.; A. A. Balkema: Rotterdam, The Netherlands, 1990; pp. 127–134. 26. Rozos, D.; Apostolidis, E.; Christaras, B. Engineering-geological map of Thessaloniki wider area. Bull. Eng. Geol. Environ. 2004, 63, 103–108. 27. Loupasakis, C.; Rozos, D. Land subsidence induced by the overexploitation of the aquifers in Kalochori village—New approach by means of the computational geotechnical engineering. Bull. Geol. Soc. Greece 2010, 43, 1219–1229. [CrossRef] 28. IGME Institute of Geological and Mineralogical Exploration. Geological Map of Greece, Scale 1:50,000; Thessaloniki Sheets; IGME: Athens, Greece, 1978. 29. Andronopoulos, V.; Rozos, D.; Hadzinakos, J. Geotechnical Study of Ground Settlement in the Kalochori Area, Thessaloniki District; Unpublished Report; IGME: Athens, Greece, 1990; p. 45. 30. Rozos, D.; Hatzinakos, I. Geological conditions and geomechanical behaviour of the neogene sediments in the area west of Thessaloniki (Greece). In Geotechnical Engineering of Hard Soils-Soft Rocks; A.A. Balkema: Rotterdam, The Netherlands, 1993; pp. 269–274. 31. Stiros, S. Subsidence of the Thessaloniki (northern Greece) coastal plain, 1960–1999. Eng. Geol. 2001, 61, 243–256. [CrossRef] 32. Loupasakis, C.; Rozos, D. Land subsidence induced by water pumping in Kalochori village (North Greece) —Simulation of the phenomenon by means of the ﬁnite element method. Q. J. Eng. Geol. Hydrogeol. 2009, 42, 369–382. [CrossRef] Remote Sens. 2020, 12, 2396 25 of 26

33. Ganas, A.; Chousianitis, K.; Argyrakis, P.; Tsimi, C.; Papanikolaou, M.; Papathanassiou, G.; Exarchos, K. Monitoring of Surface Displacements in the Kalochori Area (Thessaloniki, Greece) Using a Local Gnss Network. Bull. Geol. Soc. Greece 2017, 50, 1553–1562. [CrossRef] 34. Zervopoulou, A. Neotectonic Faults in the Broader Area of Thessaloniki, in Relation to the Foundation Soils. Ph.D. Thesis, Aristotle University of Thessaloniki, Thessaloniki, Greece, 2010. 35. Papazachos, C.; Soupios, P.; Savvaidis, A.; Roumelioti, Z. Identiﬁcation of small-scale active faults near metropolitan areas: An example from the Asvestochori fault near Thessaloniki. In Proceedings of the XXXII ESC General Assembly 2000, Lisbon, Portugal, 10–15 September 2000; pp. 221–225. 36. Paradisopoulou, P.M.; Karakostas, V.G.; Papadimitriou, E.E.; Tranos, M.D.; Papazachos, C.B.; Karakaisis, G.F. Microearthquake study of the broader Thessaloniki area (Northern Greece). Ann. Geophys. 2006, 49, 1081–1093. 37. Garlaouni, C.; Papadimitriou, E.; Karakostas, V.; Kilias, A.; Lasocki, S. Fault population recognition through microsismicity in Migdonia region (northern Greece). Boll. Geoﬁs. Teor. Appl. 2015, 56, 367–382. 38. European Center on Prevention and Forecasting of Earthquakes (ECPFE). Earthquake Planning and Protection Organization (EPPO) & Tectonic Committee of the Geological Society of Greece 1996 Neotectonic Map of Greece, Thessaloniki Sheet, Scale: 1:100000; EPPO & ECPFE: Athens, Greece, 1996. 39. Ganas, A.; Oikonomou, I.A.; Tsimi, C. NOAfaults: A digital database for active faults in Greece. Bull. Geol. Soc. Greece 2017, 47, 518. [CrossRef] 40. Fikos, I.; Ziankas, G.; Rizopoulou, A.; Famellos, S. Water balance estimation in Anthemountas river basin and correlation with underground water level. Glob. NEST J. 2005, 7, 354–359. 41. ZEPBOΠOΥΛOΥ, A.; ΠAΥΛI∆HΣ, Σ. Morphotectonic Study of the Broader Area of Thessaloniki for the Cartography of Neotectonic Faults. Bull. Geol. Soc. Greece 2005, 38, 30–41. [CrossRef] 42. Rozos, D.; Hatzinakos, I.; Apostolidis, E. Engineering Geological Map of Thessaloniki Wider Area; Scale 1 25,000; Institute of Geological and Mineralogical Exploration I.G.M.E.: Athens, Greece, 1998. 43. Thanasoulas, K. Geophysical Study of Anthemountas Area; Technical Report Institute of Geological and Mineralogical Exploration IGME: Athens, Greece, 1983. (In Greek) 44. Anastasiadis, A.; Raptakis, D.; Pitilakis, K. Thessaloniki’s Detailed Microzoning: Subsurface Structure as Basis for Site Response Analysis. Pure Appl. Geophys. PAGEOPH 2001, 158, 2597–2633. [CrossRef] 45. Nagoulis, A.; Loupasakis, C. Hydrogeological conditions of the plain area of the Anthemounta basin (Macedonia, Greece). Bull. Geol. Soc. Greece 2001, 34, 1859–1868. 46. IGME Institute of Geological and Mineralogical Exploration. Geological Map Sheets, Scale 1:50,000; Epanomi (1969), Thessaloniki (1978), Thermi (1978) and Vasilika (1978); IGME: Athens, Greece, 1978. 47. Syridis, G. Lithostromatographical, Biostromatographical and Paleostromatographical Study of Neogene-Quaternary Formation of Chalkidiki Peninsula. Ph.D. Thesis, Aristotle University of Thessaloniki, Thessaloniki, Greece, 1990. 48. IGME Institute of Geological and Mineralogical Exploration. Geological Map of Greece, Scale 1:50,000; Poliyiros and Vasilika Sheet; IGME: Athens, Greece, 1978. 49. Latinopoulos, P.; Theodossiou, N.; Xefteris, A.; Mallios, Z.; Papageoriou, A.; Fotopoulou, E. Developing Water Resources Management Plan for Water Supply-Irrigation; Final Research Project; Aristotle University of Thessaloniki: Thessaloniki, Greece, 2003. 50. Veranis, N.; Xatzikirkou, A. Hydrogeological Report Central Macedonia. Aquifers of Epanomis-Moudaniwn, Kassandras, Ormilias and Sithonias; Unpublished technical report; IGME: Thessaloniki, Spain, 2010; p. 394. 51. Siarkos, I. Developing a Methodological Framework Using Mathematical Simulation Models in Order to Investigate the Operation of Coastal Aquifer Systems: Application in the Aquifer of N. Moudania. Ph.D. Thesis, Aristotle University of Thessaloniki, Thessaloniki, Greece, 2015. 52. Xefteris, A. Investigating the Deterioration of Groundwater in “Kalamaria Plain” Emphasizing on Nitrates. Ph.D. Thesis, Civil Engineering Department, Aristotle University of Thessaloniki, Thessaloniki, Greece, 2000. 53. Gavrihlidou, E. The Long Term Study of the Variability of Hydrological Parameters and their Impacts on Groundwater Reserves at Coastal Aquifers. Master’s Thesis, Geological Department, Aristotle University of Thessaloniki, Thessaloniki, Greece, 2017. 54. 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. 2020, 12, 2396 26 of 26

55. Galloway, D.L.; Burbey, T.J. Review: Regional land subsidence accompanying groundwater extraction. Hydrogeol. J. 2011, 19, 1459–1486. [CrossRef] 56. Hoﬀmann, J.; Zebker, H.A.; Galloway, D.L.; Amelung, F. Seasonal subsidence and rebound in Las Vegas Valley, Nevada, observed by Synthetic Aperture Radar Interferometry. Water Resour. Res. 2001, 37, 1551–1566. [CrossRef] 57. Helm, D.C. One-dimensional simulation of aquifer system compaction near Pixley, California: 1. Constant parameters. Water Resour. Res. 1975, 11, 465–478. [CrossRef] 58. Helm, D.C. One-dimensional simulation of aquifer system compaction near Pixley, California: 2. Stress-Dependent Parameters. Water Resour. Res. 1976, 12, 375–391. [CrossRef] 59. Schmidt, D.A.; Bürgmann, R. Time-dependent land uplift and subsidence in the Santa Clara valley, California, from a large interferometric synthetic aperture radar data set. J. Geophys. Res. Space Phys. 2003, 108, 2416. [CrossRef]

© 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/).