Journal of Coastal Research 00 0 000–000 West Palm Beach, Florida Month 0000

Time Series Synthetic Aperture Radar Interferometry for Ground Deformation Monitoring over a Small Scale Tectonically Active Deltaic Environment (Mornos,

Central ) www.cerf-jcr.org Issaak Parcharidis{, Penelope Kourkouli{, Efthimios Karymbalis{, Michael Foumelis{, and Vassilia Karathanassi{

{Harokopio University of {National Technical University of Athens Department of Geography Laboratory of Remote Sensing El. Venizelou 70, 17671 Heroon Polytechniou 9, 15780 Athens, Greece Athens, Greece [email protected] [email protected]

ABSTRACT

PARCHARIDIS, I.; KOURKOULI, P.; KARYMBALIS, E.; FOUMELIS, M., AND KARATHANASSI, V., 0000, Time series synthetic aperture radar interferometry for ground deformation monitoring over a small scale tectonically active deltaic environment (Mornos, ). Journal of Coastal Research, 00(0), 000–000. West Palm Beach (Florida), ISSN 0749-0208.

This study deals with the estimation of subtle ground deformation at millimetric accuracy over the broader area of the Mornos River delta in Central Greece and its spatio-temporal distribution for the period between 1992 and 2009 through Persistent Scatterers Interferometry (PSI). The results showed that the majority of the scatterers, which show subsidence, are located within the delta plain with mean subsidence rates throughout the delta varying between 27.2 and +0.7 mm/y. An attempt is made to highlight the geographic distribution, the amplitude, and the causes of the observed delta plain subsidence. The positive correlation between the thickness of the fine-grained Holocene deltaic deposits and the subsidence rates reveals that the main cause is the natural compaction of sediments. The highest subsidence is observed at Bouka Karahassani village, which corresponds to the area of the most recently abandoned river mouth, which is intensely eroded by marine processes. Apart from the dominance of fine sediments in the study area, subsidence may also be attributed to submarine gravitational mass movements along the steep slopes of the prodelta as well as to the reduction of sediment load after the dam construction in 1979. The NW part of the delta seems to have been affected by aseismic slip along a NE-SW trending normal fault buried beneath the alluvial deposits of the Skala torrent fan.

ADDITIONAL INDEX WORDS: River Deltas, subsidence, Persistent Scatterers Interferometry, Mornos River Delta.

INTRODUCTION The Permanent or Persistent Scatterers Interferometry (PSI) (Ferretti, Prati, and Rocca, 2001; Hanssen, 2005; Werner Repeat-pass interferometry is a unique tool for large-scale et al., 2003), a relatively recent development used to overcome monitoring (spatially continuous) of surface deformation at the previous limitations, is a technique implemented to a low cost (Massonnet et al., 1993; Zebker et al., 1994). calculate fine motions of individual points over wide areas Regardless of the applied differential interferometric method, covering mainly urban and semiurban environments (Burg- repeat-pass interferometry is limited by a number of param- mann et al., 2006; Colesanti et al., 2003; Parcharidis et al., eters, such as 2006; Raucoules et al., 2008; Salvi et al., 2004; Teatini et al., N large baselines, which lead to low correlation attributable 2005). The technique uses a series of coregistered Synthetic to spectral shift for distributed targets; Aperture Radar (SAR) scenes to identify time persistent N loss of coherence attributable to long temporal separation scatterer (PS) points. The processing is carried out on these between acquisitions; persistent points of the SAR images having stable radiometric N difficulties to unwrap interferograms with large baseline characteristics. By examining their interferometric phase it is attributable to low correlation and high fringe rates; and possible to monitor ground stability for an area that is normally N atmospheric artifacts attributable to tropospheric water characterized by its low coherence. Additionally millimetric vapor and ionospheric electron density. target displacement along the line of sight directions can be detected, allowing the measurement of slow terrain motion (Ferretti, Prati, and Rocca, 2001; Hanssen, 2005). DOI: 10.2112/JCOASTRES-D-11-00106.1 received 14 June 2011; The unique benefit of PS interferometry is its ability to accepted in revision 19 August 2011. Published Pre-print online XX Month XXXX. provide both annual average motion rates and deformation ’ Coastal Education & Research Foundation 2012 histories for the individual targets (Parcharidis et al., 2009).

Journal of Coastal Research coas-28-05-09.3d 4/10/11 15:48:02 1 Cust # JCOASTRES-D-11-00106 0 Parcharidis et al.

River deltas are coastal features developed from the accumu- States to analyze Late Holocene deposits and assess compac- lation of sediment near the mouth of rivers. They have been and tion rates. continue to be the heart of economic developments. They In Greece, Stiros (2001) studied the subsidence of the confront the coastal regions with dynamically changing Thessaloniki coastal plain, which composes part of the delta qualitative and quantitative characteristics of the river that complex of Axios, Aliakmon, Loudias, and Gallikos Rivers in feeds them. Pollution, canalization, urbanization, and defores- northern Greece for the time period between 1960 and 1999. He tation within the catchment areas affect the delta environ- proposed that the observed subsidence should be regarded as ment. The deltas themselves are protected areas and/or are the cumulative effect of several factors, including consolidation characterized by agricultural activity and in many cases by the of near-surface sediments attributable to the decline of the existence of urban centers. On tectonically active areas, delta piezometric level and the partial abandonment of the delta, evolution may be influenced by abrupt relative sea level oxidation of peat soils in the vadose zone, synsedimentary changes attributable mainly to coseismic uplift or subsidence deformation (faulting and flow), and loading-induced consoli- (Barnhardt and Sherrod, 2006). dation of deeper sediments. Land subsidence is a common phenomenon in modern delta Deltas with significant subsidence attributable to anthropo- plains because of compaction of sediments by consolidation of genic activities such as oil and gas extraction include the Niger, the dewatered material, compression of the load of the Magdalena, Makaham, MacKenzie, and Mississippi deltas subsequent overlain deposits, and groundwater or oil pumping (Ericson et al., 2006) and the Yellow River (Huanghe). Relative (Day et al., 1995; Dumont and El-Shabrawy, 2007; Meckel, Ten contributions of geologic and anthropogenic processes to Brink, and Williams, 2007; Nageswara Rao et al., 2010; Shi subsidence of river deltas or coastal plains are vigorously et al., 2007; Stanley, 2005). Previous studies on delta rivers debated. Shallow sediment compaction is often considered to be based on field data show that deformation features of the the dominant process (Meckel, Ten Brink, and Williams, 2007). hydrographic units are greatly related to hydrogeological Understanding the rate and the nature of subsidence of properties and ground–water–level variations. An identical sediment sequences in a delta is necessary for proper analysis hydrographic unit may exhibit different deformation features of the delta-building process through time. This is especially in different locations, and additionally the hydrogeology in the true in the present context of sea-level rise and coastal erosion. same location could exhibit different deformation in different The relative sea-level rise matters most in determining the periods or seasons (Shi et al., 2008; Wu et al., 2008). Except response of a deltaic coastal zone with subsiding areas being near coastal areas, land subsidence because of ground-water more susceptible to sea invasion (Mazzotti et al., 2009). Coastal pumping generally is not a serious hazard if the whole area subsidence of low-lying delta plains results in sea-level rise, subsides. The major problems related to subsidence areas are shoreline erosion, and wetland loss, which causes a threat to the potential destructive results caused by differential subsi- coastal populations (Jelgersma, 1996). Thus quantitative dence (Holzer and Pampeyan, 1981). understanding of subsidence is important for predicting Several studies were made on land subsidence from various relative sea-level rise, storm-surge flooding, and successful river deltas around the world. Cencini (1998) investigated the wetland restoration. subsiding Po delta in Italy, the greater part of which lies below The present paper concerns the application of PSI to detect sea level today. Stanley and Warne (1998) described and ground deformation in the Mornos delta for the period 1992– discussed symptoms of a destruction phase of the Nile delta in 2009. The objective of the study is three-fold: (i) detect spatial- Egypt during the past 150 years, triggered by water regulation temporal deformation of the delta, (ii) distinguish between that has disrupted the balance among sediment influx, erosive deformation attributable to natural delta sediment compaction and deformation induced by human activities, and (iii) effects of coastal processes, and subsidence. Becker and Sultan investigate the existence of differential vertical subsidence in (2009) evaluated the modern rates of subsidence of the NE Nile the delta area caused by buried fault zones and submarine delta using persistent scatter radar interferometry techniques. mass movements. Chen and Stanley (1993) and Stanley and Chen (1993) studied The Mornos River, about 200 km west of Athens, drains the the Late Quaternary subsidence of the outer Yangtze delta in SE Pindos Mountain and is one of the two existing water China as well as the deformation of Holocene sedimentary resources that supply the greater Athens area with potable sequences by using high resolution seismic profiles and water. A dam completed in 1979 regulates its flow. The Mornos stratigraphic analysis of borings. Cohen, Gouw, and Holten delta, a subaerial delta with fan-shaped body (Piper et al., (2003) examined the sedimentary response upstream and 1990), is located in the NW (Figure 1). The Gulf downstream of a fault zone located in the central part of the is an asymmetric rift and one of the most active continental Rhine-Meuse delta in the Netherlands and used that to tectonic structures in the Mediterranean area (Armijo et al., quantify differential subsidence rates. Kuehl et al. (2005) 1996; Avallone et al., 2004). summarized the current knowledge of the Late Glacial to Holo- cene sedimentation from the upper Ganges-Brachmaputra DATA AND METHODOLOGY delta plain in India-Bangladesh to the continental shelf break indicating that the Holocene subsidence of the delta is the In this study SAR and Advanced SAR (ASAR) scenes from result of changing river discharge, basin filling, and delta-lobe ERS-1,2 and ENVISAT satellites, respectively, were used. In migration. Tornqvist et al. (2008) used a series of radiocarbon- total 71 Single Look Complex (SLC) scenes in descending dated sediment cores from the Mississippi delta in the United orbits, VV polarization, and operating in C-band (track:279,

Journal of Coastal Research, Vol. 00, No. 0, 0000

Journal of Coastal Research coas-28-05-09.3d 4/10/11 15:48:04 2 Cust # JCOASTRES-D-11-00106 SAR Interferometry over Mornos Delta 0

Figure 1. Location map of the Mornos River delta in Central Greece.

frame:2835) were used. From them 42 were acquired from were ingested in processing to enhance the accuracy of the ERS covering the time period 1992–2001, and 29 were from satellite’s orbit and estimate the initial interferometric Envisat (ERS-like mode) covering the time period 2002–09. baselines. Specifically, for ERS scenes, orbit data from Moreover, a Digital Elevation Model (DEM) produced by the DELFT Institute (NL) for Earth-Oriented Space Research Shuttle Radar Topography Mission was used with approxi- (DEOS) (Scharoo and Visser, 1998) were obtained, and for mate spatial resolution of 90 m. Finally, precise orbit vectors ENVISAT scenes orbit data from Doppler Orbitography and

Journal of Coastal Research, Vol. 00, No. 0, 0000

Journal of Coastal Research coas-28-05-09.3d 4/10/11 15:48:04 3 Cust # JCOASTRES-D-11-00106 0 Parcharidis et al.

Radio-positioning Integrated by Satellite (DORIS) instru- winter months. The flow of the river has been regulated since ment were obtained. 1980 by a dam constructed in the upper reaches of the basin in The differential interferometric methodology that will be order to supply Athens with potable water. The suspended used in this study was basically the PSI based on the sediment load of the river prior to the dam construction was Interferometric Point Target Analysis (IPTA) algorithm in- estimated between 0.5–0.8 tons/y/km2 (Piper et al., 1990). cluded in GAMMA software. The IPTA is a toolbox that can support many different methodologies including different Subaerial Morphology methodologies for candidate scatterers selection, spatial and temporal phase unwrapping, etc. Subsidence within the delta The Mornos River has a braided gravel channel aligned and plain was estimated through the IPTA. enclosed by artificial levees along its entire delta length to its In order to investigate the influence of sediment compaction present mouth in the western part of the delta. The 1945 aerial on the differential subsidence within the delta plain, boring log photographs show that there were two active distributaries sediment descriptions of 17 boreholes were considered. Bore- with the main channel located in the central SE part of the hole data were obtained from a geophysical study conducted by delta plain while the present river channel was a lesser the Ministry of Agriculture during the late 1970s at the deltaic distributary. The eastern course had a more meandering plain. These boreholes were drilled as part of a proper water pattern and was flowing through the center of the deltaic plain resources exploration program in the deltaic plain, initiated discharging 3.7 km east of the present active river mouth. This because of the disturbance of the hydgogeological conditions of channel was partially abandoned after the artificial alignment the broader delta area caused by the construction of a dam of the now active channel that took place in 1961. Following at the upper reaches of the river. For each borehole the construction of the dam in the upper reaches of the drainage percentages of coarse- and fine-grained sediments in the basin in 1980, the sediment supply decreased significantly drilling log was estimated. A buffer zone of 200 m around the downstream, and this channel has been totally abandoned. location of each borehole was created, and mean annual Karymbalis, Gaki-Papanastassiou, and Maroukian (2007) subsidence rates were calculated for the time period under mapped older abandoned channels and grouped them in four study. Then a plot of the percentage of fine sediments in the systems that begin from the apex of the delta leading to the boring logs vs. mean annual subsidence rate was created. present shoreline. Old reliable maps of 1852 and 1885 depict In an attempt to determine the main causes of variability of that there was one active distributary channel located in deformation rates in different parts of the delta plain, data and the western part of the delta during that period. This now observations regarding the geomorphology of the subaerial abandoned river path lies about 0.6 km west of the present delta plain, the subaqueous prodelta, as well as the delta channel. In the 1885 map, flooded abandoned channels are also shoreline retreat, were also collected. visible in the central part of the deltaic plain. Another abandoned channel system begins from the apex of the fan GEOLOGY AND GEOMORPHOLOGY OF THE delta ending SE of Managouli village about 1.9 km NE of the STUDY AREA Bouka Karahassani mouth and seems to be older than that because it has undergone intense retreat that is attributable to The Mornos River delta is a Gilbert-type fan-shaped alluvial marine erosion. The easternmost channel system has a W-NW– body located on the northern side of the western Gulf of E-SE direction north of Logos village leading to 0.5 km north of Corinth. It occupies an area of 28 km2 and has a mean delta Chiliadou at the eastern end of the deltaic plain (Figure 2). plain gradient of 0.004% (0.2u). This channel maintains a meandering pattern for the last According to the delta classification proposed by Galloway 1.4 km before reaching the shoreline. These traces seem to be (1975) it should be classified among those affected mainly by the oldest visible distributary channels of the Mornos River fluvial sediment supply and wave activity, while its develop- delta. The arcuate shape of the fan delta, the abandoned ment began during the Late Holocene with the relative channels, and the dominance of gravel suggests that delta stabilization of the sea-level rise. The configuration of the growth took place by gradual shifting of braided distributaries delta is the result of suitable conditions for delta formation westwards. during the Late Holocene. Intensive weathering and erosion over the catchment area of the Mornos River has resulted in the Shoreline Shift production of large quantities of sediments readily available for transportation. This process was enhanced by the lithology and Today coastal erosion is the dominant geomorphic process climate conditions of the drainage basin as well as by the along the delta coastline because of construction of the dam characteristics of the receiving basin (Gulf of Corinth). A large that has drastically reduced the sediment flux leading to part of the drainage basin (54.4% of the total area) consists of subsidence of the deltaic environment and enhancing the wave clastic sedimentary rocks, which are mainly flysch formations activity effectiveness. The eastern coast, NE of the most of great erodibility. Additionally mean annual precipitation is recently abandoned river mouth, has retreated approximately high enough, fluctuating between 600 mm near the delta to 86 m between 1945 and 1986, corresponding to a rate of 2.14 m/ more than 1200 mm in the N and NE highlands with most of y (Karymbalis, Gaki-Papanastassiou, and Maroukian, 2007). the rain falling between November and February. Mean Retreat rate accelerated to 3.25 m/y for the period between annual discharge is approximately 40 m3/s and follows the 1986 and 1998. Today, the only region where the fan delta precipitation trend with the greatest flow recorded during the progrades is the area between the present mouth and Bouka

Journal of Coastal Research, Vol. 00, No. 0, 0000

Journal of Coastal Research coas-28-05-09.3d 4/10/11 15:48:08 4 Cust # JCOASTRES-D-11-00106 SAR Interferometry over Mornos Delta 0

Figure 2. Geomorphological map of the study area also including the seismic events (geomorphic features resembling Karymbalis, Gaki-Papanastassiou, and Maroukian [2007]; seismic events after NKUA-Laboratory of Seismology and NOA-Institute of Geodynamics).

Karahassani. A progradation rate of 3.4 m/y is estimated Mornos River consists of a series of fluvial deltaic sediments between 1945 and 1998 following the redistribution of the with a thickness ranging from 60 to 150 m. sediment derived from the erosion of the abandoned mouth and The stratigraphy of the recent deltaic sediments is not the eastern shores by longshore drift from the east. uniform and is laterally variable. The subsurface sedimentary Although the main distributary prograded at a rate of sequences are characterized by alternations of coarse sediments approximately 4 m/y between 1945 and 1986, a retreat of (gravels and pebbles), fine-grained layers (silt and sand), and 11.4 m/y is observed between 1986 and 1998 because of the mixed deposits (gravels, pebbles, and silt). Detailed stratigraph- construction of the dam. ic information is unavailable for much of the delta plain, and stratigraphic heterogeneity leads to uncertainties where bor- Subsurface Stratigraphy ings are unavailable. Although the stratigraphy of this recent complex delta is obviously not uniform but laterally variable The subsurface stratigraphy of the deltaic sediments is based and deviating from an idealized scheme, the sedimentation on the description of 17 borehole logs distributed over the delta pattern seems to be typical of a fan delta. The logs of the plain (Figures 3 and 4). The basement of the Mornos River boreholes that are located near the apex and the margins of the delta is composed of a compaction-free Mesozoic substratum delta are represented by the abundance of coarse material. On overlain by a Late Holocene sediment sequence. This basement the other hand the southernmost borehole logs consist of fine- is constructed by the geological formations of flysch and grained layers interrupted by coarse materials. Fine sediments, limestone that belong to the geotectonic zones of Pindos. Test which probably represent overbank sediments, deposit during drilling and geophysical investigation detected the existence of flood events in the evolutionary phases of the delta. Pebbles and three main foldings of the rocks that form the deltaic sediments gravels correspond to channel sediments deposited by older, basement. The axes of these forms have an almost N-S trend now buried distributary braided channels. Under the delta and divide the subsurface waters in three different parts. The three distinct aquifers can be distinguished: the quaternary, borehole logs indicate that the deltaic alluvial body of the the flysch, and the karstic limestone aquifers (Louis et al., 2004).

Journal of Coastal Research, Vol. 00, No. 0, 0000

Journal of Coastal Research coas-28-05-09.3d 4/10/11 15:48:08 5 Cust # JCOASTRES-D-11-00106 0 Parcharidis et al.

Figure 3. Synthetic plot of the logs of the boreholes drilled in the Mornos delta plain by the Ministry of Agriculture.

Subaqueous Morphology distributary mouth of the Mornos, the sea floor on the prodelta slope is irregular. Seaward of the abandoned SE distributary The western Gulf of Corinth, where the study area is located, mouth there is a deeply incised valley. The prodelta slope to the is a complex asymmetrical graben and is presently the most SW of this valley is locally modified because of slumping. To the tectonically active part with geodetic extension rates reaching NE of the valley the muddy prodelta is cut by a series of gullies, up to 14–16 mm/y (Bernard et al., 1997) and intense seismicity while on the lower prodelta slope where there is a broad (Lyon-Caen et al., 2004). The morphology of the broader area of elongated depression, it is interpreted as a slump scar. Slumps the subaqueous part of the Mornos delta is shown in Figure 5. have been mapped off the Mornos delta, and slumped material Seismic profiling surveys and recent swath bathymetry provide or debris flows seem to be important locally. accuracy on the submarine topography (Piper et al., 1990; Tinti Recent swath bathymetry performed by the Hellenic Centre et al., 2007). The northern edge of the central graben is marked for Marine Research revealed that the most prominent by a sea-floor fault scarp running down the center of the Gulf. potentially unstable bodies of the western Gulf of Corinth are This fault scarp is obscured both by progradation of the Mornos found along the east-facing steep prodelta slopes off the delta and by more distal mud deposits. abandoned distributary channel of Bouka Karahassani (Tinti The southern prodelta slope of the Mornos has flat reflective et al., 2007). Much of the suspended load of the river is probably bottoms (sandy channels, lobes, and slumps) on 3.5-kHz deposited on the prodelta slope. This suspended sediment profiles and is floored by silty sand. Seaward of the SW active accumulation, which is estimated as 0.5–0.8 tons per year,

Journal of Coastal Research, Vol. 00, No. 0, 0000

Journal of Coastal Research coas-28-05-09.3d 4/10/11 15:48:11 6 Cust # JCOASTRES-D-11-00106 SAR Interferometry over Mornos Delta 0

Figure 4. The locations of the boreholes are shown in the map.

corresponds to a 2 cm/y sedimentation rate if all of this Moreover, the geocoding equations were applied on the DEM sediment is distributed uniformly over the entire prodelta in order to convert its geometry into the SLC scene geometry. slope. The morphology of the flank, which is dominated by steep slopes (gradients immediately off the river mouth are 18u) and the presence of large amounts of sediments discharged by the Main Processing Mornos River, create favorable conditions for sediment failure Based on the coregistered SLC scenes, two candidate lists of under seismic or gravitational load (Lykousis, Sakellariou, and PSs were initially estimated using two different selection Roussakis, 2003). For instance between 1889 and 1939, the approaches. The first list was generated based on phase Patras–Corinth no. 2 telegraph cable broke several times off properties. More specifically this approach identifies point the Mornos delta (Heezen, Ewing, and Johnson, 1966). Two targets with low spectral phase diversity. The spectral phase breaks were associated with seismic activity; the remaining characteristics are averaged over the stack of SLC scenes, and occurred between the months of October and early May and afterwards the average spectral behavior is used to designate was probably related to sediment slumping or turbidity the candidate PS. The second approach is based on low currents following heavy river discharge (Piper et al., 1990). intensity variability as the intensity and phase depend on the These gravitational mass movements seem to affect the point-target cross section and position. Then, the two candidate subsidence along the deltaic coastal zone and especially the point lists were combined into a single one, which finally was east-facing coastline. used for the analysis. A total number of approximately 18,000 and 20,000 candidate points for ERS-1, 2 and ENVISAT INTERFEROMETRIC PROCESSING dataset, respectively, were determined. The spatial distribu- Preprocessing tion of the candidate points is not homogeneous with high densities of PS located in built-up areas and low densities over Two separate processing procedures were carried out for fields. ERS and ENVISAT datasets, respectively. The first common The IPTA methodology requires one single scene as reference step was the coregistration of the SLC scenes so that each scene in order to form multiple pairs and produce interferograms. obtains the same geometry. For each dataset the geometry of The criteria by which the reference scene was selected were the the first scene (November 12, 1992, for ERS and October 20, following: (i) forming interferometric pairs with the minimum 2002, for ENVISAT) was used as reference geometry. The baseline (Bp), (ii) acquisition date near the central date of the achieved coregistration accuracy was satisfactory with stan- time period for which there are available SAR acquisitions, and dard deviations of individual range and azimuth offsets from (iii) reference scene to present low atmospheric distortions. The the offset regression fit, less than 0.3 pixels. scenes that fulfilled the previous criteria were June 3, 1995, for

Journal of Coastal Research, Vol. 00, No. 0, 0000

Journal of Coastal Research coas-28-05-09.3d 4/10/11 15:48:15 7 Cust # JCOASTRES-D-11-00106 0 Parcharidis et al.

Figure 5. Map of the subaqueous area of the Mornos delta showing the distribution of submarine sandy prodelta platforms, sandy channels, lobes, slumps, mud, and eroded by tidal currents sea bed. Seabed features such as gullies, slump scars, and fault scarps are also depicted (modified from Piper et al., 1990).

the ERS dataset with an average Bp of 379 m, and May 22, Because the aforementioned terms depend on the distance 2005, for the ENVISAT with an average Bp of 78 m. between points in the image, differential interferograms were Afterwards the initial differential interferograms were thus divided in patches, and for each patch the PS with the produced using the coregistered SLCs, the DEM heights, and highest quality was selected as a reference PS point. Then for the preceding point lists. This is performed by simulation of the each PS, phase differences between the reference PS point and unwrapped interferometric phase based on the initial baselines its phase values were introduced in the regression analysis in and the DEM. Then, the differential interferograms were order to estimate (i) the baseline error and correct the PS- analyzed in the temporal and spatial domain in order to obtain height value, and (ii) the deformation and atmospheric rate information on the atmospheric phase term, deformation phase relative to the reference point. term, and baseline errors. Thus, processing proceeds by Based on the regression analysis, the quality of the PS applying a least-squares regression on the differential phases candidates was further evaluated through the estimated in order to estimate terrain height and deformation rate standard deviation of the phase difference. PSs with a phase relative to a reference point target. standard deviation larger than the indicated threshold (in this

Journal of Coastal Research, Vol. 00, No. 0, 0000

Journal of Coastal Research coas-28-05-09.3d 4/10/11 15:48:17 8 Cust # JCOASTRES-D-11-00106 SAR Interferometry over Mornos Delta 0

Figure 6. Linear component of ground deformation over Mornos delta broader area for the period 1992–2000. Point targets are plotted on a high resolution satellite image. Star symbol represents the reference point.

case 1.0 radian in both datasets) was rejected, significantly compaction attributable to subjacent grain-size variability is reducing the number of points. The majority of the rejected a significant factor (Figure 8). For instance studies in the points were located over mountainous areas. A total number of Rhine-Meuse delta in The Netherlands provide an example 3353 and 3857 scatterers for the time period 1992–2000 and of .6 m of differential compaction that could be related to a 2002–09, respectively, were detected. subsurface sand body. The plot shows a positive correlation Moreover, because residual phases contain the atmospheric between subsidence rate and the thickness of the fine-grained term, nonlinear deformation, and error terms, they were sediment participation in the subsurface sedimentary se- further processed in order to compensate atmospheric and quence. Thus natural compaction of delta sediments seems to noise effects. Thus, residual phases were spatially filtered by be the main reason for the observed ground subsidence of applying a low-pass spatial filter. Atmospheric and error terms the delta plain surface. Natural compaction happens as new were reduced by the spatial filtering around the reference sediment (weight) is added to the deltaic sedimentary points, assuming that stability occurred in the region of the sequence, and the underlying sediment is reduced of its water area. Further iteration was done in order to apply additional content because of reduction of the void spaces between grains corrections in the final regression model. (Meckel, Ten Brink, and Williams, 2007). According to Meckel, The generated results consist of height corrections, linear Ten Brink, and Williams (2007), compaction rates vary over deformation rates, atmospheric phase, refined baselines, three orders of magnitude, but 80% of the rates are between 0.7 temporal coherence, and nonlinear deformation histories for and 2.2 mm/y with heterogeneity in sediment layers increasing each scatterer. Finally the deformation phases were trans- the rate of compaction. As shown in Figures 6 and 7, relatively formed into displacements and geocoded to the selected lower subsidence rates within the delta plain are observed in its cartographic reference system (UTM, wgs’84). central portion between the villages of Ag. Polykarpos and Malamata along the abandoned distributary channel. Both the RESULTS AND DISCUSSION surface and the underlain deposits of this area are dominated by coarse-grained channel sediments characterized by signif- Following the transformation of the interferometric phases icantly lower rates of natural compaction. from range–Doppler coordinates into map geometry (geographic Scatterers located over mountainous areas and also over the coordinates), PSs were imported into a GIS environment and major part of show stability or slight uplift. It is plotted on a high-resolution Quickbird image (Figures 6 and 7). observed that for both periods, point targets show the same In both time periods it is obvious that the majority of the deformation pattern. Specifically, moving from the city of scatterers, which show subsidence, are located within the delta. Nafpaktos toward the delta, the linear deformation rates of The diagram of percentage of fine sediments of the boring point targets from positive values become negative. This log vs. subsidence rate indicates that differential natural change occurs along a line separating the alluvial fan of Skala

Journal of Coastal Research, Vol. 00, No. 0, 0000

Journal of Coastal Research coas-28-05-09.3d 4/10/11 15:48:18 9 Cust # JCOASTRES-D-11-00106 0 Parcharidis et al.

Figure 7. Linear component of ground deformation over Mornos delta broader area for the period 2002–09. Point targets are plotted on a high resolution satellite image. Star symbol represents the reference point.

Figure 8. Diagram of percentage of fine sediments of the boring log vs. subsidence rates.

Journal of Coastal Research, Vol. 00, No. 0, 0000

Journal of Coastal Research coas-28-05-09.3d 4/10/11 15:48:21 10 Cust # JCOASTRES-D-11-00106 SAR Interferometry over Mornos Delta 0

Figure 9. Frequency histogram of deformation rates for the period 1992–2000.

torrent from the western Mornos delta deposits. The observed settlements show almost the same deformation pattern in both vertical motions on the two sides of this linear path, as well as periods. Most of the villages show higher subsidence during the the existence of an offshore fault scarp at its SW continuation, 2002–09 period than the 1992–2000 period. presented on maps produced by a seismic profile survey The villages Kastraki, Xiropigado, and are exceptions (Papanikolaou et al., 1987) indicate the presence of a NE-SW as they show decreased subsidence for the period 2002–09 trending normal fault buried under the alluvial deposits. compared to that of 1992–2000. Furthermore the maximum Differential vertical movements are the result of aseismic slip deformation rate of subsidence is observed at Managouli with along the fault during the observation period. rates between 22.9 mm/y and 23.5 mm/y, at Chiliadou with Concerning the period 1992–2000, the maximum observed rates between 22.6 mm/y and 23.4 mm/y, and at Malamata subsidence is approximately 27.2 mm/y, while the maximum with rates between 22.6 mm/y and 23.2 mm/y, for 1992–2000 uplift is about +2.2 mm/y. For the period 2002–09, the linear and 2002–09, respectively. Settlements located at the edge of deformation rate varies between 26.8 mm/y and +4.1 mm/y. the delta show much lower subsidence rates than villages The average estimated rate uncertainty is approximately located within the delta plain. This has to do with the sediment 60.2 mm/y. distribution and the morphology of the valley basement relief. Using statistical analysis tools, frequency histograms of It is evident that in deltaic areas where the recent deposits are deformation rates were generated (Figures 9 and 10). For the underlying by gradually varying size unconsolidated deposits, time span 1992–2000 the highest density of the point targets is the coarse-grained materials are expected to be close to the between 22 mm/y and 20.6 mm/y, while for the 2002–09 period initiation of delta. the highest density occurs between 21.5 mm/y and +0.5 mm/y. Additionally it is interesting to note that for both time Moreover, the amount of points that are characterized by periods the highest subsidence is observed at Bouka Karahas- extreme rates of maximum and minimum is quite small. sani (the old mouth of delta). This area is dominated by An additional attempt of a detailed study was made within permanent marshes and fine sediments attributable to the long the Mornos delta. Specifically, the average deformation rate distance from the apex of the delta, which also explains the was estimated for every rural community located in the delta highest rate of subsidence. Moreover, the instability because of (Managouli, Chiliadou, Logos, Malamata, Ag. Polykarpos) and the gravitational load of the SE prodelta slope unconsolidated towns and villages located at the edge of the delta (Nafpaktos, deposits affects the ground subsidence of the eastern delta Efpalio, Kastraki, Xiropigado). As shown in Figure 11, the front. Complete cessation of sediment load supply after the

Journal of Coastal Research, Vol. 00, No. 0, 0000

Journal of Coastal Research coas-28-05-09.3d 4/10/11 15:48:23 11 Cust # JCOASTRES-D-11-00106 0 Parcharidis et al.

Figure 10. Frequency histogram of deformation rates for the period 2002–09.

Figure 11. Average deformation rate for each village located in the Mornos delta covering the time span between 1992 and 2009.

Journal of Coastal Research, Vol. 00, No. 0, 0000

Journal of Coastal Research coas-28-05-09.3d 4/10/11 15:48:26 12 Cust # JCOASTRES-D-11-00106 SAR Interferometry over Mornos Delta 0

abandonment of this distributary channel attributable to the and the periphery of the delta where the free-compaction rocky modification of the natural flow of the river is also responsible basement exists. for the accelerated subsidence of the old river mouth (Bouka The highest subsidence rate (27.2 mm/y) is observed at Karahassani). River-flow anthropogenic modifications include Bouka Karahassani where the old, most recently abandoned the construction of the upstream dam and the artificial mouth of the river is located. Compaction of fine-grained deltaic confinement of the western distributary channel. Extremely sediments can only account for a part of the observed high retreat rates of the eastern shoreline (2.14 mm/y subsidence in this part of the delta. The premier force seems between 1945 and 1986 and 3.25 mm/y for the period to be submarine slumping and mass movements on the steep between 1986 and 1998) postdate the construction of the dam prodelta of the eastern delta coastline. These gravitational and seem to be the result of intense subsidence (Karymbalis, slides seem to contribute significantly to relative sea-level rise Gaki-Papanastassiou, and Maroukian, 2007). and intense erosion along the eastern delta shoreline. The fact that surface subsidence and the related marine CONCLUSIONS invasion postdate both the construction of the dam at the upper reaches of the basin and the abandonment of the eastern Persistent Scatterers Interferometry (PSI) served as a distributary channel shows that the cessation of sediment supply methodology of ground deformation detection and estimation because of manmade modifications is another factor responsible for the broader area of the Mornos River delta for the time for the delta surface subsidence and shoreline erosion. period between 1992 and 2009. Moreover, the results of the applied methodology assume the In most cases the technique has been applied to monitor existence of a possible NE-SW trending normal fault buried deformation over urban and suburban areas. Natural terrains, under the alluvial deposits of the Skala torrent at the NW flank as in this case the deltaic, attributable to surface coverage with of the delta. Aseismic slip along this fault line has affected the strong presence of vegetation are not considered optimal ground deformation at its vicinity. environments for the application of the technique. Neverthe- Future efforts should focus on the quantification of the contri- less, the technique performance was satisfactory because of the bution of each individual deformation factor through simulation and presence of dispersed settlements in the region of interest. modeling in order to allow response and mitigation actions. Although a relatively limited number of reflectors were identified, sufficient spatial distribution has allowed indirect ACKNOWLEDGMENTS conclusions to be reached regarding the causes of observed ground deformation. The present study is a part of the TERRAFIRMA-X project Subsidence seems to be a dominant process for the Mornos supported by ESA’s GMES Service Element Program, which River deltaic plain, while relative stability or slight uplift was aims to provide a pan-European ground motion hazard observed for the mountainous areas surrounding the alluvial information service. Data have been provided by the deposits. For the time period 1992–2000 the maximum European Space Agency (ESA). We would like to thank subsidence rate was estimated to be 27.2 mm/y, while the rate Professor Hampik Maroukian for his remarks and the two was approximately 26.8 mm/y between the years 2002 and anonymous reviewers for their helpful suggestions, com- 2009, however, the uplift rates range between +2 mm/y and ments, and corrections that significantly improved the paper. +4 mm/y. The uncertainty of the deformation rate estimation is almost 60.2 mm/y. LITERATURE CITED The main causes of land subsidence are natural sediment Armijo, R.; Meyer, B.; King, G.; Rigo, A., and Papathanassiou, D., compaction, aseismic slip along a buried active fault zone, 1996. Quaternary evolution of the Corinth rift and its implications submarine mass movements on the SE prodelta steep slopes, for the late Cenozoic evolution of the Aegean. Geophysical Journal and reduction of sediment supply because of human activities. International, 126, 11–53. Avallone, A.; Briole, P.; Agatsa-Balodimou, A.M.; Billiris, H.; In the study area the main cause of ground subsidence seems Charade, O.; Mitsakaki, C.; Nercessian, A.; Papazissi, K.; Paradisis, to be natural compaction of the loose deltaic sediments. The D., and Veis G., 2004. Analysis of eleven years of deformation relatively high subsidence rates are the result of high measured by GPS in the Corinth Rift Laboratory area. Comptes compaction rates attributable to the heterogeneity in sediment Rendus Geoscience, 336, 301–311. Barnhardt, W. and Sherrod, B., 2006. Evolution of a Holocene delta layers. Difference in mean subsidence depends partially on the driven by episodic sediment delivery and coseismic deformation, architecture of the underlying Holocene strata that dictate Puget Sound, Washington, USA. Sedimentology, 53, 1211–1228. differential natural sediment compaction rates primarily Becker, R.H. and Sultan, M., 2009. Land subsidence in the Nile Delta: controlled by the relative thickness of fine- and coarse-grained inferences from radar interferometry. The Holocene, 19, 949–954. sediments. The settlements of Kastraki, Xiropigado, Nafpak- Bernard, P.; Chouliaras, G.; Tzanis, A.; Briole, P.; Bouin, M.P.; Tellez, J.; Stavrakakis, G., and Makropoulos, K., 1997. Seismic and tos, and Efpalio, which are located at the edge of the delta, show electrical anisotropy in the Mornos delta, Gulf of Corinth, Greece, much lower mean annual subsidence rates (ranging from and its relationship with GPS strain measurements. Geophysical 21.7 mm/y to 20.3 mm/y) than the villages of Managouli, Research Letters, 24(17), 2227–2230. Ag. Polykarpos, Logos, Chiliadou, and Malamata (mean Burgmann, R.; Hilley, G.; Ferretti, A., and Novali, F., 2006. Resolving subsidence rates between 23.5 and 21.0 mm/y), which are vertical tectonics in the San Francisco Bay Area from permanent scatterer InSAR and GPS analysis. Geology, 34(3), 221–224. located within the delta plain. This can be attributed to the Cencini, C., 1998. Physical processes and human activities in the delta deposits’ distribution and the valley basement relief. The evolution of the Po delta, Italy. Journal of Coastal Research, 14(3), recent coarse-grained sediments are located close to the apex 774–793.

Journal of Coastal Research, Vol. 00, No. 0, 0000

Journal of Coastal Research coas-28-05-09.3d 4/10/11 15:48:29 13 Cust # JCOASTRES-D-11-00106 0 Parcharidis et al.

Chen, Z. and Stanley, D.J., 1993. Yangtze delta, eastern China: 2. Nageswara Rao, K.; Sadakata, N.; Shinde, V.; Rajawat, A.S., and Ajai, Late Quaternary subsidence and deformation. Marine Geology, 2010. Subsidence of Holocene sediments in the Godavari delta, 112(1–4), 13–21. India. Frontiers of Earth Science in China, 4(4), 410–416. Cohen, K.M.; Gouw, M.J.P., and Holten, J.P., 2003. Fluvio-deltaic Papanikolaou, D.; Chronis, G.; Lykousis, V., and Pavlakis, P., 1987. floodbasin deposits recording differential subsidence within Active tectonics in the Rion–Antirion Strait, Western Greece. a coastal prism (central Rhine-Meuse delta, the Netherlands). Proceedings of 5th Meeting European Geological Societies, Du- In: Blum, M.D., Marriott, S.B., and Leclair, S.F. (eds.), Fluvial brovnik 1987, pp. 72–73. Sedimentology VII, Special Publication 35, International Associa- Parcharidis, I.; Lagios, E.; Sakkas, V.; Raucoules, D.; Feurer, D.; Le tion of Sedimentologists. Oxford: Blackwell Scientific, pp. 295–320. Mouelic, S.; King, C.; Carnec, C.; Novali, F.; Ferretti, A.; Capes, R., Colesanti, C.; Ferretti, A.; Novalli, F.; Prati, C., and Rocca, F., 2003. and Cooksley, G., 2006. Subsidence monitoring within the Athens SAR monitoring of progressive and seasonal ground deformation basin (Greece) using space radar interferometric techniques. Earth, using permanent scatterers technique. IEEE Transactions on Planets and Space Journal, 58, 505–513. Geosciences and Remote Sensing, 41(7), part I, 1685–1701. Parcharidis, I.; Foumelis, M.; Kourkouli, P., and Wegmu¨ ller, U., 2009. Day, J.W.; Pont, D.; Hensen, P.F., and Ibanez, C., 1995. Impacts of sea-level Persistent scatterers InSAR to detect ground deformation over Rio- rise on deltas of the Gulf of Mexico and the Mediterranean: the Antirio area (Western Greece) for the period 1992–2000. Journal of importance of pulsing events of sustainability. Estuaries, 18(4), 636–647. Applied Geophysics, 68, 348–355. Dumont, H.J. and El-Shabrawy, G.M., 2007. Lake Borullus of the Nile Piper, D.J.W.; Kontopoulos, N.; Anagnostou, C.; Chronis, G., and Delta: a short history and an uncertain future. Ambio, 36(8), 677–682. Panagos, A.G., 1990. Modern fan deltas in the Western Gulf of Ericson, J.P.; Vorosmarty, C.J.; Dingsman, S.L.; Ward, L.G., and Corinth, Greece. Geo-Marine Letters, 10, 5–12. Meybeck, M., 2006. Effective sea-level rise and deltas: causes of Raucoules, D.; Parcharidis, I.; Feurer, D.; Novalli, F.; Ferretti, A.; Carnec, change and human dimension implications. Global Planetary C.;Lagios,E.;Sakkas,V.;LeMouelic,S.;Cooksley,G.,andHosford,S., Change, 50, 63–82. Ferretti, A.; Prati, C., and Rocca, F., 2001. Permanent scatterers in 2008. Ground deformation detection of the greater area of Thessaloniki SAR interferometry. IEEE Transactions on Geosciences and (Northern Greece) using radar interferometry techniques. Natural Remote Sensing, 39(1), 8–20. Hazards and Earth System Sciences Journal, 8, (4), 779–788. Galloway, W.E., 1975. Process framework for describing the morpho- Salvi, S.; Atzori, S.; Tolomei, C.; Allievi,J.;Ferretti,A.;Rocca,F.;Prati,C.; logical and stratigraphic evolution of deltaic depositional systems. Stramondo, S., and Feuillet, N., 2004. Inflation rate of the Colli Albani In: Broussard, M.L. (ed.), Deltas, Models for Exploration. Houston, volcanic complex retrieved by the permanent scatterers SAR interfer- Texas: Houston Geological Society, 87–98. ometry technique. Geophysical Research Letters, 31(12), L12606 1–14. Hanssen, R.F., 2005. Satellite radar interferometry for deformation Scharoo, R. and Visser, P., 1998. Precise orbit determination and monitoring: a priori assessment of feasibility and accuracy. gravity field improvement for the ERS satellites. Journal of International Journal of Applied Earth Observation and Geoinfor- Geophysical Research, 103, 8113–8127. mation, 6(3–4), 253–260. Shi, C.X.; Zhang, D.; You, L.Y.; Li, B.Y.; Zhang, L., and Zhang, O.Y., Heezen, B.C.; Ewing, M., and Johnson, G.L., 1966. The Gulf of 2007. Land subsidence as a result of sediment consolidation in the Corinth floor. Deep-Sea Research, 13, 381–411. Yellow River Delta. Journal of Coastal Research, 23(1), 173–181. Holzer, T.L. and Pampeyan, E.H., 1981. Earth fissures and localized Shi, X.; Xue, Y.; Wu, J.; Ye, S.; Zhang, Y.; Wei, Z., and Yu, J., 2008. differential subsidence. Water Resources, 17, 223–227. Characterization of regional land subsidence in Yangtze Delta, Jelgersma, S., 1996. Land subsidence in coastal lowlands. In: Milliman, China: the example of Su-Xi-Chang area and the city of Shanghai. J.D. and Haq, B.U. (eds.), Sea-Level Rise and Coastal Subsidence. Hydrogeology Journal, 16, 593–607. Dordrecht, The Netherlands: Kluwer Academic, pp. 47–62. Stanley, D.J., 2005. Submergence and burial of ancient coastal sites on Karymbalis, E.; Gaki-Papanastassiou, K., and Maroukian, H., 2007. the subsiding Nile delta margin, Egypt. Mediterrainee, 104, 65–73. Recent geomorphic evolution of the fan delta of the Mornos River, Stanley, D.J. and Warne, A.G., 1998. Nile delta in its destructive Greece: natural processes and human impacts. Bulletin of the phase. Journal of Coastal Research, 14(3), 794–825. Geological Society of Greece, XXXVII, 1538–1551. Stanley, D.J. and Chen, Z., 1993. Yangtze delta, eastern China: 1. Kuehl, S.A.; Allison, M.A.; Goodbred, S.L., and Kudrass, H., 2005. The Geometry and subsidence of Holocene depocenter. Marine Geology, Ganges-Brahmaputra delta. In: Bhattacharya, J.P. and Giosan, L. 112(1–4), 1–11. (eds.), River Deltas—Concepts, Models and Examples.Tulsa, Stiros, S., 2001. Subsidence of the Thessaloniki (northern Greece) Oklahoma: SEM Society for Sedimentary Geology Special Publica- coastal plain, 1960–1999. Engineering Geology, 61, 243–256. tion, 83, pp. 413–434. Teatini, P.; Tosi, I.; Strozzi, T.; Carbognin, L.; Wegmuller, U., and Louis, I.F.; Karantonis, G.A.; Voulgaris, N.S., and Louis, F.I., 2004. Rizzetto, F., 2005. Mapping regional land displacements in the The contribution of geophysical methods in the determination of Venice coastland by an integrated monitoring system. Remote aquifer parameters: the case of Mornos River delta, Greece. Sensing of Environment, 98(4), 403–413. Research Journal of Chemistry and Environment, 8(4), 41–49. Tinti, S.; Zaniboni, F.; Armigliato, A.; Pagnoni, G.; Gallazzi, S.; Manucci, Lykousis, V.; Sakellariou, D., and Roussakis, G., 2003. Prodelta slope A.; Brizuela Reyes, B.; Bresan, L., and Tonini, R., 2007. Tsunamigenic stability and associated coastal hazards in tectonically active landslides in the western Corinth Gulf: numerical scenarios. In: margins: Gulf of Corinth (NE Mediterranean). In: Locat, J. and Lykousis,V.,Sakellariou,D.,andLocat,J.(eds.),Submarine Mass Mienert, J. (eds.), Submarine Mass Movements and Their Conse- Movements and Their Consequences.NewYork:Springer,pp.405–414. quences. Dordrecht, The Netherlands: Springer, pp. 433–440. Lyon-Caen, H.; Papadimitriou, P.; Deschamps, A.; Bernard, P.; Tornqvist, T.E.; Wallace, D.J.; Storms, J.E.A.; Wallinga, J.; van Dam, Makropoulos, K.; Pacchiani, F., and Patau, G., 2004. First results R.L.; Blaauw, M.; Derksen, M.S.; Klerks, C.J.W.; Meijneken, C., and of the CRLN seismic network in the Western Corinth Rift: evidence for Snijders, E.M.A., 2008. Mississippi delta subsidence primarily caused old-fault reactivation. Comptes Rendus Geosciences, 336(4–5), 343–351. by compaction of Holocene strata. Nature Geoscience, 1(3), 173–176. Massonnet, D.; Rossi, M.; Carmona, C.; Adragna, F.; Peltzer, G.; Werner, C.; Wegmu¨ ller, U.; Strozzi, T., and Wiesmann, A., 2003. Feigl, K., and Rabaute T., 1993. The displacement field of the Interferometric point target analysis for deformation mapping. Landers Earthquake mapped by radar interferometry. Nature, 364, Proceedings of IGARSS’2003, 7, 4362–4364. 138–142. Wu, J.; Shi X.; Xue, Y.; Zhang, Y.; Wei, Z., and Yu, J., 2008. The Mazzoti, S.; Lambert, A.; van der Kooij, M., and Mainville, A., 2009. development and control of the land subsidence in the Yangtze Impact of anthropogenic subsidence on relative sea-level rise in the Delta, China. Environmental Geology, 55(4), 751–762. Fraser River delta. Geology, 37(9), 771–774. Zebker, H.A.; Rosen, P.A.; Goldstein, R.M.; Gabriel, A., and Werner, Meckel, T.A.; Ten Brink, U.S., and Williams, S., 2007. Sediment C.L., 1994. On the derivation of coseismic displacement-fields using compaction rates and subsidence in deltaic plains: numerical con- differential radar interferometry—the Landers earthquake. Jour- straints and stratigraphic influences. Basin Research, 19(1), 19–31. nal of Geophysical Research–Solid Earth B, 10, 19617–19634.

Journal of Coastal Research, Vol. 00, No. 0, 0000

Journal of Coastal Research coas-28-05-09.3d 4/10/11 15:48:30 14 Cust # JCOASTRES-D-11-00106