Land subsidence pattern controlled by old alpine basement faults in the Kashmar Valley, northeast Iran: results from InSAR and levelling

Jan Anderssohn(1) Hans-Ulrich Wetzel(1) Thomas R. Walter(1) Mahdi Motagh(1)
Yahya Djamour(2) and Hermann Kaufmann(1)
(1) GeoForschungsZentrum Potsdam, D-14473 Potsdam, Germany.
E-mail: calypso@gfz-potsdam.de
(2) National Cartographic Center of Iran, Teheran, Iran

Summmary
Fluid storage systems, such as oil, gas, magma or water reservoirs, are often controlled by the host rock structure and faulted terrain. In sedimentary basins, where no direct information about underlying structure is available, the pattern of ground deformation may allow us to assess the buried fault arrangement. We provide an example in the semi-arid area of Iran, in the Kashmar Valley, a region subject to land subsidence due to water overexploitation. Geodetically determined subsidence rates in the Kashmar Valley exceed 15-30 cm yr-1. The pattern of surface deformation is strongly non-uniform and displays NE-SW elongated bowls of subsidence. The trend resembles old Cretaceous-to-Tertiary faults that evolved during early alpine tectonic deformation. Although these early alpine structures are considered tectonically inactive in the present day, the observed land subsidence pattern indicates significant structural control on the geometry of the aquifer basin and its deformation during reservoir drainage.

Introduction
Groundwater reservoirs in arid and semi-arid regions are limited and often subject to intense and uncontrolled groundwater extraction. Currently there are no reliable estimates for the rate and extent of land subsidence in the groundwater basins of Iran. Several studies have used the remote sensing method Interferometric Synthetic Aperture Radar (InSAR) for monitoring of groundwater induced land subsidence (Galloway et al. 1998; Amelung et al. 1999; Galloway et al. 2000b; Lu & Danskin 2001; Hoffmann et al. 2003; Motagh et al. 2007). Compared to conventional techniques for monitoring land deformation (GPS, levelling), the main advantage of InSAR is its ability to provide high resolution (a few tens of metres) maps of crustal deformation with large spatial coverage (>104 km) at subcentimetre accuracy (B¨urgmann et al. 2000), Hansen 2001). In this study, we investigate land subsidence in the Kashmar region, a valley in north-eastern Iran approximately 220 km south of the province's capital, Mashhad (Fig. 1). We shall use InSAR data to measure groundwater-induced deformation in Kashmar and compare the results with precise levelling. These observations allow us to examine in detail the spatio-temporal pattern of subsidence that can be used to infer underlying structural conditions.

Figure 1. Landsat MSS, TM and ETM + colour composite image maps at different times: (A) 1972 October, (B) 1987 August and (C) 2000 September. The red pixels depict vital vegetation. Due to early and ongoing data acquisition of Landsat satellites, a significant increase in vegetation in Kashmar Valley was found for the past 35 yr. The white line in (C) displays the location of the first-order precise levelling line. Buildings and streets appear grey.

Methods
As the first method to investigate land subsidence we use radar data from the ENVISAT satellite in image mode (I2) that has a ground resolution of approximately 20 m, a scene size of about 100 × 100 km2, a fixed incidence angle of about 23°, and a revisit interval of 35 d. We used a total of 22 SAR Single Look Complex (SLC) images for the time interval from July 2003 to March 2006 at the descending satellite track 435.We utilize the 2-pass InSAR method (Massonet and Feigl 1998) implemented in the software package SARscape (www.sarmap.ch, www.creaso.com), with the 90 m resolution digital elevation model (DEM) generated by the NASA Shuttle Radar Topography Mission (SRTM) to remove the topographic phase contribution from the interferometric phase. Twenty-two interferograms with favourable baselines and temporal coverage are processed and unwrapped using the region growing algorithm (Reigber & Moreira 1997). InSAR coherence is good in the study area for time spans over a few months. For time spans exceeding ~200 d, however, we lose coherence probably because of intensive agriculture in the valley. Ground deformation, measured by C-band InSAR (wavelength ~56 mm), is generally visualized in fringes (colour cycles), each fringe representing a displacement of 2.8 cm in the line-of-sight (LOS) of the satellite. For the satellite's incidence angle of 23°, one fringe can be converted into a vertical displacement of 3.1 cm, assuming that the LOS range changes are dominated by vertical deformation alone. The perpendicular baseline of the interferograms ranges from 32 to 353 m, which involves height ambiguity (height differences corresponding to an interferometric fringe) of about 230 and 21 m, respectively. The relative height error (90 per cent probability) of SRTM elevation data over Eurasia is ~8.7 m (Rodr´iguez et al. 2005), resulting in no more than ~0.4 fringes, or 0.14 cm, of line-of-sight error in the interferogram with the longest perpendicular baseline. Moreover, the basin floor is very flat that helps mitigate any contributions from topography-related artefacts in interferograms. Other interferograms not included in this study have either poor coherence or unfavourable baselines. Using all suitable data we analysed the spatio-temporal evolution of subsidence in Kashmar Valley. In order to validate the results of space-borne remote sensing data, the InSAR measurements are compared to independent ground truth observations obtained using levelling. The National Cartographic Center of Iran (NCC) collected levelling data in the Kashmar Valley over a 10-yr time span between 1993 and 2003. The first survey was done during 1993 August-October and the second survey from May to July in 2003. Fig. 1(c) shows the location of the levelling line along the Kashmar Valley. The levelling surveys were conducted according to specifications required for first-order precise levelling; the expected standard deviation for the change in elevation differences between two benchmarks as measured in two surveys can be expressed as ±a L1/2, where a = 1 mm km-1/2 and L is the distance in kilometres along the levelling line between the benchmarks (Memarzadeh 1998). Analysis of ENVISAT's MERIS (Medium Resolution Spectrometer) multispectral measurements - done with the software solution ENVI - still show a slight increase in the vegetation cover of the spring season, as defined by the change in the value of normalized difference vegetation index (NDVI) (Tucker 1979).

Results

Figure 2: SAR differential interferogram of land-surface deformation over Kashmar Valley, Iran, during the period 2005 April 28, to 2005 July 7 (interferogram No. 16 in Fig. 3) superimposed on a corresponding amplitude image. Each full colour cycle (fringe) represents subsidence of 2.8 cm in line of sight of the satellite. Local heterogeneities of differing subsidence behaviour are clearly visible. One fringe for the valley and additional fringes for the local heterogeneities can be found, yielding subsidence of about 3 cm for Kashmar Valley and more than 6 cm for local heterogeneities.

Figure 3: Four examples of wrapped interferograms with time spans (A - 35, B - 70, C - 105 and D - 140 d). The evolution of the subsidence distribution pattern with increasing time is evident in these four interferograms, showing that the amount of displacement increases with time while the general location appears to be constant.

Figure 4: Cumulative (~2 yr) displacement map resulting from a consecutive time-series of interferograms, also shown with elevation contour lines. An established pattern of subsidence distribution is visible, indicated by local bowls and widespread homogenous subsidence. A maximum subsidence of about 58 cm is obtained at local bowls, whereas most of the area (60 per cent) is affected by subsidence up to 30 cm.

Conclusion
Our analysis shows subsidence in the east-west elongated Kashmar Valley in northeast Iran. During the study period 2003-2006, subsidence affected an area of about 750-800 km2, with common subsidence rate of 15 cm yr-1. At several isolated locations, however, the rates are even up to twofold larger, reaching almost 30 cm yr-1.We also observe that these locations probably correlate to technical buildings and well locations. Zones of widespread subsidence bowls, which are punctuated by very localized maxima, are generally elongated NE-SW along the axis of the valley. We suggest that the NE-SW elongated subsidence pattern is governed, if not controlled, by old buried faults of Cretaceous-to-Tertiary age beneath or within the sedimentary valley infill. Buried faults may accordingly form a structured aquifer basin system varying in thickness and recharge rate. Taking into account the ongoing and increased subsidence inferred by levelling and measured by InSAR, the increased demand for groundwater caused by increasing agriculture, global climate change and precipitation decline, we expect that the overexploitation of groundwater in Kashmar is likely to become a major factor for development and prosperity in this region.

Acknowlegements
We would like to thank the National Cartographic Center of Iran for providing levelling data and supporting our work. We also thank Mrs S. Chabrillat and N. Richter for advice concerning imaging spectrometry. The paper benefited from constructive comments by J.W. Bell and an anonymous reviewer. ENVISAT-ASAR, -MERIS data were provided by the European Space Agency under Category-1 research proposal No. 3188 and 2892. Mahdi Motagh was supported by a German Research Foundation grant MO 1851/1-1.

References
Anderssohn, J.; Wetzel, H.; Walter, T.; Motagh, M.; Djamour, Y. & Kaufmann, H. (2008). Land subsidence pattern controlled by old alpine basement faults in the Kashmar Valley, northeast Iran: Results from InSAR, levelling and imaging spectrometry. Geophysical Journal International (2008) 174, 287-294