This paper has to be cited as: Tomás, R., Herrera, G., Mulas, J., Cooksley, G. Persistent Scatterer Interferometry subsidence data exploitation using spatial tools: the Vega Media of the Segura River Basin case study. Journal of Hydrology, 400, 411-428. 2011. The final publication is available at Elsevier via: http://dx.doi.org/10.1016/j.jhydrol.2011.01.057
Persistent Scatterer Interferometry subsidence data exploitation using spatial tools: the Vega Media of the Segura River Basin case study
R. Tomas1, G. Herrera2, G.Cooksley3, J. Mulas2 (1) Departamento de Ingeniería de la Construcción, Obras Públicas e Infraestructura Urbana, Escuela Politécnica Superior, Universidad de Alicante P.O. Box 99, E-03080 Alicante, Spain. E-mail:
[email protected] (2) Área de Investigación en Peligrosidad y Riesgos Geológicos, Departamento de Investigación y Prospectiva Geocientífica, Instituto Geológico y Minero de España (IGME), Ministerio de Ciencia y Tecnología, c/ Alenza 1, E-28003 Madrid, Spain, e-mail:
[email protected];
[email protected] (3) Altamira Information, c/ Còrsega, 381-387, 2n 3a - E-08037 Barcelona, Spain. E-mail:
[email protected]
Abstract The aim of this paper is to analyze the subsidence affecting the Vega Media of the Segura River Basin, using a Persistent Scatterers Interferometry technique (PSI) named Stable Point Network (SPN). This technique is capable of estimating mean deformation velocity maps of the ground surface and displacement time series from Synthetic Aperture Radar (SAR) images. A dataset acquired between January 2004 and December 2008 from ERS-2 and ENVISAT sensors has been processed measuring maximum subsidence and uplift rates of -25.6 and 7.54 mm/year respectively for the whole area. These data have been validated against ground subsidence measurements and compared with subsidence triggering and conditioning factors by means of a Geographical Information System (GIS). The spatial analysis shows a good relationship between subsidence and piezometric level evolution, pumping wells location, river distance, geology, the Arab wall, previously proposed subsidence predictive model and soil thickness. As a consequence, the paper shows the usefulness and the potential of combining Differential SAR Interferometry (DInSAR) and spatial analysis techniques in order to improve the knowledge of this kind of phenomenon.
Keywords: subsidence, DInSAR, SPN, PSI, spatial analysis, conditioning and triggering factors.
3
1. Introduction Subsidence is a natural or anthropogenic hazard that produces a downward displacement of the ground surface over wide areas. It can be triggered by the excavation of ground tunnels or mining galleries, soil/rock dissolution, fluid withdrawal (petroleum, water or gas), deep erosion (piping), lateral soil creep, compaction of soil materials or tectonic activity. In this work subsidence due to soil consolidation induced by aquifer overexploitation is investigated. This type of subsidence is mainly due to the consolidation of aquitard layers composed of fine lithologies (clays and silts) during prolonged drainage periods, and also to the compaction of coarse fractions (sands and gravels). Subsidence due to water withdrawal is an extensive worldwide phenomenon affecting large areas and important cities such México City DC (Mexico), Bangkok (Thailand), Po Plain (Italy), Antelope, Santa Clara and San Joaquin Valley (USA), Nobi Plane, Niigata and Suzhou (Japan), Chosui River and Taipei (Taiwan), Shanghai, Tianjin, Beijin (China), Jakarta (Indonesia) among others. In Spain only one case of general subsidence due to water withdrawal has been reported at the Vegas Baja and Media of the Segura River. The monitoring of this phenomenon is a very important task in order to establish the mechanisms, the causes and the velocities of the subsidence process. This is useful to prevent damages on infrastructures and for land-use and water-resources planning. Moreover, subsidence measurement is necessary in order to determine the affected areal extent, to quantify the deformation or settlement velocities, to identify mechanisms and critical states of failure, and to evaluate the effectiveness of the corrective measures adopted (Tomás et al., 2005a). Murcia City, located in the Vega Media of the Segura River, was the first case in Spain where subsidence induced by piezometric changes due to aquifer overexploitation caused important damages (50 million euros) in human infrastructures (Rodríguez and Mulas, 2002; Mulas et al., 2003) during the drought period 1992-95. From 2004 to 2008 an important piezometric decline affected this area triggering additional subsidence. This work is focused on the subsidence that occurred during 2004-2008 period in the Vega Media of the Segura River located in SE Spain. We examine the evolution of ground deformation and its relation to the
3
pumping activity in this area generally during drought periods, describing the results obtained with the Stable Point Network (SPN) technique. These results will be compared with piezometric measurements, local geology, soft soil thickness, well point distribution, the river distance, the Arab city location, the basement location and the geotechnical Finite Element Method (FEM) model developed by IGME (2000a). The paper is organized as follows. Section 2 describes the geographical and geological setting of the study area. Subsidence data obtained from interferometry is presented and briefly discussed in Section 3. Then, Section 4 is dedicated to compare this data with other conditioning and triggering factors. A discussion of the results is included in section 5. The main conclusions are summarized in Section 6.
2. Description of the study area
The Vega Media of the Segura River (VMSR) is located in the East sector of the Betic Cordillera, in the so called Bajo Segura basin (Montenat, 1977; Figure 1). The materials outcropping along the boundaries of the valley vary with location (Figure 2). The South border of the VMSR consists mainly of rocks of the basin basement (Permian to Triassic) and slope deposits (Pleistocene). Meanwhile, the Northern border is formed mainly by sedimentary rocks (Upper Miocene to Pliocene) deposited in the basin (marls, sandstones and conglomerates). The valley is filled by recent materials (Holocene at ground surface, Pleistocene to Pliocene at some depth) deposited by the fluvial action of Segura and Guadalentín rivers.
These recent sediments are potentially deformable and the most problematic from a geotechnical point of view. Rodríguez Jurado et al. (2000) and Mulas et al. (2003) made a geotechnical characterization of all these materials for the VMSR showing that the sedimentary rocks protruding at the valley borders, which are also found at some depth within the flood plain, are characterized by low to negligible deformability. Above them, the recent shallow sediments are characterized by moderate to high compressibility.
From a hydrogeological point of view, the VMSR belongs to the so-called “Guadalentín – Segura Quaternary aquifer System Nº 47” (IGME, 1986). This aquifer is characterized by two units (Cerón and Pulido, 1996; Aragón et al., 2004): a) a shallow unit that consists of a semiconfined aquifer or aquitard
4
formed by silts and clays with occasional sand intercalations, and b) a “deep aquifer”, consisting on a multilayer aquifer system composed of sand and gravels confined by less permeable layers.
Figure 1. Geological setting of the Vega Baja and Media of the Segura river Basin (based on Montenat (1977) and Aragón et al. (2004)). Piezometric levels for boreholes (1 to 4) are shown in Figure 3a.
The deep aquifer, scarcely studied, is stratified and comprises several gravels of hydrological interest (CHS, 2007). The upper gravel in the deep aquifer is heavily exploited at about 20 m below the surface, although recent pumping wells drilled in 2004 (CHS, 2007) extract water from the lower gravel too.
5
When water is pumped from the upper gravels of the deep aquifer, whose piezometric level coincides with the one of the shallow unit in an undisturbed state (without pumping), a gradient is created and that implies a water flow from the shallow unit to the upper gravels of the deep aquifer. Consequently, pore pressure on the shallow unit falls and the soil suffers a consolidation process (Herrera et al, 2009a). The piezometric evolution of the shallow unit and the upper gravel of the deep aquifer is shown in Figure 3a and b.
Figure 3. a) Historical evolution of the piezometric level of the upper gravel of the deep aquifer. b) Temporal evolution of piezometric level of the shallow unit and the upper gravel of the deep aquifer. See piezometers location in Figures 1 and 2a.
6
The piezometric level in the upper gravel of the deep aquifer is found a few meters below ground surface showing significant variations over time (Figure 3a and b). These fluctuations were particularly noticeable during the period 1992-1995 (Figure 3a), where the minimum piezometric levels reached -15 m (Figure 3a). As a consequence, widespread subsidence affected the VMSR, causing damage to structures and a great public concern (Mulas et al., 2003; Martínez et al., 2004). Tomás et al. (2005b; 2008) and Herrera et al. (2009 a,b) measured ground subsidence in the metropolitan area of Murcia during this drought period by means of differential Synthetic Aperture Radar Interferometry (DInSAR), detecting maximum displacements of 12 cm in the zone. Since 2004, an important piezometric level decrease has affected the VMSR. This piezometric level has declined as much as 12 m with respect to the previously existing piezometric level with an average decline of 8 m causing soil consolidation and affecting infrastructures. Figure 4 shows several examples of the recent damages caused by subsidence in the city of Murcia. Figures 4a and c show sidewalk deformations. This deformation indicates the existence of a differential settlement between buildings piles and the terrain. A shear stress is produced along the interface of the pile and the soil known as negative skin friction. Negative skin friction produces an additional load which can exceed the allowable load of the piles inducing additional settlements (Vázquez and De Justo, 2002; De Justo et al., 2002). Notice the date of construction of the ramp marked in the concrete in Figure 4c (23-03-2007) indicating that subsidence has affected this area after that date. Figure 4b shows another typical pathology consisting of the monolithic tilt of buildings due to the differential settlement of their foundations, favoured by soil heterogeneity. This type of pathology is usually associated with rigid slabs foundations. Figure 4d shows the third common type of damage. It consists of the development of cracks on walls due to the different settlements of the foundations that cause angular distortions on structure. This kind of damage is linked with footprint foundations and non rigid slabs.
7
Figure 4. Recent damages caused by subsidence in the Murcia City.
8
3. Advanced DInSAR applied to Vega Baja of the Segura River subsidence study 3.1 Stable Point Network Technique: basic concepts and processing. The differential interferometric phase (int) obtained by combining two complex SAR images can be expressed as an accumulation of five terms (Hanssen, 2001):
int flat topo mov atmos noise
(1)
where flat is the flat-earth component related with range distance differences, topo is the topographic phase, mov is the phase component attributed due to ground displacement that occurred between the two considered SAR image acquisitions measured along the line of sight (LOS), atmos is the phase component due to atmospheric instability, and, finally, noise includes the rest of noise sources.
topo can be extracted from an external Digital Elevation Model (DEM) and flat can be expressed analytically easily. Consequently, assuming that noise and atmos are known we can determinate the phase contribution due to ground deformation (mov). This is the fundamental process of conventional DInSAR techniques that has been widely used for subsidence studies providing satisfactory results. Nevertheless, atmospheric and noise phase contribution are unknown and we can only estimate the deformation terms. The atmospheric effects superimpose on the signal of displacement supposing a serious limitation of differential interferometry, especially when the terrain deformation is small and spatially smooth (Ferretti et al., 2000). Possible temporal decorrelation caused by atmospheric artifacts, can lead to noise in the generated interferograms and to unexploitable measurements. Advanced DInSAR techniques, usually known as Persistent Scatterer Interferometry (PSI) techniques, aim at overcoming those limits. These
techniques, constitute a family of algorithms that are based on the
simultaneous processing of multiple interferograms derived from a large set of SAR images (Ferretti et al., 2000, 2001; Berardino et al., 2002; Mora et al., 2003; Arnaud et al., 2003; Werner et al., 2003; Hooper et al., 2004; Lanari et al., 2004; Blanco et al., 2008). These techniques can remarkably improve the quality of the DInSAR results because they allow the minimization of the atmospheric and noise phase contribution and, as a consequence, improve the estimate of the ground deformation phase component.
9
In this work, ground subsidence measurements were obtained with a PSI technique called Stable Point Network (SPN). A detailed description of the technique can be found in Arnaud et al. (2003) and Duro et al. (2005) but a summary is included here for the sake of completeness.
The SPN software uses the DIAPASON interferometric algorithm for all SAR data handling, e.g. coregistration work and interferogram generation. The SPN procedure generates three main products starting from a set of Single Look Complex (SLC) SAR images (Duro et al, 2005): a) the displacement rate, i.e. the average deformation velocity, along line of sight (LOS) of single PS that can be derived using a dataset of at least 6 images; b) a map of height error; and c) the LOS displacement time series of individual PS (as a function of time), which requires at least 20 images, depending on the velocity of displacement with respect to the temporal separation of image acquisitions. Nevertheless, an increase in the number of SAR images improves the quality of the measurements.
3.2 DInSAR subsidence results In this work, the SPN algorithm has been applied to 51 images acquired by the European Space Agency ERS-2 and Envisat ASAR sensors covering the period January 2004-December 2008. A subarea of about 20 km x 8 km was cropped from the original images, corresponding to the VMSR. An amount of 115 interferograms has been selected with a perpendicular spatial baseline smaller than 800 m, a temporal baseline shorter than 3 years and a relative Doppler centroid difference below 400 Hz. The Shuttle Radar Topography Mission (SRTM) DEM of the study area has been used. In this case the pixel selection was based on a combination of several quality parameters including low amplitude standard deviation (≤ 0.40) and high model coherence (≥ 0.45) providing a Persistent Scatterers (PS) density of 50 PSs per square kilometer. The 2004-08 subsidence rate (projected along the line of sight, LOS) obtained by the SPN technique is presented in Figure 5. The deformation map has been geocoded and superimposed on the available DEM and is shown in 10 colors that represent intervals of deformation. Note that reliable information is available only on urban areas and rock outcrops.
10
Maximum and minimum deformation rates in the whole area for the 2004-2008 period are -25.6 and 7.54 mm/year respectively. In absolute terms, accumulated deformations vary from -127.6 to 33.23 mm where negative and positive values represent subsidence and uplift respectively.
Figure 5. Subsidence map (mm/year) of the Vega Media of the Segura River Basin for the 2004-2008 period.
Regarding the spatial distribution of radar-derived measurements, it can be appreciated in Figure 5 that the highest density of PSs is found on the urban areas, whereas a few PSs have been detected on agricultural fields. This is because the response of the buildings to the satellite radar signal is stable in time, while the seasonal variability of the ground surface of rural fields provides an unstable radar response in time, and hence no persistent scatterers are detected.
11
In the period 2004 - 2008 a generalized subsidence is observed in the valley. The areas South and East in the Murcia City are mostly affected by subsidence with average deformation rates up to -6 mm/year and maximum rates of -16.9 mm/year. The Eastern area of the VMSR shows the highest values of subsidence with average and extreme values of -11.8 and -25.6 mm/year, respectively. Note that some of the deformation measurements might be due to settlements caused by construction, especially in surrounding areas of the city subjected to urban growth. Even though this aspect cannot be ignored, this type of subsidence is usually particularly localized and short-lived.
3.3 Comparison with the extensometers time series In this section PSI subsidence data have been validated by comparing them against in situ extensometer measurements. In February 2001 a total of 22 extensometers were installed by IGME (2001) with a monitoring depth of about 15 m providing subsidence measurements with 0.1 mm accuracy. The PS results accuracy has been assessed by comparing the retrieved deformations from the SAR data analysis with the only five operative extensometers from April 2004 to March 2009 measurements projected along the Line of Sight (Figure 6). Two quality parameters have been computed (Table 1): (1) the mean and standard deviation of the absolute difference between PS and LOS-projected extensometer deformation time series (5.9 ± 4.1 mm), (2) the mean and standard deviation of the difference between PS and LOS-projected extensometer deformation time series (-3.6 ± 5.1 mm). These values are in agreement with the ones obtained by other authors in similar experiences (e.g. Strozzi et al. (2001), Casu et al. (2006), Herrera et al. (2009a and b), Stramondo et al., 2007; Raucoules et al., 2009). Figure 6 shows the temporal evolution of the deformations measured for six extensometers projected along LOS and those estimated with the SPN technique. Notice that DInSAR subsidence data from a previously studied period (1995-2005) have been included in the plot in order to provide a vision of the long-term behavior of the aquifer system. A detailed description of the 1995-2005 subsidence data can be found in Herrera et al. (2009b).
12
Figure 6. Time series of the displacements estimated with the SPN technique and with the Line of Sight (LOS) projected measurements from the extensometers. d is the PS-extensometer distance. See extensometer location in Figure 16.
A good agreement can be observed between the ground truth and the SAR based estimations. Nevertheless V3 and Ei1 extensometers show smaller deformations than those estimated with the SPN technique since approximately October 2004. In this moment, coinciding with the start of the most recent drought period, ground water was withdrawn from the deep aquifer. A plausible explanation of these
13
differences is that the SPN technique is measuring total deformations at these points that are near two drought wells whereas the extensometers are measuring deformations only in the first 15 m below the surface not taking into account deeper compressible layers deformation (Tomás, 2009).
4. Subsidence conditioning and triggering factors spatial analysis Pumping-induced subsidence is determined by geometrical, geological, hydrogeological and geotechnical factors. These factors are usually divided into: conditioning or passive factors, which are intrinsic to the materials of the aquifer system, and triggering or active factors that cause subsidence once certain circumstances concur. The following subsections, discuss spatial and / or temporal comparisons of subsidence with several conditioning factors (geology, distribution of water extraction points, the Segura River distance, the Arab city location and soft soil thickness) and a triggering factor (piezometric level). PS results will be also compared with the geotechnical Finite Element Method (FEM) model developed by IGME (2000a) and with the basement distribution. Note that the analyses included in the next subsections have been performed not taking into account PSs with uplift and considering that the stable areas are those characterized by PSs with deformation rates between ±2 mm/year. This criterion, used for the definition of stable areas, is based on the accuracy of the PS techniques, in terms of linear velocity, computed by other authors (e.g. Lanari et al., 2007; Stramondo et al., 2007; Raucoules et al., 2009).
4.1. Comparison with piezometric level evolution Usually groundwater exploitation in urban and agricultural areas of the VMSR increases during drought periods. Meanwhile ground water recharge is reduced significantly due to the reduction of precipitations, and thus triggering the drop of the piezometric levels. The temporal evolution of DInSAR measurements shows a good relationship with piezometric level changes. Figure 7 shows the correlation between DInSAR measured Line of Sight (LOS) surface
14
deformation and the piezometric level of the upper gravel of the deep aquifer for six representative sites located over deformable sediments.
Figure 7. Piezometric level and DInSAR subsidence correlation (d is the PS-piezometer distance and D is the accumulated subsidence for the studied period).See piezometers location in figure 2.
During this period (2004-2008) piezometric level shows a general decreasing trend of 8-10 meters with seasonal fluctuations of 4-6 meters related to wet and dry periods (Figure 7). Note that a general
15
subsidence decline trend with some small uplift and subsidence that deviates from the general trend is shown in Figure 7. A stress-displacement relationship (Sk), that provides the linear relation between subsidence (D) and water level decline (h), has been computed for the six piezometric-subsidence temporal series shown in Figure 7 using the expression:
Sk
D h
(2)
Following expression (2), a coarse estimation of the skeletal storage coefficient (S k) has been computed as the rate of the slopes of the straight line fitted to the whole piezometric and PSI 2004-2008 data plotted in Figure 7. Sk represents the deformability of the aquitard and varies with the stress state, adopting a different value depending on the piezometric level (H) and its position above or below the minimum recorded piezometric level - preconsolidation head - (Hp) (Hoffmann et al., 2001; Tomás et al., 2010). Hp separates the elastic recoverable subsidence from the inelastic, unrecoverable, subsidence:
S if H H p S k ke S kv if H H p
(3)
The computed skeletal storage coefficients are assumed to be inelastic (Skv) because they have been computed for the period of maximum known piezometric decline (Figure 3a). The minimum and maximum calculated inelastic skeletal storage coefficient (Skv) of the aquifer system, are 1.1 x 10-4 and 5.7 x 10-3 respectively with an average value of 3.7 x 10-3. These values are similar to those computed by Tomás et al. (2010) using stress-displacement curves for the 1993-1995 time period in the City of Murcia using CPT DInSAR technique that vary from 4.9 x 10-4 to 3.1 x 10-3 with an average value of 1.5 x 10-3.
4.2. Comparison with geology As discussed in section 2 the VMSR basin is filled by recent Holocene sediments (Figure 2). These sediments belong mainly to the flood plain and the channel of the Segura River where high deformation rates are expected. The bordering and adjacent deposits of the flood plain are composed of older and
16
less deformable materials belonging to the Pleistocene (sedimentary formations as alluvial fan, colluviums, piedemonts, etc.) and the Permotriassic (Figure 2).
Figure 8. Deformation rates of the different geological units. See geological units distribution in Figure 2.
A spatial analysis has been performed in order to determine the subsidence related to each geological unit. The greatest amount of pixels has been detected on the flood plain zone of the Segura River
17
(54.4%). These points show an average deformation rate of -5.8±3.4 mm/year (Figure 8). For the same period, maximum average subsidence rates have been measured over the abandoned fluvial channel (terraces and abandoned meanders) with an average value of -6.1±2.9 mm/year (Figure 8). However, the greatest extent of this geological unit in the study area is low with an area of 7.4 km2 and a PSs concentration of only 1.9%. Older lithologies located on the borders of the basin (Permo-Triassic and Pliocene units) show low values of average subsidence rates varying from -1.7 to -3.1 mm/year (Figure 8). Therefore these results demonstrate, as expected, that young units composed of unconsolidated sediments have been affected by subsidence whereas older materials are relatively less affected or unaffected by this process.
Figure 2. (a) Geological map of the Vega Media of the Segura river (based on IGME, 2000b). Piezometric levels for boreholes (1 to 4) are shown in Figure 3a.
18
4.3. Comparison with soft soil thickness The thickness of soft sediments that fill the VMSR flood plain was established by IGME (2000b) using geotechnical information obtained from hundreds of boreholes drilled in the study area. For the elaboration of this map, soft soil thickness has been established as the sum of the silt and clay layers located over the upper layer of gravels of the deep aquifer. The results show that soft sediments thickness increases towards the centre and the East of the valley reaching more than 25 m thick in some points of the Eastern areas (Figure 9). On the other hand, in the areas bordering the flood plain only thin soft soil is found.
Figure 9. Soft soil thickness map (IGME, 2000).
19
The relationship between subsidence rates of all the PSs located within the different compressible thickness areas has been computed. Figure 10 shows, as it was expected, that higher subsidence is found in areas where soft soil is thicker. One can observe that cumulated subsidence varies in the range 3.7 ± 2.5 mm/year and from -9.5 ± 3.4 mm/year for soft soil thickness varying from 0 to 2 m and >25 m, respectively.
Figure 10. Relationship between soft soil thickness and subsidence rates projected along LOS.
4.4. Comparison with Segura river distance In order to analyze the relationship between ground subsidence and the fluvial system of the valley, we have computed the average and standard deviation of the deformation rates measured in PSs included within the different buffer areas defined by the distance to the Segura River. In Figure 11 it is observed that the maximum average subsidence rates (from -5.7 to -6.2 mm/year) are found in those PSs included within the area defined by a 1500 m buffer from the Segura river course. This spatial analysis evidences
20
how subsidence rate decreases proportionally to river distance. A similar relationship was noticed by Hu et al. (2004), Manunta et al. (2008) and Chen et al. (2007) showing that subsidence is higher near the river than far from it, partially due to the existence of recent soft and unconsolidated flood or overbank deposits related with the activity of the river.
Figure 11. (Maximum (Max), minimum (Min), average (Mean) and standard deviation (±) of subsidence
4.5. Comparison with well distribution Wells provide a water source during prolonged drought periods. Hundreds of wells have been drilled along the VMSR basin. More than 600 wells are estimated to be placed in the VMSR. These wells are generally domestic and agricultural wells that extract water from the shallow aquifer. Unfortunately little precise information about them is available. Nevertheless it is documented that up to 100 wells withdraw groundwater from the deep aquifer (CHS, 2007). Figure 12 shows the location of the documented wells. If
21
we analyze the proximity of these wells to the Segura River axis, we observe that the greatest concentration of wells (35%) is closer than 500 m from the river (Figure 13) and 80% are closer than 2000 m.
Figure 12. Distribution of the pumping wells of the study area.
In 2004 the Water Hydrographic Authority drilled a total of 24 wells (CHS, 2007) in order to maintain water supply during drought periods. These wells are called “drought wells” and have depths varying from 138 to 304 m and an average value of 202 m (CHS, 2007). At least 71 % of them were drilled closer than 500 m from the river axis. These wells currently withdraw important water flows from the deep aquifer (gravel unit) with depths greater than 240 m, pumping 80 to 160 l/s with an average value of 111.8 l/s (CHS, 2007).
22
Municipal Authority of Murcia city drilled 54 wells for urban uses (public gardens watering, street clean-up, etc.) during 1993-95 drought period. These wells pump less than 10 l/s from the first aquifer gravel unit of the deep aquifer system (CHS, 2007). Note that municipal wells are erratically distributed in Murcia City, with 70% of them closer than 2000 m from the river axis.
Figure 13. Inventoried pumping wells distribution respect to the Segura river axis.
A total of 10 industrial wells used for car cleaning-houses, refrigeration, etc. (CHS, 2007) are inventoried in the VMSR, with 60% of them closer than 500 m to River Segura. These wells withdraw water probably from the upper aquifer of the deep aquifer system unit because their depths vary from 32 to 40 m, with the average depth of the pumps at 18.5 m (CHS, 2007). Finally, 45 legalized agricultural wells have been considered, with 82% of them closer than 2000 m from the river axis. The average pumping flow of these wells is 55 l/s from an average depth of 31 m. In this case groundwater is used to irrigate 715 Hectares (CHS, 2007).
23
In order to analyze the relationship between ground surface deformation and the distance to the wells, we have computed the average and the standard deviation of the deformation rate measured in those PSs included within the different buffer areas defined by the distance from the wells to the Segura River.
Figure 14. Variation of subsidence rates from agricultural, drought, industrial and municipal wells axis.
24
Figure 14 shows that generally the maximum subsidence rates are found near the wells, gradually decreasing with the distance from them. For the agricultural wells, from 0 to 1750 m from the wells, maximum average subsidence of as much as -5 mm/year is observed. A similar relation can be observed for drought wells recently drilled in the study area with subsidence occurring from 0 to 3000 m, with a maximum average subsidence rate of -7.3 mm/year located 200 m from the wells. Industrial wells show similar subsidence values as the agricultural wells with a higher influence area (up to 2500 m from the wells). The municipal wells show a small subsidence influence area with a radius of 500 m. A second subsidence influence area is observed from 1250 m to 1750 m distance from the well. In general, it is observed that the shape of the deformation curve with respect to the distance to the wells is similar to the depletion curve of a pumping test. Nevertheless, we have to consider that the spatial variability of soils (deformational properties and thickness) can distort the shape of the cone of depletion, causing deviations on the results of this spatial analysis. Furthermore, it is a well-known fact that the interference of two pumping cones of depression causes a water head decline equals to the sum of the water head declines caused by each one, therefore increasing the magnitude of the settlements on these points. As expected, drought wells, that extract important water flows from significant depths, cause the highest subsidence rates and affect greater extensions of the ground surface. On the contrary, municipal wells, that only produce small quantities of water from the uppermost layer of gravels, originate lower maximum subsidence rates and affect the smallest area.
4.6. Comparison with Arab city location The origin of the city of Murcia dates back to the time of the Arab period of Abderraman II (8th and 9th centuries). The old city was developed on the left side of Segura River and was protected by a defensive wall (Figure 12). This wall had an external defensive moat 2-4 m depth called “cava” that was filled due to the city historical expansion. The progressive growth of the city has been always linked to the Segura River (Figure 12). Nowadays, the city of Murcia occupies part of both sides of the river covering a wider extension up to 12 km2.
25
Previously occupied areas of the city are expected to be overconsolidated and therefore less subsidence, due to the increase of effective stresses caused by water pumping, should occur compared with the rest of the city. A spatial analysis has been performed in order to study this effect (Figure 15). Obtained results show that average subsidence rate inside the wall perimeter (-3.5 ± 1.1 mm/year) is lower than outside (-5.4 ± 2.0 mm/year). The accumulated subsidence for the same period show similar results with average accumulated deformations of -16.5 ± 5.8 mm and -25.6 ± 9.9 mm, respectively.
Figure 15. Subsidence rates projected along LOS inside and outside the Arab defensive wall of the city.
4.7. Comparison with the IGME subsidence model The Geological and Mining Survey of Spain –IGME– developed a one-dimensional coupled model for subsidence prediction using Finite Elements Method, FEM, (IGME, 2000a; Mulas et al., 2003). This model assumes that subsidence occurs when water is withdrawn from the topmost layer of gravels of the deep aquifer. As a consequence of pumping, a vertical gradient is created that causes a descending flow of ground water from the shallow aquifer (surface unit) towards the upper gravel layer in the deep aquifer
26
causing a piezometric level decline. As a result, pore pressure in the shallow aquifer system goes down causing the consolidation of the aquifer.
Figure 16. Settlement map obtained using a FEM for a 10 m progressive piezometric fall during 2 years (IGME, 2000).
This model computes subsidence for several stratigraphic soil columns that have been previously defined, taking into account the silt and clay thickness, the intercalated sand layers, the temporal evolution of the piezometric level and the geotechnical properties of the different soil layers. A two years progressive piezometric level decline has been considered in the subsidence model for several maximum values (5, 10, 15, 20 and 25 m). The settlements map obtained by IGME (2000a) using the above mentioned model for a 10 m piezometric drop is shown in Figure 16. These values are expected to be similar to those caused by
27
2004-2008 piezometric crisis that, as previously discussed, caused a real average piezometric decline of 8 m with extreme declines of 12 m.
Figure 17. Comparison between predicted subsidence using a FEM model and deformation rate obtained from DInSAR.
Figure 17 shows the results of the cross analysis of 2004-08 DInSAR results and the predicted settlements using FEM. Maximum average DInSAR subsidence rates correspond to the areas of the model where maximum settlements are expected, whereas minimum average values correspond to low settlement areas of the model with good correlation for all intermediate situations. Cumulated DInSAR subsidence values have been directly compared with the FEM predicted subsidence. Figure 18 shows a very good correlation between both kinds of data with differences that are generally less than 1 cm. The greatest differences correspond to those values of subsidence in the model with a small areal extent where only a few number of PSs are available.
28
Figure 18. Comparison between predicted subsidence using a FEM model and accumulated deformation obtained from DInSAR for the 2004-2008 period.
4.8. Comparison with the basements and urban tunnel location A spatial analysis has been performed in order to analyze the influence of the number of floors, which vary from 1 to 5, of underground basements and urban tunnels of the city of Murcia (Figure 19). Basements pump low water flows from chests located on the upper part of their slabs in order to reduce sub-charges and basements flooding that are directly spilled to the public sewage. Several authors pointed out that this pumping could contribute to the subsidence process developed due to the caused piezometric depletion. However, no significant relationships have been observed for this factor.
5. Discussion The calculation of ground settlement is a common task in subsidence studies. Settlements are usually calculated by considering three parameters: (a) the stress state increase, (b) the thickness of potentially deformable soil, and (c) a deformation modulus that relates the two previous parameters. If we consider
29
that variations in stress state are due to piezometric level changes, this relationship can be written as (Hoffmann, 2003; Hoffmann et al., 2003; Galloway and Hoffmann, 2007; Tomás et al., 2010):
D h D S sk h S k
(4)
where D is the settlement of a deformable layer D meters thick due to a h piezometric level decrease (all expressed in m). Sk and Ssk are the skeletal storage coefficient (dimensionless) and the specific skeletal storage coefficient of the aquitard (m-1) respectively. As it was previously mentioned in subsection 4.1, Sk or Ssk represents the deformability of the aquitard and depends on the position of the piezometric level compared to the maximum historical recorded piezometric decline.
Figure 19. Basements and urban tunnels distribution on the city of Murcia.
One variable considered in the expression (4) for subsidence computation is piezometric level decline (h), which is the responsible of effective stress increase. A direct relationship between subsidence and
30
piezometric level decline has been demonstrated for several time series of piezometric level evolution. Pumping well spatial distribution has been also used to relate subsidence with the piezometric level decline because it is responsible of water head descent. The analysis mostly reveals the higher rate of subsidence near the wells than far from them according with the expected theorical shape of the depletion curves around a pumping well. Subsidence has been also demonstrated to be larger near the Segura River than far from it due, principally, to the existence of an important quantity of wells near the river channel. The direct relationship between soft soil thickness and ground subsidence, according to expression (4), has been proved by means of the simple spatial analysis conducted in section 4.3 which shows that the thinner the accumulated soft soil is the lower the subsidence. In expression (4) the third parameter that affects subsidence is the specific skeletal storage coefficient (Ssk) related with soil deformability. This parameter depends on the nature of materials and its geological history. Usually, older and consolidated lithologies present a minor deformability (minor S sk) than younger and unconsolidated sediments. In section 4.2 we have shown that the higher values of subsidence correspond to the younger Holocene deposits that fill the valley in comparison with the border area and the reliefs, where generally older rocky lithologies are present, showing low subsidence values. The deformations measured inside the old defensive Arab wall of the city of Murcia are lower than those measured in the rest of the city probably due to the soil overconsolidation induced by the buildings of the old city. This is because under a certain stress a higher deformation is expected over normally consolidated (computed using Skv) soil than over an overconsolidated soil (computed using Ske). Concluding, PSI subsidence data has been used to study the spatiotemporal evolution of the subsidence affecting the VMSR allowing to perform a simple spatial analysis of the most common conditioning and triggering factors that control subsidence mechanisms due to ground water extraction. In the future, the jointly use of deformational, hydrological and geotechnical data will allow to rank the different conditioning and triggering factors of subsidence.
31
6. Conclusions A Persistent Scatterer Interferometry Technique (PSI), named Stable Point Network (SPN), has been used in order to study the subsidence affecting the Vega Media of the Segura River basin. This technique provides subsidence information of the whole area from January 2004 to December 2008 with maximum settlements up to -12 cm. Settlements are mainly concentrated in the flood plain of the valley where younger deposits are placed. The borders of the valley that comprise older deposits do not suffer any significant deformation. The relationship between the piezometric level changes and subsidence has been demonstrated comparing both temporal series of data for several locations. Relationship among subsidence and soft soil thickness has been demonstrated showing that deformations are higher where soft soil is thicker. Subsidence has been demonstrated to be higher near the Segura River than far from it. Subsidence is probably due to the existence of normally consolidated young deformable soils and the more intensive water withdrawn from wells. Another factor directly related to the magnitude of subsidence is the proximity of the pumping wells, which in fact are mainly located near the river. Subsidence rates are higher near the pumping wells and decrease with the distance, describing a depletion curve (subsidence bowl). Spatial distribution of DInSAR data demonstrate that subsidence rates are lower inside the old Arab city than outside from it. This is probably due to previous soil overconsolidation produced by the weight of pre-existing buildings. A very good correlation was shown between measured subsidence using PSI and subsidence calculated with a predictive model. Finally, subsidence has been compared with basement buildings distribution, observing that there is no spatial correlation. In this case, PSI-DInSAR derived information has allowed studying subsidence phenomena affecting the Vega Media of the Segura River and its relationships with other triggering and conditioning factors. These results have improved the knowledge about the mechanisms and factors that govern subsidence in Murcia City. Therefore these results may be used by decision makers to improve territorial management and planning. Acknowledgements
The European Space Agency (ESA) Terrafirma project has funded all the SAR data processing with the SPN technique. Additionally, this work has been partially financed by the Spanish Geological and Mining Institute (IGME) with the collaboration of the Regional Government of Murcia and the universities of
32
Alicante (UA). This work has been also supported by the Spanish Ministry of Science and Research (MICINN) and EU FEDER under project TEC2008-06764-C02-02 and by the Generalitat Valenciana (ACOMP/2010/082) The Cartographical Service of Murcia (CARTOMUR) has provided DEM data used in this work.
REFERENCES Aragón, R., García-Aróstegui, J.L., Lambán, J., Hornero, J., Fernández-Grillo, A.I. (2004). Impacto de la explotación intensiva de aguas subterráneas en la ciudad de Murcia (España). Análisis hidrogeológico. Proc. XXXIII Congress of IAH-ALHSUD, Conference on Groundwater Flow Understanding from local to regional scales, Zacatecas, Mexico, pp. 2622-2624.
Arnaud, A., Adam, N., Hanssen, R., Inglada, J., Duro, J., Closa, J., Eineder, M. (2003). ASAR ERS interferometric phase continuity. Proc. IEEE International Geoscience and Remote Sensing Symposium (IGARSS-03), 21-25 July, Toulouse, France, 2, 1133- 1135.
Berardino, P., Fornaro, G., Lanari, R., Sansosti, E. (2002). A new algorithm for surface deformation monitoring based on small baseline differential SAR interferograms, IEEE Transactions on Geoscience and Remote Sensing 40, 2375-2383.
Blanco, P., Mallorquí, J.J., Duque, S., Monells, D. (2008). The Coherent Pixels Technique (CPT): an Advanced DInSAR Technique for Non-linear Deformation Monitoring, Pure and Applied Geophysics 165, 1167-1193.
Casu, F., Manzo, M., Lanari, R. (2006). A Quantitative Assessment of the SBAS Algorithm Performance for Surface Deformation Retrieval from DInSAR Data. Remote Sensing of the Environment 102, 195-210
Cerón, J.C., Pulido, A. (1996). Groundwater problems resulting from CO2 pollution and overexploitation in Alto Guadalentín aquifer (Murcia, Spain). Env. Geol., 28, 223-228.
33
Chen, C., Hu, J., Lu, C., Lee, J., Chan, Y. (2007). Thirty-year land elevation change from subsidence to uplift following the termination of groundwater pumping and its geological implications in the Metropolitan Taipei Basin, Nothern Taiwan. Engineering Geology 95, 30-47.
CHS (2007). Nuevas aportaciones al conocimiento hidrogeológico del entorno urbano de Murcia. 363 pp. Technical report. Available in: www.chs.es
De Justo, J.L., Vázquez, N.J., De Justo, E. (2002). Rozamiento negativo en pilotes producidos por la subsidencia en Murcia. Ingeniería Civil 125, 95-103.
Duro J., Inglada J., Closa J., Adam N., Arnaud, A. (2005) High resolution differential interferometry using time series of ERS and ENVISAT SAR data. In: FRINGE 2003, Frascati, Italy, 1–5 December.
Duro, J., Inglada, J., Closa, J., Adam, N., Arnaud, A. (2005) High resolution differential interferometry using time series of ERS and ENVISAT SAR data. In: FRINGE 2003, Frascati, Italy, 1–5 December
Ferretti, A., Prati, C., Rocca, F. (2000). Nonlinear subsidence rate estimation using permanent scatterers in differential SAR interferometry. IEEE Trans. Geosci. Remote Sensing, 38, 2202–2212.
Ferretti, A., Prati, C., Rocca, F. (2001). Permanent scatterers in SAR interferometry, IEEE Trans. Geosci. Remote Sensing, 39, 8–20.
Galloway, D.L., Hoffmann, J. (2007). The application of satellite differential SAR interferometry-derived ground displacements in hydrogeology. Hydrogeology Journal 15, 133-154.
Hanssen, R.F. (2001). Radar interferometry. Data interpretation and error analysis. (308 pp.). The Netherlands: Kluwer Academic publisher.
Herrera, G., Tomás, R., Lopez–Sanchez, J.M., Delgado, J.,
Vicente, F.,
Mulas, J., Cooksley, G.,
Sánchez, M., Duro, J., Arnaud, A., Blanco, P., Duque, S., Mallorquí, J.J., Vega–Panizo, R., Monserrat, O. (2009b). Validation and comparison of Advanced Differential Interferometry Techniques: Murcia metropolitan area case study. ISPRS Journal of Photogrammetry and Remote Sensing 64, 501–512.
34
Herrera, G.; Fernández, J. A.; Tomás, R.; Cooksley, G.; Mulas, J. (2009a). Advanced interpretation of subsidence in Murcia (SE Spain) using A-DInSAR data – modelling and validation. Natural Hazards and Earth System Science 9, 647-661.
Hoffman, J. (2003). The application of satellite radar interferometry to the study of land subsidence over developed aquifer systems. PhD Thesis, University of Stanford, 211 pp. Hoffmann, J., Galloway, D.L., Zebker, H.A. (2003). Inverse modeling of interbed storage parameters using land subsidence observations, Antelope Valley, California: Water Resour. Res. 39, 5-10.
Hoffmann, J., Zebker, H.A., Galloway, D.L., Amelung, F. (2001). Seasonal subsidence and rebound in Las Vegas Valley, Nevada, observed by synthetic aperture radar interferometry: Water Resources Research 37, 1551-1566.
Hooper, A., Zebker, H., Segall, P., Kampes, B. (2004). A new method for measuring deformation on volcanoes and other natural terrains using InSAR Persistent Scatterers. Geophys. Res. Lett., 31, DOI:10.1029/2004GL021737.21.
Hu, R.L., Yue, Z.Q., Wang, L.C., Wang, S.J. (2004). Review on current status and challenging issues of land subsidence in China. Eng. Geol., 76, 65-77.
IGME (1986). Calidad de las aguas subterráneas en la Cuenca Baja del Segura y costeras de Alicante. Instituto Geológico y Minero de España, Madrid.
IGME (2000a). Estudio geotécnico para el análisis, prevención y corrección de la patología derivada de los cambios en el subsuelo de la ciudad de Murcia. Instituto geológico y Minero de España y Consejería de Obras Públicas y Ordenación territorial de la Región de Murcia. 208 pp. Internal report.
IGME (2000b). Estudio geotécnico del subsuelo del área metropolitana de Murcia. Instituto geológico y Minero de España y Consejería de Obras Públicas y Ordenación territorial de la Región de Murcia. 120 pp. Internal report.
35
IGME, 2001. Seguimiento y control instrumental de la subsidencia del terreno en Murcia: Fase I, Instituto Geológico y Minero de España, 120 pp. (In spanish).
Lanari, R., Mora, O., Manunta, M., Mallorquí, J.J., Berardino, P., Sansosti E. (2004). A small Baseline DIFSAR Approach for Investigating Deformations on Full Resolution SAR Interferograms, IEEE Transactions on Geoscience and Remote Sensing, 7, 1377-1386.
Manunta, M., Marsella, M., Zeni, G., Sciotti, M., Alzori, S., Lanari, R. (2008). Two-scale surface deformation analysis using the SBAS-DInSAR technique: a case study of the city of Rome, Italy. Remote Sens. Environ. 29, 1665-1684.
Martínez, M., Mulas, J., Herrera, G., Aragón, R. (2004). Efectos de una subsidencia moderada por extracción de agua subterránea en Murcia, España. Proc. XXXIII Congress of IAH-ALHSUD, Zacatecas, Mexico. Conference on Groundwater Flow Understanding from local to regional scales, CD ROM.
Montenat, C. (1977). Les basins néogènes et quaternaries du Levant d’Alicante à Murcie (Cordillères Bétiques orientales, Espagne). Stratigraphie, paléontologie et evolution dynamique. Doc. Lab. Geol., Univ. Lyon, 69, 345 pp.
Mora, O., Mallorquí, J.J., Broquetas, A. (2003). Linear and nonlinear terrain deformation maps from a reduced set of interferometric SAR images. IEEE T. Geosci. Remote 41, 2243-2253.
Mulas, J., Aragón, R., Martínez, M., Lambán, J., García-Arostegui, J.L., Fernández-Grillo, A.I., Hornero, J., Rodríguez, J., Rodríguez, J.M. (2003). Geotechnical and hydrological analysis of land subsidence in Murcia (Spain). Proc. 1ª International Conference on Groundwater in Geological Engineering, 22-26 September, Bled, Slovenia, 50, 249-252.
Raucoules, D., Bourgine, B., de Michele, M. , Le Cozannet, G., Closset, L., Bremmer, C., Veldkamp, H., Tragheim, D., Bateson, L., Crosetto, M., Agudo, M., Engdahl, M. (2009). Validation and intercomparison of Persistent Scatterers Interferometry: PSIC4 project results. Journal of Applied Geophysics 68, 335-347
36
Rodríguez Jurado, J., Martínez Corbella, M., Mulas, J., Rodríguez Ortiz, J.M. (2000). Establecimiento de un modelo geológico para el estudio de la subsidencia por rebajamiento del nivel freático. Geotemas, 1, 155-158.
Rodríguez Ortiz, J.M., Mulas, J. (2002). Subsidencia generalizada en la ciudad de Murcia (España). In Ayala Carcedo, J. & Olcina Cantos Coords., Riesgos Naturales (pp. 459-463). Barcelona: Editorial Ariel.
Stramondo S., Saroli M., Tolomei C., Moro M., Doumaz F., Pesci A., Loddo F., Baldi P., Boschi E (2007). Surface movements in Bologna (Po Plain -Italy) detected by multitemporal DInSAR. Remote sensing of Environment 110, 304-316.
Strozzi T., Wegmüller, U., Tosi, L., Bitelli, G., Spreckels, V. (2001) Land Subsidence Monitoring with Differential SAR Interferometry. PE&RS, Photogrammetric Engineering & Remote Sensing 67, 12611270.
Tomás, R. , 2009. Estudio de la subsidencia de la ciudad de Murcia mediante interferometría SAR diferencial avanzada. Ph. D Thesis. In Spanish. Unpublished.
Tomás, R. Herrera, G. Delgado, J. Lopez-Sanchez, J.M. Mallorquí, J.J. Mulas, J. (2010). A ground subsidence study based on DInSAR data: calibration of soil parameters and subsidence prediction in Murcia City (Spain). Engineering Geology 111, 19-30.
Tomás, R., Delgado, J., López-Sánchez, J.M. (2005a). Técnicas de ingeniería cartográfica empleadas en el estudio de subsidencia y movimientos de ladera: principales características y análisis comparativo. Proc. XV ADM – XVII INGEGRAF, Sevilla, 1-3 junio, CD-ROM
Tomás, R., Lopez–Sanchez, J.M., Delgado, J., Mallorquí, J.J., Herrera G. (2008). DInSAR monitoring of aquifer compaction due to water withdrawal: Vegas Baja and Media of the Segura river (SE, Spain) case study. In: J.M. Sánchez Ed.: Droughts: Causes, Effects and Predictions, Nova Publishers, New York, 253–276.
37
Tomás, R., Márquez, Y., Lopez-Sanchez, J.M., Delgado, J., Blanco, P., Mallorqui, J.J., Martinez, M., Herrera, G., Mulas, J. (2005b). Mapping ground subsidence induced by aquifer overexploitation using advanced Differential SAR Interferometry: Vega Media of the Segura River (SE Spain) case study. Remote Sensing of the Environment 98, 269 – 283.
Vázquez, J.N., De Justo, J.L. (2002). La subsidencia en Murcia. Implicaciones y consecuencias en la edificación. Ed. COPOT, Murcia. 262 pp.
Werner, C., Wegmüller, U., Strozzi, T., Wiesmann, A. (2003). Interferometric point target analysis for deformation mapping. Proc. IEEE International Geoscience and Remote Sensing Symposium (IGARSS03), 21-25 July, Toulouse, France 7, 4362-4364.
38
