This work is focused on the Almonte-Marismas aquifer located within Doñana Natural Space (SW Spain); this aquifer is threatened by droughts, irrigation-driven groundwater overexploitation, urban use, and the potential reactivation of gas extraction and storage projects. We present ground deformation measurements from Sentinel-1 Interferometric Synthetic Aperture Radar (InSAR) data ranging from 1 to 2.5 cm, covering ∼2,500 km2 from 2014 to 2020. Detecting spatially distributed ground deformation over this agricultural area is challenging due to the low-coherence radar signals; consequently, the ground movement results are on the same order of magnitude as the errors associated with InSAR data. We have approached this issue by considering auxiliary information such as groundwater levels, climatic time series, and pumping rates and analyzing their spatio-temporal connections to ground displacements. We correlate InSAR and hydrogeological information through wavelet analysis, developing a Python package that allows applying the same methodology in other similar study areas. For the first time in the Doñana area, a significant relationship between distances to pumping wells and the displacement extent is detected. Moreover, other subsidence-related triggering factors are identified, such as the soil moisture balance, clay shrinking-swelling processes, and creep of geological formations. These results are highly valuable to support aquifer management decisions in the Doñana Natural Space; in this border region, three groundwater bodies were officially declared overexploited in 2020. Our findings provide a ground motion baseline assessment to help differentiate historical variations from any future anthropogenic effect in this complex marsh land ecosystem.
Integrated Interferometric Synthetic Aperture Radar and hydrogeological analysis has shown displacements up to 2.5 cm/year on the aquifer that hosts the Doñana Natural Space
Wavelet method with a new Python package allows for integrated displacement, groundwater level, climatic and pumping time series analyses
The results provide a ground motion baseline to differentiate natural variations from future anthropogenic effects in the Doñana area
Plain Language Summary
This work is focused on Doñana Natural Space (SW Spain), which integrates the Doñana National Park and the Doñana Natural Park in a single space. Doñana has acquired the highest protected area categories that a natural area can receive from international conservation organizations. However, Doñana's ability to support biodiversity is under constant threat due to droughts, groundwater overexploitation for irrigation, urban use, and the reactivation of a gas extraction and storage project. This study presents ground deformation measurements in the Doñana area derived from Earth observation data ranging from 1 to 2.5 cm from 2014 to 2020. The work relates ground displacement, groundwater and climatic information through mathematical tools. For the first time in the Doñana area, a significant relationship between distances to pumping wells and the displacement extent is detected. These results are highly valuable to support aquifer management decisions in the Doñana Natural Space. Our findings provide a ground motion baseline assessment that will help differentiate historical variations from any future anthropogenic effect in this complex marsh land ecosystem.
Groundwater is a fundamental resource for ecosystems, population supply and agriculture (Taylor, 2014; Zektser & Everett, 2004). Aquifers, characterized by having very large water storage capacities in relation to their average annual resources, are less vulnerable to the effects of droughts, and their exploitation can be maintained during dry periods (Shah, 2007). However, not all groundwater is a renewable resource, and aquifers are being exploited at a higher rate than they can be recharged. The urban and agricultural water demands have vastly increased, while climate change is affecting the timing and distribution of this water (Famiglietti, 2014; Taylor, 2014; UNESCO, 2022; Wada & Bierkens, 2014).
Two of the most significant risks of aquifer overexploitation are land subsidence and permanent groundwater storage loss caused by mechanical hysteresis (Béjar-Pizarro et al., 2017; Chen et al., 2022; Ezquerro et al., 2014; Pacheco-Martínez et al., 2013; Raju et al., 2022; Teatini et al., 2006). The compaction of fine-grained aquifer materials results in land subsidence when pumping exceeds the recharge rate, and this subsidence can be permanent if the lowering of the groundwater increases the effective stress above the preconsolidation value (Galloway et al., 1998). Subsidence due to groundwater depletion is a slow and gradual process that develops on large time scales (months to years), producing progressive land elevation lowering (centimeters to decimeters per year) over very large areas (tens to thousands of square kilometers) and variably impacting urban and agricultural areas worldwide (Herrera-García et al., 2021).
Interferometric synthetic aperture radar (InSAR) is a cost-efficient Earth observation technique used to monitor land surface deformation with a high measurement point density (Castellazzi, Martel, Galloway, et al., 2016; Tomás & Li, 2017). Multitemporal InSAR (MT-InSAR) techniques allow the measurement of land movements with an accuracy of a few mm/year, similar to geodetic devices such as leveling and global positioning system (GPS) instruments (Béjar-Pizarro et al., 2016; Chen et al., 2020; Ezquerro et al., 2020; Shamshiri et al., 2018). However, the density of MT-InSAR measurements depends on several factors, including the processing approach, the temporal stability of the scattering properties of the ground surface, the number of available data, the time interval between the acquisitions, the baseline configuration or the characteristics of the deformation signal (e.g., Hooper et al., 2012; Morishita & Hanssen, 2015). Higher densities usually are achieved in terrains with temporally stable scattering properties (e.g., urban environments) and lower densities are generally obtained in terrains with varying scattering properties (e.g., agricultural areas).
MT-InSAR data have proven to be useful for studying the relationship between ground surface displacement and groundwater withdrawals (Galloway & Burbey, 2011; Mahmoodinasab & Mohseni, 2021) in many locations around the world. For example, in big cities or extensive regions like the plains and basins of Iran (Motagh et al., 2008), China (Yang et al., 2014) or California (Borchers & Carpenter, 2014; Riel et al., 2018). In California, one of the most productive agricultural areas in the world (The Central Valley), InSAR records reveals that land subsidence has reached values of 10 m, affecting an area of up to 13,500 km2 (Faunt et al., 2016; Neely et al., 2021). Among the case studies performed over big cities, stand out those of Mexico city (Chaussard et al., 2014), New Orleans (Jones et al., 2016), Tehran (Mahmoudpour et al., 2013), Beijing (Chen et al., 2016), Madrid (Ezquerro et al., 2014) or Bangkok (Aobpaet et al., 2013). In all these studies, MT-InSAR data have allowed the assessment of land subsidence and the identification of groundwater pumping as its main triggering factor. In addition, the inclusion of InSAR data in hydrogeological models also allows to have a better knowledge about aquifer heterogeneities, hydraulic conductivity values or structural boundaries of aquifers (Guzy, A. & Malinowska, 2020).
The Sentinel-1 mission, from the Copernicus European Union's Earth Observation Programme, comprises a constellation of two polar-orbiting SAR satellites that provide an enhanced revisit frequency (every 6 or 12 days, depending on the satellites available), medium spatial resolution (20 m) and worldwide coverage under a free, full, and open data policy. Sentinel-1 data are highly suitable for the continuous and broad-scale monitoring of ground displacements. In addition, beginning in 2022, the Copernicus European Ground Motion Service (EGMS) provides an updated annual ground displacement map for all of Europe derived from time-series analyses of Sentinel-1 data. To handle and process the large and constantly increasing Sentinel-1 archive, the Geohazards thematic exploitation platform (GEP-TEP) web tool was launched in 2015 by the European Space Agency (ESA). In this online processing service, SAR images and Advanced Differential Interferometric Synthetic Aperture Radar (A-DInSAR) algorithms are located together in a user-friendly interface. The processing chains run automatically in the server with very little user interaction, and the time processing requirements are relatively fast. The results derived from the GEP MT-InSAR algorithms (named Thematic Applications), such as the Parallel Small Baseline Subset (P-SBAS) tool, can be found in recent publications (Cigna and Tapete, 2021a, 2021b; Ikuemonisan & Ozebo, 2020). These contributions are helping to validate GEP results derived with other more refined and user-guided tools (Cigna et al., 2021).
Wavelet transforms are widely used in natural science (Labat, 2005). In hydrology they have been applied to identify intermittent periodicities, cross-correlation analysis, de-noising, or simulation and prediction of hydrological time series (Sang, 2013). Moreover, the continuous wavelet transform (CWT) of bivariate time series has enabled the comparison of the time series of groundwater levels and ground displacements through the cross wavelet transform (XWT) to identify potential physical relationships between them (Bonì et al., 2020), or the inference of cause-effect relationships in groundwater pumping processes (Gao et al., 2018). CWT is an alternative method to the Fourier transform that facilitates the analysis of time series in the time-frequency domain, thus allowing the identification of intermittent periodicities (Grinsted et al., 2004) and time delays between groundwater and displacement time series (Tomás et al., 2020).
The study area considered in this work is the Doñana area (SW Spain), a UNESCO World Heritage property, a Biosphere Reserve and a Ramsar Site. Approximately 40% of all the water that infiltrates into the aquifer units around Doñana (predominantly from rain) is used for human activities through groundwater abstraction (UNESCO et al., 2021). The abstraction volume currently exceeds sustainable limits, stressing the aquifer systems. For this reason, in 2020, the Spanish Authorities declared three of the five groundwater bodies overexploited (i.e., the management unit under the Water Framework Directive) in the Doñana area. This has raised fears that the Outstanding Universal Value of Doñana is at risk. Furthermore, the Doñana region is one of the few areas in southern Spain that would be geologically suited for gas storage, and it has been subsequently designated as a strategic gas storage location by the Spanish government. In 2020, a UNESCO/IUCN/Ramsar mission alerted that the impacts of gas storage projects are not adequately assessed (UNESCO et al., 2021) and recommended systematic monitoring of the possible impacts derived from gas mining activities. The compound challenges now faced by water planners require a new generation of more efficient monitoring tools for aquifer management, that the changes in aquifer storage be addressed and that the responses of groundwater system to various stresses (natural vs. anthropogenic pressures) be analyzed.
This work aims to detect potential land displacements in the Doñana area and to identify the triggering mechanism by analyzing Sentinel-1 InSAR data and groundwater level measurements recorded between 2014 and 2020 through wavelet tools. The results will help decision-makers implement adequate strategies for the sustainable management of the Doñana aquifer, which is threatened by increasing anthropic pressure and even more recurrent droughts. The implemented methodology was compiled in an open source Python package, available at GitHub, that can be reproduced for other aquifers when similar InSAR data and hydrogeological information are available.
2 Study Area
The Doñana Natural Space is located in the south‒southwest region of the Iberian Peninsula (Figure 1) over the Almonte-Marismas aquifer (Salvany & Custodio, 1995; Salvany et al., 2010, 2011). It is an area with gentle slopes and elevations ranging from 190 m above mean sea level (a.m.s.l) in the north to less than 1 m a.m.s.l. in the marsh. It is situated in the lower basins of the Guadalquivir and Tinto Rivers and has a total area of 2,409 km2. The Almonte-Marismas aquifer is connected laterally with the aquifer of El Aljarafe in the northeast. This area borders the Guadalquivir River to the east, the Atlantic Ocean to the southwest and the impermeable materials of the Sierra Morena and the Tinto River to the north (Figure 1; Salvany et al., 2011).
The materials that constitute the aquifers were deposited between the Miocene and the Quaternary, forming a monoclinal series that dips gently to the south‒southwest (Salvany & Custodio, 1995). From oldest to most modern, the series begins with the Miocene blue marls, which form the substrate of the aquifer on which the Pliocene silt-sandy units are deposited, with thicknesses between 10 and 150 m, growing to the south, with overlying aeolian mantle and ancient dunes. Above these layers lie Pleistocene alluvial units; these units do not form outcrops and comprise gravels, sands and clay lentils. On top of the Pleistocene units are the Holocene marsh materials, formed by clays that fill the mouth of the Guadalquivir River with thicknesses reaching 200 m. Finally, the most recent materials are the alluvial deposits of rivers and streams and the sands of today's coastal beaches and dunes. The latter, formed by siliceous sands extended parallel to the coast of the Atlantic Ocean (Figure 1b), originate voluminous mobile dunes with thicknesses of up to 150 m.
Two hydrogeological behaviors are distinguished in the Almonte-Marismas aquifer system. There are unconfined parts of the aquifer, with a thickness varying from 15 m in the north to 150 m of the Eolic mantle in the Matalascañas area. In the southeast area, under the marshlands-silt and salt unit, the aquifer presents a confined behavior and has a thickness ranging from 50 to more than 250 m. There are also other confined areas under the clays that are underneath thick layers of sands closer to the coast or to La Rocina stream. Several upwelling levels in those areas corroborate this behavior (Custodio et al., 2008). A semiconfined transition zone separates the confined and unconfined behaviors (Custodio et al., 2008). Detailed hydrogeological studies show the presence of numerous clay intercalations and lentils, which, in some specific areas, transform the aquifer into a multilayer model (Figure 1c; Salvany & Custodio, 1995).
The overall groundwater flow features N‒S or NW‒SE directions. Recharge is mainly produced by infiltration of rainfall, with an annual average of 5.8 × 105 m3/d (Guardiola-Albert et al., 2009). The main discharge takes place by the drainage of surface watercourses (streams and rivers), outflows to the sea, discharges at the boundaries between aeolian sands and marshlands, flow exchanges with permanent and temporary lagoons and, to a remarkable extent, the extraction of water through wells for agricultural, industrial, tourism, and human consumption.
3 Materials and Methods
3.1 Data Sources
3.1.1 InSAR Data
We processed InSAR data from the ESA Sentinel-1A satellite (12 days revisit time in a single pass) to estimate line-of-sight (LOS) ground motion in the 2014–2020 period. The SAR data consist of Interferometric Wide Single Look Complex (IW SLC) images from ascending orbit 147 and descending orbit 154. These images have been processed using the P-SBAS approach (Manunta et al., 2019), which is the parallel computing solution for the SBAS processing chain (Berardino et al., 2002) used to deal with the current scenario characterized by vast SAR archives (Casu et al., 2014). The P-SBAS was implemented at the ESA Geohazards Exploitation Platform (GEP) by the Institute for Electromagnetic Sensing of Environment (CNR-IREA) (De Luca et al., 2015). The P-SBAS processing chain includes methods for the phase unwrapping mitigation and for the atmospheric phase screen removal (De Luca et al., 2019; Lanari et al., 2020). The main limitation of this tool is that it is unsupervised, meaning that the process runs automatically and there is limited user interaction. Nevertheless, the processing pipeline automation, robustness and results validation has been assessed in several works (Cigna et al., 2021b; Manunta et al., 2019).
The P-SBAS results cover an area of approximately 10,000 km2 for the descending data sets and 23,000 km2 for the ascending datasets. However, we have only analyzed the MPs located within the Almonte-Marismas aquifer system area. The temporal coherence threshold applied to select coherent pixels is 0.85. The spatial resolution is ∼90 m (multilook applied: 20 in range × 5 in azimuth). The main results are terrain deformation mean LOS velocity maps and LOS displacement time series derived for each measurement point (MP).
For each satellite direction (ascending and descending), one set of data was obtained, covering two different periods: from October 2014 to January 2020 for the descending trajectory and from January 2017 to January 2020 for the ascending. The InSAR-derived deformation rates obtained for each data set are shown in Figure 2. An initial reference point was established at the port of Mazagón, which is the nearest global navigation satellite system (GNSS) station in the study area. Then P-SBAS algorithm searched for stable points near this initial one. Final reference points were all located at the Mazagón village. The mean LOS velocity standard deviation and total number of MPs in each data set are shown in Table 1.
|Trajectory_period||Satellite||Track||Frames||Number S-1 dates||Standard deviation||Number MPs AOI||Number interferograms|
- Note. Columns show satellite trajectory, frames, number of S-1 dates, standard deviation of all MP velocity values, number of MPs in the area of interest (AOI) and number of interferograms for each InSAR data set.
InSAR has the potential to detect ground surface movement with displacement accuracies up to the millimeter level (Adam et al., 2009; Lanari et al., 2007) in the best-case scenarios. However, there are several factors that can affect the surface deformation accuracy, such as unwrapping errors and atmospheric artifacts (Hanssen, 2001). Standard deviation values are between 0.2 and 0.4 cm/year in all the presented data sets. To keep as much data as possible in this complex area, the standard deviation is used as the stability threshold, and all displacements below 0.4 cm/year are considered noise. It is not possible to analyze these data points without access to detailed auxiliary data, such as pumping rates or piezometric levels.
The criterion used to select MPs, which can be set based on exploiting either the amplitude or the phase components of the SAR acquisitions, is also relevant for the accuracy and MP density. The Doñana region is a complex mosaic of shrub, forest, wetland and agricultural areas, alongside rice crops in many cases, crossed by human structures and infrastructures. These characteristics make it difficult to obtain MPs from anthropic surfaces even when selecting pixels by coherence-based MT-DInSAR methods. The lack of continuity in the MPs reduces the redundancy in the displacement calculation, thereby increasing the probability of error accumulation and translation over low-connectivity areas. Moreover, some studies assessing short temporal baseline multilook interferograms have reveal a systematic noise signal (named the fading effects) (Ansari et al., 2020) that could be related to physical sources such as changes in soil moisture (De Zan et al., 2015; Zwieback et al., 2016). To overcome this limitation, we applied the SBAS technique, which has been demonstrated to guarantee fully consistent solutions (Lanari et al., 2020). In addition, we have selected a restrictive coherence threshold (0.85) to ensure the quality of the selected pixels at expenses of point density measurement loss. Furthermore, as explained in the methodology section, we have imposed strict criteria to select the InSAR MPs and considered the displacement uncertainties when discussing the triggering factors associated with land subsidence.
3.1.2 Hydrological and Hydrogeological Information
Groundwater level information was provided by the piezometric networks of the Geological and Mining Institute of Spain and the Guadalquivir Hydrographic Confederation, which have a good spatio-temporal distribution since 1994. We considered piezometers with more than 60 measurements in the 2010–2020 time span, partly coinciding with InSAR data. This resulted in 242 piezometers (Figure 3). As will be explained in the methodology section, for the wavelet analysis, it is required that groundwater level sampling series are regularly spaced in time; hence, monthly groundwater level data were obtained using an arithmetic mean. Additionally, to fill in the data gaps, we averaged time interpolations using cubic splines and a random forest method (R package MissForest, Stekhoven, 2022). Although the InSAR data analyzed in the present work represent measurements obtained after 2014, we considered all the records obtained since 1994 to increase the amount of data used in the groundwater level interpolations. This year was chosen because since then, time series have been more densely measured. Additionally, 25 continuous daily monitoring groundwater level sensors from the Mining and Geological Institute of Spain were used to obtain a 12-day period database with the aid of polynomial interpolation (Figure 3). The values of each sensor were upscaled into 12-day time series since this interval is the measurement interval of the InSAR deformation series. In this way, the measurement dates of the deformation series are matched with those of the piezometric series of the sensors for their subsequent joint analysis using wavelet tools.
Likewise, other variables that could explain the observed ground displacement have also been compiled, including time series of precipitation, evapotranspiration (ETP) and aquifer extractions that occurred in the surroundings of the Matalascañas tourist resort (Figure 1a). Daily evapotranspiration and precipitation records were obtained from the agroclimatic stations belonging to the Junta de Andalucía (https://www.juntadeandalucia.es/agriculturaypesca/ifapa/riaweb/web/, Figure 3). Both the original evapotranspiration values and the cumulative deviation from the historic mean were studied. For visual purposes only the cumulative deviation from the mean was represented for precipitation. The groundwater extraction rate data compiled in Matalascañas consist of monthly values for the 2014–2020 period, generated based on recorded pumping rates during the 2007–2010 period and assuming no significant groundwater consumption changes in this area for the most recent years.
Also a map with the location of temporary ponds was included in the study (Figure 3). This map was based on the first LiDAR Coberture Map of the Geographic National Institute (IGN; Mapa LiDAR cob1 2020 CC-BY 4.0 scne.es) by looking for pixels of ponds color (i.e., blue value between 222 and 234) and discharging areas smaller than 0.01 ha.
A scheme of the methodological approach is shown in Figure 4. MPs were aggregated into polygons using the Aggregate Points tool from ArcMap 10.7 software (ESRI, 2011). Each polygonal entity was generated by grouping the nearest MPs. This tool was applied to the descending InSAR data set covering the 2014–2020 period, which is the period that presents the highest amount of data and best-distributed MPs. The geometries of these aggregates were slightly modified in the other InSAR data sets to accommodate the maximum number of nearby points without altering the location or identification of each of the polygons.
Mean velocity deformation higher than the standard deviation range.
At least 3 MPs with velocities higher than the standard deviation, with the same direction of displacement and within a distance buffer of 200 m.
Located over expected highly coherent terrain patches (e.g., an urban fabric).
Following these criteria, 40 aggregated polygons were studied (Figure 3). Once the aggregates were formed a polygonal shapefile was obtained. Then, the average deformation of each aggregate was studied in combination with the climatic variables with the newly developed Python package WaSAR.
3.2.1 Integrated Analysis Compiled in a Python Package: WaSAR
WaSAR is an open source Python package built from the developed methodology. It allows a combined spatio-temporal study between MT-DinSAR time series and climatic variables such as piezometry, precipitation or evapotranspiration. The code is formed by several scripts and based on an object-oriented design, which makes it more reusable and readable. The main class of the code is Model, which hosts the aggregates spatial layer. MT-InSAR datasets and climate databases (spatial or not) need to be defined and associated to this object. Once the Model object has been created, it is possible to study the average deformation of any aggregate together with the evolution of the groundwater levels in the vicinity of the polygon, as well as other variables of interest (Figure 4). Further wavelet analyses have also been implemented in WaSAR through the R WaveletComp package (Roesch & Schmidbauer, 2018). The use of this package, and not others available in MATLAB (Grinsted et al., 2004) or Python (Gregory et al., 2019), was motivated by its possibility of being executed on a free platform and by having the XWT and WTC tools included in a robust way. Details and explanations about wavelet theory, the tool used to apply it, and how to interpret the corresponding results are given in Section 3.2.2.
WaSAR builds mainly on GeoPandas (Jordahl et al., 2021) and Pandas (Reback et al., 2020) for the spatio-temporal time series analysis and data storage, and make use of Matplotlib (Hunter, 2007) and Folium (Filipe et al., 2021) for the data and spatial visualization, respectively. The use of Folium package allows the user an interactive visualization of any of the model variables, and the possibility of representing them in different background layers, such as the satellite image or Google maps layers.
The package contains a minimal reproducible model to familiarize the user with all the WaSAR functionalities, with real data from public datasets. The code and several Jupyter Notebook tutorials are freely available under a General Public License v3.0 (GPL) at https://github.com/MiguelonGonzalez/wasar (González-Jiménez & Guardiola-Albert, 2022), and is listed in the Python Packaging Index (PyPI). The inclusion of WaSAR in the ecosystem of open source programs in Python wills to incentive the collaboration with other users to improve and update the software.
All these above characteristics, such as the object-oriented design, the availability of the code at open repositories like GitHub or Zenodo, or code readability tend to follow Hutton et al. (2016) guidelines to help move toward reproducibility hydrology.
3.2.2 Wavelet Transforms
The CWT is a data-processing method that enables the analysis of time series in the time-frequency domain, allowing the identification of intermittent periodicities in nonstationary processes (Grinsted et al., 2004). These results are depicted as a two-axis plot defined by the time (X-axis) and the period (i.e., inverse of the frequency) of the time patterns (Y-axis), in which the areas exhibiting high values indicate the presence of significant time patterns (i.e., significant periodicity) at particular dates (Bonì et al., 2020; Tomás et al., 2020). The color scale represents the squared root of the wavelet power values to accentuate the contrast of the image. A white contour designates the 10% significance level that encloses the area in which periodicity is significant against a white noise process fitted to the data. Finally, a cone of influence is plotted that excludes edge-effect areas by showing a relatively light shadow.
WTC is defined as the coherence between two CWTs and is calculated as the smoothed normalized XWT. WTC shows the correlation within a local phase, showing statistical significance only in areas where the series involved actually share significant periods.
To successfully implement CWT, the time series input parameter needs to be regularly spaced in time (Tomás et al., 2016). Sentinel-1 satellites have a revisit time of 12 days, making CWT a suitable tool for analyzing Sentinel-1 P-SBAS-InSAR-derived vertical deformation time series without the need for any further transformation. As explained in Section 3.1.2, the groundwater level change time series were interpolated to make them temporally consistent with the InSAR time series.
Grinsted et al. (2004) stated that cross wavelet analysis is a powerful method for testing proposed connections between two time series, but that the outcome of the analysis has to consider a linking mechanism. They advised to verify cross wavelet results with the physical process being analyzed. In this way, Tomás et al. (2016, 2020) and Bonì et al. (2020) verified their cross wavelet analysis by proposing a conceptual model for the possible mechanisms of landslides or subsidence physical processes. In the present study, the same procedure has been followed. First, the physical relations between groundwater drawdown and land displacements were identified. Then, possible temporal correlation between surficial displacement data obtained from Sentinel-1 satellite images and piezometry was analyzed using the wavelet tools CWT, XWT, and WTC in the R package WaveletComp (Roesch & Schmidbauer, 2018).
Wavelet analysis allows for the analysis and decomposition of signals or data into different frequency components, specifically different seasonal and noise components of the signals. Additionally, and to facilitate the visual comparison of the InSAR time series with other variables, we isolated the underlying long-term behavior of the signal by extracting the trend of the vertical displacement time series and detrending it. The vertical component was computed through the projection of the LOS unit vector, while the detrending of the trend was performed using the Python package statsmodels (Seabold & Perktold, 2010).
4 Results and Discussions
The SBAS approach (Berardino et al., 2002) proposes a complete MT-DInSAR procedure that uses a coherence-based selection criterion to choose the points at which deformation will be measured. These MPs correspond to averaged or multilooked terrain patches whose scatter properties are not altered with time. Therefore, the SBAS is better suited for natural terrains than the persistent scatterer interferometry (PSI) approach (Ferretti et al., 2000). The latter works at full resolution and performs better in urban areas where anthropogenic structures act as nonfluctuating scatters. However, the presence of vegetation or agricultural fields with rapid coherence loss, such as those in the studied area, prevents MP detection. Hence, the results in the study area are sparsely distributed, with greater concentrations of measurements in urban and suburban areas than in other regions (Figure 2). In this section, we discuss the results obtained in four representative areas of Doñana Natural Space with different hydrological, geological, and environmental conditions.
4.1 Isla Mayor Zone
The Isla Mayor area corresponds to the northeast part of the Almonte-Marismas aquifer system (Figure 3) and belongs to an administrative groundwater body named Marismas, one of the three groundwater bodies within the Doñana area that were officially declared overexploited in 2020. Rice agriculture surrounds the Isla Mayor village, covering the western part of the Isla Mayor zone (Figure 5b). In the eastern part, citrus, rice, and other crops are grown over the largest piezometric depression cone of the Almonte-Marismas aquifer system (Figure 5b). As established in the methodology (Section 3.2), MPs were aggregated into 13 polygons in this Isla Mayor subarea of study.
Aggregates 1 and 2 completely cover the urban area of Isla Mayor village (Figure 5d). The averaged InSAR displacement time series shows a clear descending trend in both descending data sets in aggregate 1, with an accumulated value of almost 3 cm in the vertical direction (Figure 6a). However, negligible displacement is seen in the ascending data sets over the 2017–2020 period (Figure 2b). P-SBAS differences for the ascending and descending trajectories could be attributed to horizontal movements, spatial and temporal decorrelations, signal delays due to atmospheric artifacts, orbital or topographic errors, and other systematic errors in the processing procedure. These differences in SBAS results were also found and analyzed by Pawluszek-Filipiak and Borkowski (2020). Aggregates 1, 2, 3, 4, and 26 were located over a clay layer 50-m thick (Figure 1c) and were thus prone to subsidence effects. However, the piezometric levels of the nearest measurement point from aggregates 1 and 2 (11418096, 4.5 km away, Figure 5a) did not show any declining trend in the studied period (Figure 6b).
In 2017, a drawdown groundwater depression up to 15–20 m was reported in the Guadalquivir Hydrographic Confederation (Figure 5) approximately 10 km away from aggregates 1 and 2. This groundwater drawdown was caused by nearby pumps reducing the pore fluid pressure, subsequently increasing the effective stress and stimulating the soil skeleton to compact (Galloway & Hoffmann, 2007). The presence of thick Pleistocene clay layers (Figure 1c), which may be up to two orders of magnitude more compressible than sand (Chilingar & Knight, 1960), can explain land subsidence in this part of the Almonte-Marismas freshwater system. Analogous effects have been identified by Rafiee et al. (2022), who related significant subsidence to a high thickness of fine-grained sediments; Mahmoodinasab and Mohseni (2021), who found a significant relationship between the distances to pumping wells and the displacement extent; and Raju et al. (2022), who determined that the groundwater deficit explained the spatial extent of land subsidence in their monitoring area. As stated by Bozzano et al. (2015), the spatial and temporal evolution of the subsidence process related to the large groundwater drawdown are the local geological conditions (i.e., the thickness of the compressible deposits overlying the exploited aquifer) driving the magnitude of the deformation process and the groundwater level variations driving the subsidence triggering timing over the area. Moreover, this can be the explanation of the differences among ascending and descending trajectories in this area (Figure 2). In Isla Mayor the ascending time series shows a stable trend, meanwhile the descending orbits show displacement moving away from the satellite in the LOS (Figure 6a). This indicates that the movement has a strong horizontal component toward West. The origin of the decrease on the pore pressure is located 10 km to the West of aggregates 1 and 2. Along this line, there seems to be a temporal link between the maximum piezometric drawdown, which is always registered in August/September (Figure 6b), and the local subsidence minima, which is difficult to identify by looking at the InSAR time series (Figure 6a) but that can be clearly seen in the detrended InSAR time series (Figure 6b). This minimum in the displacement series varies depending on the year (February 2016, November 2016, December 2017, December 2018 or December 2019). XWT power spectra and WTC can help us estimate this time lag. Figure 6e shows high wavelet coherence for the 1-year period and phase angles of year 2015/2016 of approximately 230°, which, using Equation 1, can be found to correspond to delays of approximately 240 days and agree with the behavior seen in Figure 6b. This time delay can be explained by drainage from clayey beds generating a lag behind the drainage of productive sands and gravels and thus causing delayed (or secondary) land subsidence, which may manifest itself after well shutdown (Gambolati & Teatini, 2021). In future works, relatively long displacement time series taken from other satellites would allow the further study of this secondary consolidation effect. Although aggregates 1 and 2 are 10 km away from the center of the drawdown cone, they are still in the radius of influence (Figure 5b); as they are in the confined area of the aquifer system, mass losses derived from extracted water can spread horizontally to such distances because of low storativity (Castellazzi, Martel, Rivera, et al., 2016).
The deformation signal presents two significant periods of 12 and 18 months, as identified with the red, high-power areas in the CWT power spectra enclosed in the white contour line (Figure 6c). The annual periodicity appears in 26 of the 40 InSAR aggregates studied all over the Almonte-Marismas aquifer. The groundwater levels at 114180096 piezometer also show this annual periodicity, which can be clearly seen just by looking at the raw information (Figures 6b and 6d), but not the 18-month period. As these groundwater levels measure a confined aquifer, these oscillations correspond to the response of the mechanical load on the formations due to the seasonal changes in the soil moisture and storage at the water table (Van der Kamp & Maathuis, 1991). To find evidence for this hypothesis, we compared displacement data with the cumulative deviation from the mean moisture balance in Figure 6a (i.e., precipitation minus areal evapotranspiration taken from the Isla Mayor climatic gauge). This graph shows the seasonal and cumulative moisture losses that could explain the descending ground movement registered in the InSAR data. The good correlation between the cumulative deviation of soil moisture and the descending-processed InSAR results during the study period from 2014 to 2020 is reflected by a determination coefficient R2 of 0.86. Further research on how changes in atmospheric pressure affect the groundwater level in this confined part of the aquifer system should be performed to firmly establish this poroelastic response (Van der Kamp & Schmidt, 2017) and differentiate it from other phenomena that could also explain this volume loss, such as the creep of soft sediments. Creep consolidation (Chigira, 1992), which is generally referred to as secondary consolidation, can contribute to subsidence long after drawdown in pumped aquifers has stabilized, a phenomenon that is traditionally attributed to “hydrodynamic lag” (Kooi & Erkens, 2020).
There is another group of relevant aggregates in the Isla Mayor zone located on the top of the highest piezometric depression cone within the Almonte-Marismas aquifer (Figure 5). These are aggregates 5, 6, 7, and 8. When comparing surface displacement with piezometry during the study period (Figure 7a), we detected a similar behavior to that observed in the Isla Mayor village; therefore, the same triggering factors can be expected for these aggregates. However, aggregates 5, 6, 7, and 8 are not located over a clay layer of tens meters. In this zone sands are the predominant lithology covering the gravel and sand productive aquifer layer (Figure 1c), which are much less compressible than clays. This is the explanation of why these aggregates do not show velocity rates of the same order of magnitude as in aggregates 1 and 2. If we look at a relatively long historic period, we see a drawdown trend of 6 m in this piezometer within more than 20 years (Figure 7b). In this case, the CWT power spectra of the subsidence series do not show the 18-month period that we identified in former aggregates located far from the depression cone (Figure 7c). We can explain the lack of this longer period, as the location of the aggregates is precisely at the same location as the areas of highest groundwater extraction. The direct relation of both groundwater levels and subsidence is corroborated by XWT power spectra in aggregate 8 in 2018, where both signals are in phase or synchronized. However, in former years, the phase differences indicated lags between 270 and 365 days, thus reflecting the coupling of more complex processes. This coupling of 365-day phase can be evidently recognized in the same seasonal time variation of piezometry and InSAR detrended data shown in Figure 7b.
4.2 North Zone
In the North zone, there are fewer agricultural fields than in Isla Mayor. Rociana del Condado village presents a good MP density, clustered around its urban limits (Figure 8a). Aggregated polygon 10 is not located within the urban fabric, but the soil in this area seems to have good coherence. Its InSAR time series registered an accumulated deformation of 2.5–3 cm in the study period (Figure 9a). This aggregate 10 present a probability of error accumulation as there is low continuity in the MPs due to the presence crops. However, it is interesting to analyze InSAR measurements of this aggregate within this potential error framework because it has a piezometer within its limits (Figure 8). The measured piezometry at this point follows the accumulated deviation of precipitation (Figure 9b), showing the control of the precipitation factor on the temporal behavior of the groundwater level; this feature is present in this and other areas of the Almonte-Marismas aquifer system (Naranjo-Fernández et al., 2020; Rodríguez-Rodríguez et al., 2021).
The wavelet analysis performed for aggregate 10 and the nearest piezometer 114150111 is shown in Figures 9c–9f. Figure 9c displays the CWT results for the individual time series of InSAR displacement in aggregate 10, presenting a significant periodicity of 12 months during the whole study period; this periodicity corresponds to the seasonal hydrogeological signal, to the seasonal moisture deficit, or to both factors jointly with the effects of the irrigation practices in the area. In aggregate 10, we also identify smaller nonsignificant periods of 6 months at the end of 2016. These periodicities can be also identified in the detrended InSAR series in Figure 9b. For the groundwater level measurements (Figure 9d), there is also a clear significant periodicity of 12 months, varying to 10 months toward the end of the period, and other significant periodicities corresponding to fluctuations of 3–6 months can be identified during 2018. Similar results have been found in other sedimentary aquifers in central Spain (Bonì et al., 2020), attributing the yearly signal to seasonal natural climate changes in winter-summer cycles. This piezometer is located at the unconfined part of the aquifer, highlighting the relation between the groundwater level and the cumulative deviation from the mean of precipitation data (Figure 9b). Looking at the XWT power spectra (Figure 9e), there is high joint for yearly and half-yearly seasonal variations, although the latter ones do not extend to all the observation period but only to 2016 and the last months of 2018. This fact could be an indication of the relationships between the piezometric level and InSAR displacement time series (Bonì et al., 2020). The wavelet coherence WTC (Figure 9f), which shows statistical significance only in areas where the series involved actually share significant periods, corroborates the analysis results of the yearly and half-yearly common periodicities regarding both the groundwater level and land subsidence time series; the first periodicity is present throughout the entire study period, while the second periodicity is only present in certain months. In any case, the low-connectivity of MPs in aggregate 10 suggests that relationships shown by the wavelet analysis should be taken with care and further analysis should be performed to corroborate the mentioned links.
4.3 Matalascañas Zone
The InSAR-derived ground deformation information in the urban area of Matalascañas was divided into four aggregated polygons numbered 21, 22, 23, and 24 (Figure 10). The results obtained are similar for all aggregates. We took aggregate 23 as an example to discuss the results in this area because it has inside its limits a daily groundwater level recording sensor (Cuartel S41 in Figure 10).
InSAR time series is plotted in Figure 11a, where moderate descending trend can be appreciated with the 2014–2020 descending data set. Local minimum groundwater level and maximum pumping rates are correlated with local minimum detrended InSAR displacements (Figure 11b). Moreover, in Figures 11c and 11d, the InSAR and groundwater level CWT results show similar periodicities of 12 months throughout the whole study period. In Figure 11e, the horizontal arrows point to the right, indicating that the piezometry and InSAR series are completely in phase. However, in Figure 11f, the arrows point to the left, indicating that the pumps are completely in anti-phase with the InSAR displacements; in other words, when the pumping rates are higher, the piezometric level decays and the ground surface subsides. This relation of both the groundwater level and pumping rate signal phases with the InSAR displacement can be linked to the hydrogeological triggering factor of the ground movements registered in the Matalascañas area, which reached 1.5 cm in the studied period.
4.4 Temporary Ponds
Ten aggregated polygons corresponding to areas of temporary ponds, identified from the map created from the LiDAR information, were analyzed in the present work. Figure 12 shows the temporal analysis results of the relationship between the displacement InSAR data and groundwater levels at two temporary pond locations, corresponding to aggregates 36 and 38 (Figure 3). The figure indicates significant similarities between the seasonality of the two time series, although their characteristics are quite different. Aggregate 36 is located at the Marismas wetland border in the southeastern part of the aquifer. It constitutes a drainage area from the aquifer to the wetland throughout the dune sands. Aggregate 38 is at a place named Laguna de los Arrayanes and does not present water at present. However, the LiDAR analysis and topological name indicate that there have been periods of accumulated surface water in this small area. It is located in the northwest part of the aquifer system, 3 km away from the nearest red fruit crops. The closest piezometers to aggregates 36 and 38 are 114360013 and 104160019, respectively. Both aggregates 36 and 38 present springtime uplift corresponding to the increased groundwater level and subsidence during summer, when the aquifer presents a lower groundwater level (Figures 12a–12d). These ground surface movements can reach 1 cm from uplift to subsidence. This seasonality can be physically explained by shrink-swell processes, the volume changes resulting from variations in the moisture contents of clay-rich soils. Swelling pressures can cause the heaving or lifting of structures, while shrinkage can cause settlement or subsidence (Jones et al., 2020). Similar small surface elevation changes at the millimeter scale, which are a result of frost heave and the shrinkage of clayey soils, have also been observed from InSAR measurements in an agricultural field in the Netherlands (Brake et al., 2013). Likewise, a multiyear trend and seasonal variations in the InSAR-observed ground displacement were observed by Li et al. (2020) over the southern Ontario region, and the authors also attributed these findings to changes in water storage. The cross-wavelet analysis performed for InSAR and groundwater level information at both locations identified similar seasonality of the one-year period with similar WTC and XWT power spectrum (Figures 11e and 11f). The temporal delay detected was analogous in both pond locations, with a value of 45 days, as the arrows indicating the phase differences had inclinations of 45° (Figures 12e and 12f).
InSAR-derived land subsidence is an excellent proxy for assessing groundwater overexploitation in cases where a steady groundwater level drawdown has been observed. However, the interdependence of the findings on pumping effects and other triggering factors might not be so evident. The proposed approach interrelating InSAR displacement, groundwater level and rainfall time series through wavelet analysis within a Python package allows for finer analysis and identification of other subsidence-related triggering factors, such as soil moisture balance, creep phenomena or clay shrinking-swelling processes.
In this study, integrated InSAR and hydrogeological analyses were applied to recognize, define and explain potential ground motions on the Almonte-Marismas aquifer, which hosts the Doñana Natural Space (southwest Spain). The results for the time interval from the end of 2014 to the beginning of 2020 reveal that the ground movement magnitude in this area is similar to the errors associated with the GEP InSAR data. This means that one must consider the associated uncertainties when interpreting InSAR information and support the results with auxiliary data such as groundwater levels, pumping rates and climate series. Furthermore, precision leveling or other advanced measures would provide better insight into the problem in the near future.
In the areas of the aquifer that presented relatively great potential displacements (i.e., the aggregated MPs following the robust criteria imposed), InSAR statistical and visual relations with geological and hydrogeological information (i.e., presence of clays, piezometry, precipitation and soil moisture balance) have pointed out different subsidence-triggering factors depending on the study zone within the aquifer system: long-term groundwater drawdowns, moisture balance in the presence of clays, and the vicinity to cone depressions. The short length of the studied period may be one of the reasons why, in the northern part of the aquifer system close to the most-exploited area, the InSAR displacement data do not show any visual linkage between subsidence rates and groundwater table variations. However, when performing wavelet analysis, common periodicities for both groundwater level and land subsidence are observed. Moreover, different time lags ranging from 8 to 2 months are measured, which may be associated with the capability of the aquifers to drain pore pressure variations caused by piezometric level changes coupled with other subsidence-triggering processes.
From 2014 to 2020, the subsidence rate in the Almonte-Marismas system aquifer ranged from 1 cm in Matalascañas and in the temporary pond zones to 2.5 cm in the northern study area. For the first time, in the Doñana area, a significant relationship between distances to pumping wells and the displacement extent was detected. Areas with higher potential displacement rates (e.g., Isla Mayor village) should be better monitored in the future to prevent possible deformation hazards. In situ GPS campaigns or installation of corner reflectors at strategic points would guarantee better accuracy and spatial density of InSAR measurements.
Additionally, a nonanthropogenic component of the water storage change was identified as a potential cause of ground motion, but this component was similar to the InSAR detection threshold of a few mm/year. This was the case in the temporary pond locations, where both deformational behaviors were detected: subsidence and uplift. These movements may be related to shrinkage and swelling processes in clay soils due to the groundwater table and soil moisture fluctuations.
The results demonstrate that although InSAR coherence is low over the area, this technique offers a relatively inexpensive means to monitor areas for overexploitation-driven effects with a potential time lag of less than 6 months to a year. Moreover, in view of the different proposals to reactivate gas extraction and storage projects in Doñana's vicinity, all of the above findings provide a ground motion baseline assessment that can help differentiate natural and historical variations from any future anthropogenic effects in the Doñana area.
We acknowledge the Geological and Mining Institute of Spain and Guadalquivir Hydrographic Confederation for providing the piezometric data, as well as Junta de Andalucía for providing the meteorological information. The authors wish to extend special thanks to Fernando Ruiz Bermudo and Antonio N. Martínez Sánchez de la Nieta from the Geological and Mining Institute of Spain. Without their special dedication, the observations would not have been available. We also thank Ayterra enterprise for sharing their information and expertise in the study area and the anonymous referees for their useful suggestions. Copernicus Sentinel-1 IW SAR data were provided and processed in ESA's GEP in the framework of the GEP Early Adopters Programme. Data processing was carried out with the P-SBAS and FASTVEL services developed and integrated by CNR-IREA and TRE-Altamira, respectively, in the GEP. This work is part of the subsidized activities within the National Youth Guarantee System (PEJ2018-002477), funded by the Tripartite Foundation for Training in Employment, the Youth Employment Initiative (YEI) initiative and the European Social Fund (ESF), the RESERVOIR project, which is part of the PRIMA Programme supported by the European Union (Grant Agreement number:  [RESERVOIR] [Call 2019 Section 1 Water RIA]), the SARAI project funded by the Ministry of Science and Innovation of Spain (MCIN/AEI/10.13039/501100011033, PID2020-116540RB-C22), and the extraordinary grants for excellence IGME-CSIC (AECEX2021).
Data Availability Statement
The deformation, climatic and hydrogeological data used for analysis of the aquifer-system deformation in the present study are available at Zenodo via https://doi.org/10.5281/zenodo.7180022 with Creative Commons Attribution 4.0 International (Guardiola-Albert et al., 2022). v.0.0.1 of WaSAR used for analyzing the ground deformation of a region and to compare it with other climatic variables, such as groundwater levels or rainfall is preserved at [DOI, persistent identifier link], available via https://doi.org/10.5281/zenodo.6334995 and developed openly at https://github.com/MiguelonGonzalez/wasar/tree/v0.0.1 (González-Jiménez & Guardiola-Albert, 2022).
- 2009). Practical persistent scatterer processing validation in the course of the Terrafirma project. Journal of Applied Geophysics, 69(1), 59–65. https://doi.org/10.1016/j.jappgeo.2009.07.002
- 2020). Study of systematic bias in measuring surface deformation with SAR interferometry. IEEE Transactions on Geoscience and Remote Sensing, 59(2), 1285–1301. https://doi.org/10.1109/TGRS.2020.3003421
- 2013). InSAR time-series analysis of land subsidence in Bangkok, Thailand. International Journal of Remote Sensing, 34(8), 2969–2982. https://doi.org/10.1080/01431161.2012.756596
- 2017). Mapping groundwater level and aquifer storage variations from InSAR measurements in the Madrid aquifer, Central Spain. Journal of Hydrology, 547, 678–689. https://doi.org/10.1016/j.jhydrol.2017.02.011
- 2016). Interpolation of GPS and geological data using InSAR deformation maps: Method and application to land subsidence in the Alto Guadalentín aquifer (SE Spain). Remote Sensing, 8(11), 965. https://doi.org/10.3390/rs8110965
- 2002). A new algorithm for surface deformation monitoring based on small baseline differential SAR interferograms. IEEE Transactions on Geoscience and Remote Sensing, 40(11), 2375–2383. https://doi.org/10.1109/TGRS.2002.803792
- 2020). 3D groundwater flow and deformation modelling of Madrid aquifer. Journal of Hydrology, 585, 124773. https://doi.org/10.1016/j.jhydrol.2020.124773
- 2014). Land subsidence from groundwater use in California, Full report of findings, report (p. 151). California Water Foundation.
- 2015). Understanding the subsidence process of a quaternary plain by combining geological and hydrogeological modelling with satellite InSAR data: The Acque Albule Plain case study. Remote Sensing of Environment, 168, 219–238. https://doi.org/10.1016/j.rse.2015.07.010
- 2013). Satellite-based radar interferometry to estimate large-scale soil water depletion from clay shrinkage: Possibilities and limitation. Vadose Zone Journal, 12(3), 1–13. https://doi.org/10.2136/vzj2012.0098
- 2016). Assessing groundwater depletion and dynamics using GRACE and InSAR: Potential and limitations. Groundwater, 54(6), 768–780. https://doi.org/10.1111/gwat.12453
- 2016). Groundwater depletion in Central Mexico: Use of GRACE and InSAR to support water resources management. Water Resources Research, 52(8), 5985–6003. https://doi.org/10.1002/2015WR018211
- 2014). SBAS-DInSAR parallel processing for deformation time-series computation. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 7(8), 3285–3296. https://doi.org/10.1109/JSTARS.2014.2322671
- 2014). Land subsidence in central Mexico detected by ALOS InSAR time-series. Remote Sensing of Environment, 140, 94–106. https://doi.org/10.1016/j.rse.2013.08.038
- 2020). Land subsidence and its relation with groundwater aquifers in Beijing Plain of China. Science of the Total Environment, 735(15), 139111. https://doi.org/10.1016/j.scitotenv.2020.139111
- 2022). Assimilation of SBAS-InSAR based vertical deformation into land surface model to improve the estimation of terrestrial water storage. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 15, 2826–2835. https://doi.org/10.1109/JSTARS.2022.3162228
- 2016). Imaging land subsidence induced by groundwater extraction in Beijing (China) Using satellite radar interferometry. Remote Sensing, 8(6), 468. https://doi.org/10.3390/rs8060468
- CHG (Confederación Hidrográfica del Guadalquivir). (2019). Propuesta para la declaración de la masa de agua subterránea de La Rocina en riesgo de no alcanzar un buen estado cuantitativo y químico. Retrieved from https://www.chguadalquivir.es/documents/10182/41541/la_rocina.pdf/79a36acc-2581-c18f-ffb4-d94c2bf6cefb
- 1992). Long-term gravitational deformation of rocks by mass rock creep. Engineering Geology, 32(3), 157–184. https://doi.org/10.1016/0013-7952(92)90043-X
- 1960). Relationship between pressure and moisture content of kaolinite, illite, and montmorillonite clays. AAPG Bulletin, 44(1), 101–106. https://doi.org/10.1306/0BDA5F81-16BD-11D7-8645000102C1865D
- 2021). Accuracy of Sentinel-1 PSI and SBAS InSAR displacement velocities against GNSS and geodetic leveling monitoring data. Remote Sensing, 13(23), 4800. https://doi.org/10.3390/rs13234800
- 2021b). Accuracy of Sentinel-1 PSI and SBAS InSAR displacement velocities against GNSS and geodetic leveling monitoring data. Remote Sensing, 13(23), 4800. https://doi.org/10.3390/rs13234800
- 2021a). Present-day land subsidence rates, surface faulting hazard and risk in Mexico City with 2014–2020 Sentinel-1 IW InSAR. Remote Sensing of Environment, 253, 112161. https://doi.org/10.1016/j.rse.2020.112161
- 2021b). Satellite InSAR survey of structurally-controlled land subsidence due to groundwater exploitation in the Aguascalientes Valley, Mexico. Remote Sensing of Environment, 254, 112254. https://doi.org/10.1016/j.rse.2020.112254
- 2008). Perspectiva general del papel y gestión de las aguas subterráneas en el Área de Doñana, Sudoeste de España. Boletin Geologico y Minero, 119(1), 81–92. https://www.igme.es/boletin/2008/119_1_2008/ARTICULO%207.pdf
- 2015). An on-demand web tool for the unsupervised retrieval of earth’s surface deformation from SAR data: The P-SBAS service within the ESA G-POD environment. Remote Sensing, 7(11), 15630–15650. https://doi.org/10.3390/rs71115630
- 2019). A genetic algorithm for phase unwrapping errors correction in the SBAS-DInSAR approach. In IGARSS 2019-2019 IEEE international geoscience and remote sensing symposium (pp. 266–269).
- 2015). Phase inconsistencies and multiple scattering in SAR interferometry. IEEE Transactions on Geoscience and Remote Sensing, 53(12), 6608–6616. https://doi.org/10.1109/TGRS.2015.2444431
- ESRI. (2011). ArcGIS desktop: Release 10. Environmental Systems Research Institute.
- 2014). A quasi-elastic aquifer deformational behavior: Madrid aquifer case study. Journal of Hydrology, 519, 1192–1204. https://doi.org/10.1016/j.jhydrol.2014.08.040
- 2020). Improving multi-technique monitoring using Sentinel-1 and Cosmo-SkyMed data and upgrading groundwater model capabilities. Science of the Total Environment, 703, 134757. https://doi.org/10.1016/j.scitotenv.2019.134757
- 2014). The global groundwater crisis. Natural Climate Change, 4(11), 945–948. https://doi.org/10.1038/nclimate2425
- 2016). Water availability and land subsidence in the Central Valley, California, USA. Hydrogeology Journal, 24(3), 675–684. https://doi.org/10.1007/s10040-015-1339-x
- 2000). Nonlinear subsidence rate estimation using permanent scatterers in differential SAR interferometry. IEEE Transactions on Geoscience and Remote Sensing, 38(5), 2202–2212. https://doi.org/10.1109/36.868878
- 2021). Python-visualization/folium v0.12.1. https://doi.org/10.5281/zenodo.4447642
- 2011). Review: Regional land subsidence accompanying groundwater extraction. Hydrogeology Journal, 19(8), 1459–1486. https://doi.org/10.1007/s10040-011-0775-5
- 2007). The application of satellite differential SAR interferometry-derived ground displacements in hydrogeology. Hydrogeology Journal, 15(1), 133–154. https://doi.org/10.1007/s10040-006-0121-5
- 1998). Detection of aquifer system compaction and land subsidence using interferometric synthetic aperture radar, Antelope Valley, Mojave Desert, California. Water Resources Research, 34(10), 2573–2585. https://doi.org/10.1029/98WR01285
- 2021). Land subsidence and its mitigation (p. 92). The Groundwater project. Retrieved from https://gw-project.org/books/land-subsidence-and-its-mitigation/
- 2018). Regional land subsidence analysis in eastern Beijing plain by InSAR time series and wavelet transforms. Remote Sensing, 10(3), 365. https://doi.org/10.3390/rs10030365
- 2022). WASAR: Analysis of A-DInSAR time series. (v0.0.1) [Software]. Zenodo. https://doi.org/10.5281/zenodo.6334996
- 2019). PyWavelets: A Python package for wavelet analysis. Journal of Open Source Software, 4(36), 1237. https://doi.org/10.21105/joss.01237
- 2004). Application of the cross wavelet transform and wavelet coherence to geophysical time series. Nonlinear Processes in Geophysics, 11(5/6), 561–566. https://doi.org/10.5194/npg-11-561-2004
- 2022). Data for the analysis of aquifer-system deformation in the Doñana Natural Space (Spain) using unsupervised cloud-computed InSAR data and wavelet analysis (v0) [Dataset]. Zenodo. https://doi.org/10.5281/zenodo.7180022
- 2009). Management of groundwater resources in the Donana area with the support of the Almonte-Marismas aquifer mathematical model; Gestión de los recursos hídricos subterráneos en el entorno de Doñana con el apoyo del modelo matemático del acuífero Almonte-Marismas. Boletín Geológico y Minero de España, 120(3), 361–376. Retrieved from https://www.igme.es/Boletin/2009/120_3_2009/6-ARTICULO%203%20361-376.pdf
- 2020). State of the art and recent advancements in the modelling of land subsidence induced by groundwater withdrawal. Water, 12(7), 2051. https://doi.org/10.3390/w12072051
- 2001). Radar interferometry: Data interpretation and error analysis (Vol. 2). Springer Science & Business Media.
- 2021). Mapping the global threat of land subsidence. Science, 371(6524), 34–36. https://doi.org/10.1126/science.abb8549
- 2012). Recent advances in SAR interferometry time series analysis for measuring crustal deformation. Tectonophysics, 514–517, 1–13. https://doi.org/10.1016/j.tecto.2011.10.013
- 2007). Matplotlib: A 2D graphics environment. Computing in Science & Engineering, 9(3), 90–95. https://doi.org/10.1109/MCSE.2007.55
- 2016). Most computational hydrology is not reproducible, so is it really science? Water Resources Research, 52(10), 7548–7555. https://doi.org/10.1002/2016WR019285
- 2020). Characterisation and mapping of land subsidence based on geodetic observations in Lagos, Nigeria. Geodesy and Geodynamics, 11(2), 151–162. https://doi.org/10.1016/j.geog.2019.12.006
- 2016). Anthropogenic and geologic influences on subsidence in the vicinity of New Orleans, Louisiana. Journal of Geophysical Research: Solid Earth, 121(5), 3867–3887. https://doi.org/10.1002/2015JB012636
- 2020). In D. P. Giles & J. S. Griffiths (Eds.), Swelling and shrinking soils geological Hazards in the UK: Their occurrence, monitoring and mitigation. Engineering geology special publications 29 (pp. 223–242). Geological Society. https://doi.org/10.1144/EGSP29.8
- 2021). Geopandas/Geopandas, v 0.10.2 [software]. Zenodo. https://doi.org/10.5281/zenodo.3946761
- 2020). Creep consolidation in land subsidence modelling; integrating geotechnical and hydrological approaches in a new MODFLOW package (SUB-CR). Proceedings of the International Association of Hydrological Sciences, 382, 499–503. https://doi.org/10.5194/piahs-382-499-2020
- 2005). Recent advances in wavelet analyses: Part 1. A review of concepts. Journal of Hydrology, 314(1–4), 275–288. https://doi.org/10.1016/j.jhydrol.2005.04.003
- 2020). Automatic generation of Sentinel-1 continental scale DInSAR Deformation time series through an extended P-SBAS processing pipeline in a cloud computing environment. Remote Sensing, 12(18), 2961. https://doi.org/10.3390/rs12182961
- 2007). An overview of the small baseline subset algorithm: A DInSAR technique for surface deformation analysis. Pure and Applied Geophysics, 164(4), 637–661. https://doi.org/10.1007/s00024-007-0192-9
- 2020). Surface deformation observed by InSAR shows connections with water storage change in Southern Ontario. Journal of Hydrology: Regional Studies, 27, 100661. https://doi.org/10.1016/j.ejrh.2019.100661
- 2021). A spatiotemporal analysis of the relationship between groundwater level and ground surface displacement using Sentinel-1 SAR data. Arabian Journal of Geosciences, 14(12), 1–10. https://doi.org/10.1007/s12517-021-07497-2
- 2013). Characterization of regional land subsidence induced by groundwater withdrawals in Tehran, Iran. Geopersia, 3(2), 49–62. https://doi.org/10.22059/jgeope.2013.36014
- 2019). The parallel SBAS approach for sentinel-1 interferometric wide swath deformation time-series generation: Algorithm description and products quality assessment. IEEE Transactions on Geoscience and Remote Sensing, 57(9), 6259–6281. https://doi.org/10.1109/TGRS.2019.2904912
- 2015). Deformation parameter estimation in low coherence areas using a Multisatellite InSAR approach. IEEE Transactions on Geoscience and Remote Sensing, 53(8), 249–252. https://doi.org/10.1109/TGRS.2015.2394394
- 2008). Land subsidence in Iran caused by widespread water reservoir overexploitation. Geophysical Research Letters, 35(16), L16403. https://doi.org/10.1029/2008GL033814
- 2020). Clustering groundwater level time series of the exploited Almonte-Marismas aquifer in southwest Spain. Water, 12(4), 1063. https://doi.org/10.3390/w12041063
- 2021). Characterization of groundwater recharge and flow in California's San Joaquin Valley from InSAR-observed surface deformation. Water Resources Research, 57(4), e2020WR028451. https://doi.org/10.1029/2020WR028451
- 2013). Land subsidence and ground failure associated to groundwater exploitation in the Aguascalientes Valley. Engineering Geology, 164, 172–186. https://doi.org/10.1016/j.enggeo.2013.06.015
- 2020). Integration of DInSAR and SBAS techniques to determine mining-related deformations using sentinel-1 data: The case study of Rydułtowy mine in Poland. Remote Sensing, 12(2), 242. https://doi.org/10.3390/rs12020242
- 2022). Artificial neural network modeling of the subsidence induced by overexploitation of groundwater in Isfahan-Borkhar Plain, Iran. Bulletin of Engineering Geology and the Environment, 81(5), 170. https://doi.org/10.1007/s10064-022-02646-7
- 2022). Multi-temporal analysis of groundwater depletion-induced land subsidence in Central Ganga Alluvial plain, Northern India. Geocarto International, 37(26), 11732–11755. https://doi.org/10.1080/10106049.2022.2060322
- 2020). Integration of DInSAR and SBAS Techniques to determine mining-related deformations using sentinel-1 data: The case study of Rydułtowy mine in Poland. Remote Sensing, 12(2), 242. https://doi.org/10.3390/rs12020242
- 2018). Quantifying ground deformation in the Los Angeles and Santa Ana Coastal Basins due to groundwater withdrawal. Water Resources Research, 54(5), 3557–3582. https://doi.org/10.1029/2017WR021978
- 2021). Climate influence Vs. local drivers in surface water-groundwater interactions in eight ponds of Doñana national park (Southern Spain). Wetlands, 41(25), 25. https://doi.org/10.1007/s13157-021-01425-6
- 2018). WaveletComp: Computational wavelet analysis. R package, version 1.1 [software]. CRAN. https://CRAN.R-project.org/package=WaveletComp
- 2010). Las formaciones Plio-Cuaternarias de El Abalario, en el litoral de la provincia de Huelva (España). Estudios Geológicos, 2(66), 209–225. https://doi.org/10.3989/egeol.39953.072
- 1995). Características litoestratigráficas de los depósitos plio-cuaternarios del bajo Guadalquivir en el área de Doñana: Implicaciones hidrogeológicas. Revista de la Sociedad Geologica de Espana, 8(1–2), 21–31.
- 2011). Chronology and tectono-sedimentary evolution of the Upper Pliocene to Quaternary deposits of the lower Guadalquivir foreland basin, SW Spain. Sedimentary Geology, 241(1–4), 22–39. https://doi.org/10.1016/j.sedgeo.2011.09.009
- 2013). A review on the applications of wavelet transform in hydrology time series analysis. Atmospheric Research, 122, 8–15. https://doi.org/10.1016/j.atmosres.2012.11.003
- 2010). Statsmodels: Econometric and statistical modeling with Python. In Proceedings of the 9th Python in science conference (SCIPY) (pp. 92–96). Retrieved from https://conference.scipy.org/proceedings/scipy2010/pdfs/seabold.pdf
- 2007). The groundwater economy of south Asia: An assessment of size, significance and socio-ecological impacts. In M. Giordano & K. G. Villholth (Eds.), The agricultural groundwater revolution: Opportunities and threats to development, Comprehensive assessment of water management in agriculture series 3 (pp. 7–36). CABI. https://doi.org/10.1079/9781845931728.0007
- 2018). Efficient ground surface displacement monitoring using sentinel-1 data: Integrating distributed scatterers (DS) identified using two-sample t-test with persistent scatterers (PS). Remote Sensing, 10(5), 794. https://doi.org/10.3390/rs10050794
- 2022). missForest: Nonparametric missing value imputation using random forest. R package version 1.5.
- 2014). When wells run dry. Nature, 516(7530), 179–180. https://doi.org/10.1038/516179a
- 2006). Groundwater pumping and land subsidence in the Emilia-Romagna coastland, Italy: Modeling the past occurrence and the future trend. Water Resources Research, 42(1), W01406. https://doi.org/10.1029/2005WR004242
- 2017). Earth observations for geohazards: Present and future challenges. Remote Sensing, 9(3), 1–10. https://doi.org/10.3390/rs9030194
- 2016). Using wavelet tools to analyse seasonal variations from InSAR timeseries data: A case study of the Huangtupo landslide. Landslides, 13(3), 437–450. https://doi.org/10.1007/s10346-015-0589-y
- 2020). Wavelet analysis of land subsidence time-series: Madrid Tertiary aquifer case study. Proceedings of the International Association of Hydrological Sciences, 382, 353–359. https://doi.org/10.5194/piahs-382-353-2020
- UNESCO. (2022). The united nations world water development report 2022. United Nations Educational, Scientific and Cultural Organization.
- UNESCO, IUCN, & Ramsar. (2021). Joint reactive monitoring mission to Doñana national park (Spain). Retrieved from https://rsis.ramsar.org/RISapp/files/RAM/RAM_095_ES_en.pdf
- 1991). Annual fluctuations of groundwater levels as a result of loading by surface moisture. Journal of Hydrology, 127(1–4), 137–152. https://doi.org/10.1016/0022-1694(91)90112-U
- 2017). Review: Moisture loading – The hidden information in groundwater observation well records. Hydrogeology Journal, 25(8), 2225–2233. https://doi.org/10.1007/s10040-017-1631-z
- 2014). Sustainability of global water use: Past reconstruction and future projections. Environmental Research Letters, 9(10), 104003. https://doi.org/10.1088/1748-9326/9/10/104003
- 2014). Monitoring land subsidence and fault deformation using the small baseline subset InSAR technique: A case study in the Datong Basin, China. Journal of Geodynamics, 75, 34–40. https://doi.org/10.1016/j.jog.2014.02.002
- 2004). Groundwater resources of the world and their use. UNESCO IHP-VI Series on Groundwater No 6 Retrieved from http://unesdoc.unesco.org/images/0013/001344/134433e.pdf
- 2016). A statistical test of phase closure to detect influences on DInSAR deformation estimates besides displacements and decorrelation noise: Two case studies in high-latitude regions. IEEE Transactions on Geoscience and Remote Sensing, 54(9), 5588–5601. https://doi.org/10.1109/TGRS.2016.2569435