Key points are not available for this paper at this time.
The Vega de Granada aquifer stands out as one of the primary detrital aquifers in the "Alto Genil" Basin in Southern Spain. Its significance lies in its vast extension, covering nearly 200 km2, and its substantial renewable water resources amounting to approximately 160 hm3/yr. Positioned strategically in the metropolitan area of Granada, it holds great relevance from a social point of view. Historically, it has been a crucial water source for meeting agricultural and urban water demands in various municipalities within the Vega de Granada. Over recent decades, groundwater extraction has escalated significantly, driven by urban expansion, and especially during severe droughts that periodically impact the region, resulting in highsubsidence ratesrelated to substantial groundwater level depletions. Historical subsidence rates have been monitored using remote sensing techniques, specifically Differential Interferometric Synthetic Aperture Radar (DInSAR). Previous studies utilized 3independent sets of images from different satellites: the ENVISAT satellite (C-band) and Sentinel-1A satellites (C-band) from the European Space Agency, and the Cosmo-skyMed constellation (X-band) from the Italian Space Agency. The integration of these datasets has enhanced the definition of the affected area by ground deformation and its temporal evolution. Presently, theEuropean Ground Motion Servicefrom Copernicus provides user-friendly information about ground deformation rates across Europe. EGMSrepresents a novel tool for the study of natural/induced processes such as land subsidence. We utilized compiled historical information to devise a preliminary method for assessing groundwater level depletion and its associated subsidence rates in potential future scenarios. The method simulates future groundwater level drawdowns through the application of a straightforward lumped balance equation proposed by Scott (2011). Various approaches, including simple conceptual models and machine learning techniques, were tested to simulate groundwater level dynamics. These approaches aided in a more comprehensive assessment, considering the structural uncertainty associated with different simulation methods. Additionally, we explored linear regression models and neural network approaches (such as NAR or ELMAN) to assess subsidence resulting from groundwater level depletion. Machine learning techniques proved effective in providing better insights into non-linear subsidence processes. In selected points, potential future subsidence in the horizon of 2071-2100 may double in a business-as-usual scenario within the aquifer. Based on the analysis of potential future subsidence values, we identified constraints that should be imposed on groundwater policies due to the associated risk of land subsidence resulting from groundwater level depletion. Acknowledgments: This research has been partially supported by the projects: STAGES-IPCC (TED2021-130744B-C21) and SIGLO-PRO (PID2021-128021OB-I00), from the Spanish Ministry of Science, Innovation and Universities.
Mateos et al. (Mon,) studied this question.