Inferring the storage of aquifer systems from InSAR measurements via flow and geomechanical modelling

In hydrogeological science, it is widely acknowledged that the response of an aquifer to groundwater pumping is predominantly influenced by two key parameters characterizing the aquifer system: saturated hydraulic conductivity Ks and the oedometric bulk compressibility cm. The response must be viewed in terms of changes of pore water pressure p and the deformation of the pore volume which traduces in a movement of the land surface. In a confined aquifer system subjected to groundwater pumping, the variation in groundwater pressure is linearly dependent on Ks, while the deformation of the pore volume is linearly dependent on cm. However, cm also impacts p, particularly the speed of pressure variation over time, as aquifer specific storage Ss is also dependent on cm. The dependency p - cm can be considered “weaker” than that p - Ks.  The RESERVOIR Project, funded by the EU-PRIMA Programme, aims to characterize aquifer properties, with a focus on Ss, by optimizing the use of the available measurements. Pressure measurements from piezometers provide fundamental information to quantify Ks through the groundwater flow equation. Additionally, displacement measurements of the land surface provided by InSAR can be optimally used in equilibrium equations to constrain cm (and consequently Ss). This objective is achieved through a novel procedure utilizing a 3D groundwater flow simulator (MODFLOW) and a 3D geomechanical simulator (GEPS3D) in an iterative one way coupled approach. Spatial variations of Ss and Ks are mathematically described as stationary Gaussian random fields. The procedure is applied to characterize the properties of the alluvial aquifer system in the eastern portion of the Gediz River basin, Turkey. In this region, groundwater withdrawal for irrigation has led to a general decline in pore water pressure and land subsidence of up to 10 cm/year over the past decade. The convergence of the procedure was achieved after four iterations, highlighting the presence of considerable heterogeneity in the distribution of parameters. This heterogeneity cannot be effectively constrained without the aid of satellite-based earth observation measurements.