SENSITIVITY ANALYSIS OF HYDROGEOLOGICAL AND MECHANICAL PARAMETERS TO MODEL 1D LAND SUBSIDENCE IN THE HIGHLY COMPRESSIBLE HETEROGENEOUS AQUITARD OF MEXICO CITY
We employ automated inverse modeling to conduct a sensitivity analysis on a 1D nonlinear flow and subsidence model of the highly compressible heterogeneous aquitard of Mexico City. We focus our investigation on a research site that was instrumented with piezometers and extensometers and was monitored for 10 years. The estimated thickness of the aquitard at the research site is 100 m. Our algorithm to model groundwater flow and deformation considers the nonlinearity arising from the fact that hydraulic conductivity (K), void ratio (e) and the storage coefficient (Ss) depend on the effective stress. We use the PEST software to conduct the inverse simulation. Following the conceptual hydrogeological and geomechanical analysis of Zapata-Norberto et al. “Analysis and numerical modeling of data from piezometers and extensometers reveal the main drivers of land subsidence in Mexico City” (related abstract and talk in this RAUGM), we model the deformation due to groundwater pumping imposing a known drawdown rate at the base of the 1D domain to represent the drawdown rate observed in the regional aquifer. The different strata of the aquitard are represented in the 1D model (upper clay formation UCF, hard layer HL, lower clay formation LCF, deep deposits DD, deep clay formation DCF and deep stratified formation DSF) and the aquifer is represented by a section of 36 m length. Parameters within each stratum of the aquitard are assumed uniform (given that K, e and Ss are stress-dependent, and hence change with time, only their initial value are calibrated during the inverse procedure). The results show that in general, hydraulic conductivity K is the parameter with the largest sensitivity; in addition, the sensitivity is larger at the base of the aquitard (DSF) and diminishes for the upper strata (the UCF and HL showed zero sensitivity because pore pressure has not changed). Only the compression index (Cc) of the deepest strata of the aquitard (DSF) showed a significant sensitivity value, but its magnitude is one order of magnitude smaller than the sensitivity of K. The sensitivity of void ratio e of DSF and DD showed a significant sensitivity but also one order of magnitude smaller than the sensitivity of K. Our results highlight the importance of hydraulic conductivity K in modeling land subsidence driven by groundwater pumping.