Mapping regional subsidence rate from electricity consumption-based groundwater extraction

Abstract

Study region: Choushui River alluvial fan in central Taiwan. Study focus: Excessive land subsidence is occurring in the Choushui alluvial fan, Taiwan. Groundwater pumping in the region is not monitored so pumping volumes are estimated from information based on electricity consumption. Groundwater pumping data alone cannot be used to accurately estimate the spatial rate and distribution of subsidence due to the low spatial correlation between estimated pumping volumes and subsidence from heterogeneous geologic materials. Time-dependent spatial regression is a reliable tool for estimating the spatial distribution of annual rate of subsidence based on spatial mapping of pumped groundwater volumes and monitored land subsidence. New hydrological insights for the region: Estimates of pumping quantities indicate that the locations of subsidence from overexploitation of groundwater from the underlying aquifer system is not always spatially correlated with major pumping areas or with the observed locations of maximum water-level drawdown. Considering time-dependent spatial regression models, the average RMSE of annual subsidence is 0.65 cm for the Choushui alluvial fan, indicating the high level of correlation between measured and estimated subsidence. These findings show that annual subsidence can be reliably estimated and predicted under the field pumping conditions of the study site. The influences of reduced withdrawals based on the relation between annual subsidence and pumping volumes can be evaluated for improved best management practices.

Keywords

• Time-dependent spatial regression
• Land subsidence
• Groundwater pumping volume