Spatial downscaling of GRACE TWSA data to identify spatiotemporal groundwater level trends in the Dougherty Plain, South Georgia

Observations from the Gravity Recovery and Climate Experiment (GRACE) have become a valuable tool to identify regions across the world experiencing groundwater depletion. While many published studies focus on downscaling coarse (1) GRACE products, few studies have spatially downscaled GRACE to produce fine resolution maps that are more useful to resource managers. This study trained a boosted regression tree model to downscale GRACE total water storage anomaly to monthly five-kilometer groundwater level anomaly maps in the karstic Upper Floridan Aquifer in southwest Georgia using an array of hydrologic datasets. Model validation indicated satisfactory performance (R = 0.79, NSE = 0.61). Results demonstrate that groundwater levels were stable from 2002-2016 and highlighted areas where pumping may be exerting stronger influence on groundwater levels. Results demonstrate the applicability of machine learning methods for spatial downscaling of GRACE data and highlight the need for proper input data collection.