Uncertainties in GRACE-Groundwater Assessments and Their Implications for Groundwater Drought

Abstract

Groundwater is an essential freshwater resource that underpins ecological resilience, agricultural productivity, and long-term water security. Despite its importance, large-scale assessment remains constrained by the limited spatial and temporal coverage of conventional monitoring networks. The Gravity Recovery and Climate Experiment (GRACE) and its follow on mission (GRACE-FO) have transformed groundwater monitoring by providing global estimates of terrestrial water storage anomalies (GRACE-TWSA), from which groundwater storage anomalies (GRACE-GWA) can be derived. However, isolating GRACE-GWA remains methodologically challenging due to uncertainties in auxiliary water storage components, such as soil moisture, snow, and surface water, and limited validation against in-situ groundwater observations. This thesis addresses these challenges by refining the derivation and evaluation of GRACE-GWA and examining the implications of associated uncertainties for groundwater drought characterisation. A systematic integration of diverse auxiliary datasets into the water balance framework reveals that discrepancies among water budget components introduce significant bias and variability in GRACE-GWA estimates. Analysis across three hydrologically complex basins demonstrates strong sensitivity to input selection, with in-situ correlation coefficients ranging from 0.32 to 0.89. To improve evaluation fidelity, a multi objective comparison framework, combining Nash–Sutcliffe Efficiency (NSE) and Kling Gupta Efficiency (KGE), is implemented, offering a more holistic assessment of time series agreement than traditional correlation-based methods. This approach reveals substantial inter model divergence in GRACE-GWA trends and amplitudes, reinforcing the value of ensemble based strategies. Building on these insights, the GRACE-Groundwater Drought Index (GGDI) is revisited using multi-model GRACE-GWA realisations to investigate how uncertainty influences drought indicators across 37 major aquifers. Results reveal that GRACE-GWA uncertainty compromise drought reliability and that aquifers with high memory, quantified via GGDI autocorrelation, experience fewer but longer and more severe droughts. Overall, this research delivers a scalable, uncertainty-aware framework to advance the accuracy and applicability of GRACE-based groundwater monitoring under intensifying climatic and anthropogenic pressures.