Monitoring groundwater storage change through joint assimilation of GRACE terrestrial water storage and SMOS soil moisture observations

Monitoring groundwater storage change is of considerable value to freshwater supply and irrigated agriculture. Quantifying groundwater storage change remains challenging at large scale owning to the lack of monitoring infrastructure. The groundwater depletion caused by water withdrawals from pumping wells is often unmonitored and unable to be modeled. Recent studies have demonstrated that combining auxiliary hydrological datasets with terrestrial water storage (TWS) estimates from the Gravity Recovery and Climate Experiment (GRACE) can offer improved groundwater storage change estimation albeit at a coarse scale ( 300 km resolution). However, accurate soil moisture, surface water and snow data are critical to isolate groundwater component from TWS. Near-surface soil moisture (SM) retrievals from the Soil Moisture and Ocean Salinity (SMOS) satellite provide opportunity to better disaggregate groundwater storage change from TWS. In this study, we assimilated GRACE TWS and SMOS SM into a water balance model to quantify groundwater change in Australia. Through the assimilation, the satellite observations updated the individual water storage components (from top-layer soil to groundwater stores) at daily time steps over a one-month assimilation window. Our results showed that combining TWS and SM observations together can achieved more accurate and realistic estimates of model simulated groundwater storage, SM profile and TWS. The assimilation resulted in improved modeled estimates of profile moisture (0-90 cm) for 80% of in situ monitoring sites, and groundwater storage for 70% of bore sites across the country compared to open-loop (unconstrained) model simulations. The temporal correlation with in-situ measurements and linear trend of model simulated groundwater storage were improved by up to 0.9 compared to model open-loop simulations. Our analysis of the temporal trend across Australia found a significant reduction in groundwater over the southeastern Australia during 2002-2010 and an increase during 2010-2013. These trends we not observed in the model open-loop simulations, suggesting that the joint assimilation offers a new tool for monitoring global groundwater storage change.