Two decades (2003-2021) of storage changes in the soil water and groundwater of CONUS

The climate change along with anthropogenic water use have been affecting the (re)distribution of water storage and water flux across the Continental United States (CONUS). Understanding these changes requires tools that provide a big picture of the processes that drive these changes. These processes are implemented in this study by combining remote sensing data with available modeling techniques, where the data contains a sample of hydro-climatological signals and models reflect our understanding of these processes. Time series of Terrestrial Water Storage Changes (TWSC) from the Gravity Recovery and Climate Experiment (GRACE, 2003-2017) and its Follow-on (GRACE-FO, 2018-2021) are integrated into the W3RA water balance model within CONUS, where the model is run at 9-km resolution using ERA5 forcing data. The gap in the GRACE and GRACE-FO TWSC products is filled following https://doi.org/10.3390/rs12101639. To achieve the best possible statistical combination, the Markov Chain Monte Carlo-based Data Assimilation (MCMC-DA) approach (https://doi.org/10.1016/j.scitotenv.2020.143579) is applied to use GRACE and GRACE-FO TWSC as a vertical integration constraint to update W3RA's individual water storage estimates. This approach rigorously accounts for the uncertainties of model and observations.

The outputs of MCMC-DA are evaluated using in-situ USGS groundwater level data and the European Space Agency (ESA) Climate Change Initiative (CCI) soil moisture product. The results indicate changes in trend and seasonality of water storage variations, for example, in southwestern (California and Nevada), southeastern (including Florida, South and North Carolina, Virginia, and Georgia), and south-central CONUS (including Texas, New Mexico, Colorado, Kansas, and Oklahoma). MCMC-DA improves the estimation of soil water in regions with high forest intensity, where ESA CCI and models reveal difficulties in capturing the soil-vegetation-atmosphere continuum. The representation of El Nino Southern Oscillation (ENSO)-related variability in water storage are found to be considerably improved after integrating GRACE(-FO) into W3RA. This new hybrid approach is found efficient for understanding the linkage between the climate variability and large-scale hydrological processes.

Keywords: CONU; Data Assimilation; Bayesian Method; MCMC; GRACE(-FO); W3RA; groundwater storage; soil water storage; USGS; ESA CCI.