Investigating the Potential for Assimilation of Terrestrial Water Storage Observations from GRACE

Kevin Ellett, Jeffrey Walker, Andrew Western and Matthew Rodell

Power Point Presentation

Early results from the GRACE satellite mission are providing novel assessment of the global-scale terrestrial water cycle. GRACE observations of changes in total water storage also present new opportunities for model verification and data assimilation. This paper investigates the latter and identifies the primary challenges to utilizing GRACE in this context as (1) GRACE observation resolution and error, (2) downscaling from the observation to model scale and (3) decomposing the vertically integrated total terrestrial water storage change observation into the relevant components (e.g., soil moisture and ground water storage). Results from synthetic twin studies undertaken for the Murray-Darling Basin in Australia (106 km2) indicate that assimilation of GRACE observations can improve model prediction at the catchment scale (104-105 km2) for the case of relatively large errors in the initial soil moisture conditions. Improvement in prediction is a function of the catchment size since the variational-type scheme used in this study relies on areal weighting to implicitly downscale the GRACE observations. This initial investigation suggests that the effectiveness of data assimilation of GRACE observations will depend on the spatial resolution and inherent accuracy of the hydrological model, the size of the basin of interest, and the magnitude of the natural water storage change signal.