Abstract:
Results are presented from one of the first attempts to assimilate remotely sensed terrestrial water storage anomaly data into a land surface model (LSM). The data used were 10-day average anomalies for each of the four major sub-basins of the Mississippi River Basin, derived from Gravity Recovery and Climate Experiment (GRACE) satellite observations of Earth's time variable gravity field. Assimilation was performed using an Ensemble Kalman Filter (EnKF) for 30-member ensemble simulations of the Catchment LSM. In addition to simulating soil and snow water storages, the Catchment LSM accounts for variations in the elevation of the water table, making it appropriate for total terrestrial water storage applications. For these experiments the model was configured to run over Pfafstetter level 5 catchments with 2.0 x 2.5 degree atmospheric forcing, resulting in 1950 1-D model tiles for the Mississippi River basin. The simulations that included EnKF assimilation of GRACE data produced water storage time series which more closely resembled the original GRACE observations, relative to the open loop simulations. Improvements included a larger annual range in water storage for the combined Lower Mississippi / Red / Arkansas River Basin, substantial summer drying of the Ohio-Tennessee basin in 2003, and relatively mild summer drying for the Ohio-Tennessee in 2004. Pending independent validation, these results emphasize the potential for GRACE to improve the accuracy of hydrologic model output, which will benefit water cycle science and water resources applications. Furthermore, data assimilation enables coarse resolution, vertically integrated terrestrial water storage anomalies from GRACE to be spatially and temporally disaggregated and attributed to different levels of the snow-soil-aquifer column in a physically meaningful way.