Movement of Amazon surface water from time-variable satellite gravity measurements and implications for water cycle parameters in land surface models

The large-scale observations of terrestrial water storage from GRACE satellites over the Amazon are analyzed with land surface model (LSM) outputs of runoff and soil moisture. A simple yet effective runoff routing method based on a continuity equation is implemented to model horizontal transport of surface water within the Amazon basin. The GRACE observations are analyzed separately for soil moisture and surface water storages (generated from runoff), relying on their distinct spatial patterns, being disperse for soil moisture and localized for surface water. Various effective velocities for storage transport are tested against the GRACE observations. When the model runoff is routed with an uniform velocity of 30 cm/s, the annual variation of the resulting surface water storage is generally found to be larger than the satellite measurements and ground gauge data by a factor of 1.5 or higher. The peak annual anomaly of surface water storage is observed around the midstream of the Amazon main stem. However, the runoff routing simulations present the peak amplitude consistently around the delta (downstream), unless the increasing velocity in a downstream region is used. As complements to the ground gauge data, the satellite observations provide unique \textquoteleftspatial\textquoteright information of water cycle parameters. Our analysis indicates possible shortcomings in the certain LSM mass transport scheme between atmosphere and land surface, particularly the production of too large seasonal variations in runoff (and maybe too little variations in evapotranspiration), and the dynamic characteristics of surface water transport within the Amazon basins.