Closing the water balance of large watersheds using satellite gravimetry

At the level of a watershed, the conservation of mass imposes that the net moisture transport through the atmospheric boundaries is balanced by the river discharge and an accumulation/depletion in terrestrial sources such as the soil, surface waters and groundwater. There are considerable uncertainties connected with modelled water balance components, especially since most models only simulate part of the system: either the atmosphere, the surface or the subsurface. Uncertainties in boundary conditions propagate as biases in the simulated results. For example, not accounting for anthropogenic groundwater extraction potentially introduces biases in arid regions, where groundwater is a non-negligible source of moisture for the atmosphere. The use of observations is therefore an important aid to evaluate model performances and to detect possible biases in water balance components. In this contribution, we compare total water storage changes derived from the Gravity Recovery Climate Experiment (GRACE) and its follow-on mission, with modelled components of the water balance. We use ERA5 reanalysis data to compute (net) atmospheric transports, and river discharge from GloFAS (Global Flood Awareness System). Furthermore, we use precipitation estimates (e.g. from GPCC) together with evapotranspiration from the Surface Energy Balance System (SEBS). We finally perform an accounting of the water balance components for the world’s largest watersheds and show to what extent we can find agreements, inconsistencies and biases in the data and models