Basin-scale terrestrial water storage changes inferred from GRACE-based geopotential differences: a case study of the Yangtze River Basin, China

The Gravity Recovery and Climate Experiment (GRACE) mass concentration (mascon) solutions provide enhanced signal and spatial resolution of surface mass changes by using regularization techniques to reduce striping errors. To further improve the computational efficiency and capture the same benefits as GRACE mascon solutions, we presented an estimation of regional mascon solutions from GRACE-based geopotential differences by using spatio-temporal constraints with the unconstrained spherical harmonic solutions as a priori information. As a case study, the changes in the basin-scale terrestrial water storage (TWS) over the Yangtze River Basin (YRB) on 2^◦ × 2^◦ grids at monthly intervals were estimated using GRACE-based geopotential differences, for the period of 2003 January—2013 December. The estimates were validated through official GRACE mascon solutions and in situ observations (i.e. time derivative of TWS change derived from precipitation, evapotranspiration and river run-off based on the water mass balance equation). The results demonstrate that the spatial and temporal patterns of TWS changes in the YRB inferred from geopotential differences adequately agree with the official mascon solutions; however, differences in amplitudes can be observed at the subbasin scale because of different regularizations applied in different solutions. In situ validations demonstrate that seasonal changes of mascon solutions and in situ observations agree well in the YRB; however, there are evident discrepancies in amplitudes over the subbasins owing to leakage biases in mascon solutions. For the entire YRB, the statistical evaluation and cross-wavelet transform demonstrate that our regional mascon solutions appear more consistent with in situ observations than the official mascon solutions. In addition, compared with the results estimated by spatial constraints, regional mascon solutions estimated by spatio-temporal constraints using observations from three consecutive months adjacent to the given month were improved. Our method provides an alternative option to use different regularization constraints, which is helpful for fine-tuning analysis of basin-scale TWS changes.