Estimating GRACE Monthly Water Storage Change Consistent with Hydrology by Assimilating Hydrological Information

Abstract

A sequential estimation approach is used for constraining GRACE monthly estimates of mass changes with observed hydrological data, which is available for 20% of the land area, in order to improve the overall quality of the GRACE dataset. It is expected that the hydrological data constrains GRACE by utilising the correlations within the spherical harmonic coefficients, which is described by a simulated covariance matrix. Due to the dependancy of the approach on the stochastic information of GRACE, the influence of different structures of the GRACE covariance matrix were also tested. Initial results show that the hydrology constraints replace GRACE completely in the constrained areas, and contribute only meagrely outside the constrained areas. This hints at better parametrization of the model. The tests with different structures of the GRACE covariance matrix indicate that the block-diagonal structure approximates the full covariance matrix very well. © Springer-Verlag Berlin Heidelberg 2010.