Estimation of Terrestrial Water Storage Changes in Brazil From the Joint Inversion of GRACE-Based Geopotential Difference and GNSS Vertical Displacement Data

Gravity Recovery and Climate Experiment (GRACE) satellite gravimetry and Global Navigation Satellite System (GNSS) surface displacement measurements offer complementary advantages for monitoring terrestrial water storage (TWS) changes. We propose a new joint inversion model based on the combination of GRACE-based geopotential difference (GPD) observations using the mascon method and GNSS vertical displacements through the Green's function method to obtain reliable TWS changes in Brazil. The performance of the jointly inverted TWS changes is assessed through closed-loop simulations and comparisons with hydrometeorological data (precipitation-P, evapotranspiration-ET, and runoff-R) using water budget closure (P-ET-R) and river water level observations from satellite altimetry. The simulation results indicate that the joint inversion results exhibit higher accuracy and reliability than GRACE GPD-based mascon (GPD-mascon) and GNSS solutions, and the standard deviations (STDs) of joint results decrease by ∼6.98 and ∼37.5 mm compared to those from GPD-mascon and GNSS solutions. The joint inversion of real GRACE and GNSS data demonstrates notably lower uncertainty than that of GPD-mascon solutions and exhibits significant improvement than GNSS-only solutions. The STDs and correlation coefficients between monthly R time series derived from three inversion methods (joint inversion, GPD-mascon and GNSS solutions, combined with P and ET through water budget closure) and in situ observations are 19.607 mm and 0.912, 20.879 mm and 0.904, and 31.370 mm and 0.778, respectively. The joint estimates also yield better correlation with river water level observations compared to GPD-mascon and GNSS solutions. Furthermore, at the weekly time scale, the joint inversion results show better consistency with P-ET-R data than those from GPD-mascon and GNSS solutions. © 2024. The Author(s).