According to the regional concentration characteristics of the Slepian basis function in the frequency and spatial domains,we established the Slepian basis function inversion model for recovering regional surface mass variations and the corresponding algorithm for solving the ill-posed problem using GRACE intersatellite geopotential difference(GPD)data.The accuracy and effectiveness of the inversion method had been evaluated through the estimation of terrestrial water storage change(TWSC)in the Amazon River basin.Firstly,the performance of the Slepian basis function method(GPD SBF),GPD-based mascon method(GPD Mascon),and GPD-based spherical harmonic method(GPD SH)for recovering the TWSC over the Amazon River basin in 2005 had been investigated through a closed-loop simulation study.The results demonstrate that the GPD SBF reduces the uncertainty caused by downward continuation in the inversion process by limiting the frequency and spatial domains of the signals,and effectively weakens the degree of ill-posed problem for solving the normal equations,and the stability of the solutions estimated by GPD SBF is significantly better than that of GPD Mascon.Furthermore,the accuracy and reliability of GPD SBF solutions are generally better than those of GPD Mascon solutions and GPD SH solutions,and GPD SBF can better recover the TWSC signal in edge regions and effectively reduce the leakage error.Secondly,the TWSC time series of the Amazon River basin were estimated by the measured GRACE GPD data from 2004 to 2015.The results indicate that the TWSC time series estimated by GPD SBF are more consistent with the official mascon solutions(CSR,JPL and GSFC RL06 mascon)than those estimated by GPD Mascon,and the GPD SBF solutions present more spatial details and better signal-to-noise ratio compared with the GPD Mascon solutions.Finally,using the hydrometeorological data(GPCP precipitation and ERA5 evapotranspiration)and the dS/dt(first derivative of TWSC)series derived from different GRACE solutions,the monthly averaged runoff time series at the Obidos hydrological station in the Amazon River basin were estimated through the water balance equation,and the runoff estimates were verified through in-situ runoff measurements.The results show that the residual standard deviations between the time series of estimated runoff and in-situ runoff with seasonal components removed are 12.35 mm and 14.54 mm for GPD SBF and GPD Mascon,and the corresponding correlation coefficients are 0.71 and 0.69,respectively.Moreover,the runoff estimates from the GPD SBF solutions and official mascon solutions are closer to the in-situ runoff measurements.This study demonstrates that the GPD SBF can weaken the dependence on regularization constraints for solving the ill-posed problem,and GPD SBF solutions have higher accuracy and reliability than the traditional GPD Mascon solutions,which provides a new solution strategy for estimation of high-precision and high-resolution regional surface mass variations.
维普
万方数据