Joint inversion of GNSS and GRACE/GRACE-FO data for terrestrial water storage changes in Southwest China
The global navigation satellite system (GNSS) and gravity recovery and climate experiment (GRACE), along with its follow-on mission (GRACE-FO) have revolutionized studies of terrestrial water storage changes (TWSC). However, TWSC estimates obtained from these two techniques exhibit different characteristics and accuracy in terms of temporal and spa-tial scales. Therefore, it is crucial to integrate these observations to obtain high-precision TWSC at a consistent scale. In this study, we designed a joint inversion model that combines GNSS and GRACE/GRACE-FO data for regional TWSC based on the GNSS inversion model (GNSS-IM). The summation operator was used for the Jet Propulsion Laboratory (JPL) GRACE/GRACE-FO Mascon data to construct constraints in the joint inversion model. The optimal model parameters were selected using the Akaike's Bayesian information criterion. To confirm the feasibility of the joint inversion model for estimating TWSC in Southwestern China, we conducted 1000 simulation experiments. The simulation results indicate that the joint inversion yields an average root mean square error of 10mm, which is 47％ lower than that of GNSS-IM. Subsequently, we applied the joint inversion to estimate the TWSC in Southwest China from January 2011 to June 2022, utilizing GNSS and JPL Mascon da-ta. The comparison of the joint inversion results with the results of GNSS-IM, GRACE/GRACE-FO, and the high-resolution global land data assimilation system (GLDAS) shows that the TWSC annual amplitudes of the joint inversion exhibit the most consistent spatial distribution characteristics with those of GLDAS. It indicates that the joint inversion achieves better spatial resolution for TWSC compared to GNSS-IM and GRACE/GRACE-FO. Furthermore, we estimated the regional hydrological budget through the water balance equation and evaluated the uncertainty of the monthly rate of equivalent water height (d(EWH)/dt) of TWSC using the generalized three-cornered hat method. The results demonstrate that the uncertainty of d(EWH)/dt estimated by the joint inversion is 8mm/month, which is 33％, 68％, and 62％ lower than the results of GRACE/GRACE-FO, GNSS-IM, and the water balance equation, respectively. Our study indicates that the joint inversion ap-proach improves the accuracy of TWSC compared to GNSS and GRACE/GRACE-FO, and provides more reliable data for wa-ter resources management and hydrology research.
GNSSGRACE/GRACE-FOjoint inversionterrestrial water storage changesAkaike's Bayesian information criterion
万方数据
维普
