Publication Date:
2023-06-27
Description:
Due to different reanalysis methods and input data, there exist big discrepancies between hydrological models from regions to regions. Therefore, accuracy assessment of different models is necessary before applications in specific areas. In this report, six hydrological models, including GLDAS, FLDAS, ERA5, MERRA-Land, NCEP and WGHM were evaluated and analyzed through inter-comparison between models and outer-comparison with Global Positioning System (GPS) station height time series, GRACE/GRACE-FO RL06 Mascon solutions, and Precipitation data from Global Precipitation Climatology Centre (CPCC) for a comprehensive assessment of model differences. We then try to combine the six models through variance component estimation (VCE), entropy weight method (EWF), coefficient of variation method (CVM) and other mathematical models, so as to improve the accuracy, integrity and applicability of hydrological models. Our results show that the root mean square (RMS) of global GPS height time series improves by up to 17% after correcting the hydrological effect obtained from combined models compared with that from individual models, while the correlation with rainfall also improves by 30% at most. Compared with TWSC derived from satellite gravity inversion, the correlation coefficient increases from 0.2 to 0.8 at the highest. Finally, the combined methods exhibit much higher signal-to-noise ratio (SNR) value than that of the pre-combined models, and the VCE combined model performs as the optimal hydrology model to correct the GPS height. Therefore, we conclude that the combined hydrological model could provide a better data source for monitoring hydrological changes and surface load deformation at a global scale.
Language:
English
Type:
info:eu-repo/semantics/conferenceObject
Permalink