Comparative analysis of recent hydrological models and an attempt to generate new combined products for monitoring terrestrial water storage change

Hydrological models are crucial for characterizing large-scale water quantity variations and correcting GNSS reference station vertical displacements. We evaluated the robustness of multiple models, such as the Global Land Data Assimilation System (GLDAS), the Famine Early Warning System Network Land Data Assimilation System (FLDAS), the National Centers for Environmental Prediction (NCEP), and the WaterGAP Global Hydrology Model (WGHM). Inter-model and outer comparisons with Global Positioning System (GPS) coordinate time series, satellite gravity field Mascon solutions, and Global Precipitation Climatology Centre (GPCC) guide our assessment. Results confirm WGHM's 26% greater effectiveness in correcting nonlinear variations in GPS height time series compared to NCEP. In the Amazon River Basin, a 5-month lag between FLDAS, GLDAS, and satellite gravity results is observed. In eastern Asia and Australia, NCEP's Terrestrial Water Storage Changes (TWSC)-derived surface displacements correlate differently with precipitation compared to other models. Three combined hydrological models (H-VCE, H-EWM, and H-CVM) utilizing Variance Component Estimation (VCE), Entropy Weight Method (EWM), and Coefficient of Variation Method (CVM) are formulated. Correcting nonlinear variations with combined models enhances global GPS height scatter by 15%–17%. Correlation with precipitation increases by 25%–30%, and with satellite gravity, rises from 0.2 to 0.8 at maximum. The combined model eliminates time lag in the Amazon Basin TWSC analysis, exhibiting a four times higher signal-to-noise ratio than single models. H-VCE demonstrates the highest accuracy. In summary, the combined hydrological model minimizes discrepancies among individual models, significantly improving accuracy for monitoring large-scale TWSC.