GNSS evaluation of GRACE-assimilated water storage models over 89 river basins worldwide

TL;DR

This study compares two water storage models using GNSS data to evaluate their accuracy in predicting water storage changes across 89 river basins worldwide.

Contribution

The paper introduces a novel evaluation method using GNSS data to assess GRACE-assimilated hydrological models globally.

Findings

CLSM-DA shows better agreement with GNSS data in regions like Africa and Southeast Asia.

Both models struggle during extreme events like droughts, highlighting their limitations.

Phase delays in CLSM-DA improve alignment with GNSS in some regions.

Abstract

The gravity recovery and climate experiment (GRACE) and GRACE follow-on (GFO) gravity observations have significantly improved our understanding of the terrestrial water cycle. However, GRACE-assimilated (GA) hydrological models still differ significantly. This paper uses global navigation satellite system (GNSS) data to assess two global GA datasets: Global land water storage release 2 (GLWS2.0) and catchment land surface model GRACE data assimilation (CLSM-DA). From 2004 to 2019, the mean annual amplitude of equivalent water thickness (EWT) of these datasets differs by more than 25 mm over 40% of the modeled land area, and the timing of peak water storage diverges by as much as 30-days across 50% of their domain. We compare the modeled hydrological loading vertical displacement predicted from these models with GNSS uplift data to compare and contrast the model quality. Using river basin boundary information from 89 rivers, we cluster 9,163 global GNSS stations, each with at least three years of daily data. Results show that CLSM-DA generally agrees better with GNSS data across more river basins. Its 100–300 mm larger annual water variation accounts for better agreement in Africa, Southeast Asia, and parts of South America. In regions like the Western United States and Eastern Europe, where both models estimate similar annual amplitudes, CLSM-DA’s 30–60 day phase delay improves alignment with GNSS. This evaluation also reveals key limitations in both models, especially during extreme hydrological events such as droughts, and highlights the value of geodetic observations in advancing GA hydrological modeling.

The online version contains supplementary material available at 10.1038/s41598-025-31887-1.

Figures