Similarities and divergent patterns in hydrologic fluxes and storages simulated by global water models

Global water models (GWMs) are critical tools for understanding the Earth’s water cycle and water resource management under a changing climate and accelerating human interventions. Although GWMs have been evaluated for hydrologic fluxes (for example, river discharge) and the role of representing human activities, there is a persistent gap in understanding models’ ability to simultaneously reproduce fluxes and storages (for example, terrestrial water storage (TWS)). Here we show that eight state-of-the-art GWMs do not consistently reproduce discharge and TWS with the same efficacy across varied geographic and climatic regions. Furthermore, model performance for discharge deteriorates as human impacts intensify. While a general agreement between simulated and observed TWS trends is found in two-thirds of major global river basins, models tend to underestimate the trends in both directions. Likewise, no single model simulates TWS trends and seasonality accurately and uniformly across major global river basins. Although improvements in capturing basin-averaged TWS trends, spatial distributions and seasonal fluctuations have been achieved compared with previous reports, challenges remain in accurately reproducing both fluxes and storages, owing primarily to inadequate representation of human activities in heavily managed regions. This study underscores critical disparities in GWM performance, emphasizing the need for further model enhancements, which is crucial for improved and more robust hydrologic assessments and predictions under climate change. This study evaluates global water models in capturing both water fluxes and storage, identifying well-performing regions and areas of discrepancy, ultimately emphasizing the need for further model refinements to better represent human activities.