Benchmarking soil moisture and its relationship to ecohydrologic variables in Earth System Models

Abstract. Soil moisture (SM) is a key regulator of ecosystem biogeophysics, influencing plant water relations and land-atmosphere energy exchanges. This study evaluates the representation of SM in Earth System Models (ESMs) using the International Land Model Benchmarking (ILAMB) framework, focusing on both surface (0–5 cm, 0–10 cm) and rootzone (0–100 cm) depths. We benchmark Coupled Model Intercomparison Project Phase 6 (CMIP6) models against multiple observational and assimilated datasets to evaluate their performance in simulating SM, as well as their relationships with ecohydrological processes and vegetation traits such as gross primary productivity (GPP), leaf area index (LAI), and evapotranspiration (ET). Results show that while surface SM is generally well represented (r > 0.87), rootzone SM variability is overestimated (normalized standard deviation > 1). Simulated ET agrees strongly with observations (r > 0.9; normalized standard deviation 0.8–1.2), whereas GPP and LAI exhibit greater discrepancies (r > 0.7; normalized standard deviation mostly > 1). The strength of SM–ecohydrology relationships varies with model structure and observational dataset, with better consistency observed when assimilated SM products are used. Regional analyses using Köppen classifications reveal distinct model behaviors, with stronger performance in tropical zones and reduced skill in high-latitude regions, likely due to challenges in simulating freeze–thaw and permafrost dynamics. These findings offer quantitative benchmarks of model performance, highlighting specific areas for improving SM representation and its coupling with vegetation and hydrological processes in future ESM development.