Large biases in soil moisture limitation across CMIP6 models

Abstract

Accurate soil moisture representation is crucial in climate modeling, due to its significant role in land-atmosphere interactions. Our study focuses on water storage dynamics and analyzes how soil moisture limitation is represented in simulations from the land component (land-hist experiment) of seven models within the Coupled Model Intercomparison Project phase 6 (CMIP6). We quantified the annual maximum depletion in soil moisture, contrasting model results with observations of terrestrial water storage from the Gravity Recovery and Climate Experiment (GRACE). Our analysis shows that CMIP6 models mostly underestimate these annual extremes in soil moisture reductions, with the Amazon consistently emerging as the most biased region. We further computed the critical soil moisture thresholds and quantified the frequency of soil moisture limitation in CMIP6 simulations, comparing model estimates against solar-induced fluorescence (SIF) and GRACE observations. We found consistent results with the annual maximum depletion in soil moisture, with models almost always overestimating the frequency of soil moisture limitation globally compared to observations. We validated our findings with data from 128 eddy-covariance sites from eight biomes worldwide. Our study illuminates the biases in soil moisture storage and dynamics between CMIP6 models and empirical observations, highlighting the importance of improving the representations of soil moisture and land-atmosphere interactions in Earth System Models.