Large biases in the frequency of water limitation across Earth system models

Scripts and instructions to reproduce the analysis presented in "Large biases in the frequency of water limitation across Earth system models" by Giardina et al. You are required to cite Zenodo data repository when using this data as follows:Giardina et al. (2025). Large biases in the frequency of water limitation across Earth system models. doi://10.5281/zenodo.10810324 Abstract Water availability limits evapotranspiration on land, shaping the energy balance, land carbon uptake, and climate extremes. Despite its importance, Earth System Models struggle to capture where and how often water-limited conditions occur. Here we investigate the representation of water limitation and its link to land water storage capacity in simulations from the Sixth Coupled Model Intercomparison Project (CMIP6) driven by consistent observational atmospheric forcing. Using observations of remotely sensed solar-induced vegetation fluorescence and terrestrial water storage, together with ecosystem flux data, we find that CMIP6 models overestimate the frequency of water limitation by 14% over land and 26% in the tropics. Model overestimation occurs over 57% of the land area, and 78% in the tropics. These too frequent water-limited conditions are not conclusively linked to a potential underestimation of land water storage capacity in the models, hinting at gaps in how ESMs represent rooting depths, plant water uptake, and plant water-use strategies. Our study highlights the need for model development in these areas, with implications for projections of future climate on land. Licensing All code is licensed under AGPL-v3, and the manuscript and data are licensed as CC-BY.