Observation-constrained projections reveal less increase in future global terrestrial evapotranspiration

Terrestrial evapotranspiration ( ET ) is a key component of the water cycle and energy balance. While the Earth system models (ESMs) in the Coupled Model Intercomparison Project Phase 6 (CMIP6) project that global ET will increase through the end of this century, the magnitude of this change remains highly uncertain because future trends cannot be directly validated against historical observations. This partly explains why the Intergovernmental Panel on Climate Change Sixth Assessment Report assigns only “ medium confidence ” to projected hydrological changes over the same period. Here, we provide more reliable future (2015–2099) global ET projections by leveraging water-balance observations in world’s large river basins. We show that future global ET trends are significantly correlated with historical basin-wide ET trends across certain ESMs skillful in capturing long-term trends. Applying an Emergent Constraint approach, we find that the raw CMIP6 ESM ensemble overestimates future global ET trend by 30% under both low- and medium-emission scenarios, and by 43% under high-emission scenario. Even for a subset of 18 well-performing ESMs, the unconstrained ensemble overpredicts the future global ET trend by 15–22% relative to the constrained estimates under three emission scenarios. The observational constraint also reduces the inter-model variance in future global ET projections by approximately one-third. Our conclusions are robust to the precipitation dataset in the water-balance calculation, projection length, or the ESM selection, although the absolute constrained trend values vary slightly with these choices. These findings enhance confidence in future global ET changes and underscore the crucial role of historical trends measured at the basin scale in refining global hydrological projections.