Laura Braschoss, Nils Weitzel , Jean-Philippe Baudouin, Kira Rehfeld

Reliable projections of the future hydrological cycle are needed for designing adaptation and mitigation measures under global warming. However, uncertainties in the projected sign and magnitude of effective precipitation changes (precipitation minus evaporation, P-E) remain high. Here, we examine the state-dependency of circulation, temperature, and relative humidity contributions to P-E changes in simulations of the Last Glacial Maximum (LGM), mid-Holocene, and abrupt quadrupling of the atmospheric carbon dioxide concentration. To this purpose, we apply a moisture budget decomposition and a thermodynamic scaling approximation to CMIP6/PMIP4 simulations with the Earth system model MPI-ESM1.2. We find that the importance of thermodynamic and dynamic contributions to P-E changes and the patterns of dynamic contributions depend strongly on the underlying forcing. Greenhouse gas forcing leads to a stronger thermodynamic than dynamic response. The LGM ice sheets yield a large dynamic contribution with zonally heterogeneous patterns. Orbital forcing induces a predominantly dynamic response with a hemispherically anti-symmetric structure. We also identify state invariant features: the importance of temperature and relative humidity contributions to specific humidity changes is consistent across states, and the wet-get-wetter-dry-get-drier paradigm proposed for global warming holds in almost all regions dominated by thermodynamic contributions. By definition, the P-E budget is closed in the global mean. We find that, additionally, the respective thermodynamic, dynamic, and transient eddy contributions vanish in the global mean. Moreover for increasing length scales, the spatial variability of these contributions decays with similar rates. We suggest repeating our analysis for more models and states which could help constraining hydroclimate projections.