Understanding the Intermodel Spread in Global-Mean Hydrological Sensitivity

This paper assesses intermodel spread in the slope of global-mean precipitation change P with respect to surface temperature change. The ambiguous estimates in the literature for this slope are reconciled by analyzing four experiments from phase 5 of CMIP (CMIP5) and considering different definitions of the slope. The smallest intermodel spread (a factor of 1.5 between the highest and lowest estimate) is found when using a definition that disentangles temperature-independent precipitation changes (the adjustments) from the slope of the temperature-dependent precipitation response; here this slope is referred to as the hydrological sensitivity parameter . The estimates herein show that is more robust than stated in most previous work. The authors demonstrate that adjustments and estimated from a steplike quadrupling CO2 experiment serve well to predict P in a transient CO2 experiment. The magnitude of is smaller in the coupled ocean-atmosphere quadrupling CO2 experiment than in the noncoupled atmosphere-only experiment. The offset in magnitude due to coupling suggests that intermodel spread may undersample uncertainty.Also assessed are the relative contribution of , the surface warming, and the adjustment on the spread in P on different time scales. Intermodel variation of both and the adjustment govern the spread in P in the years immediately after the abrupt forcing change. In equilibrium, the uncertainty in P is dominated by uncertainty in the equilibrium surface temperature response. A kernel analysis reveals that intermodel spread in is dominated by intermodel spread in tropical lower tropospheric temperature and humidity changes and cloud changes.