Why is Hydrological Sensitivity Larger in AMIP Uniform Warming Experiments than in CMIP Experiments?

The sensitivity of global mean precipitation to the Earth's surface temperature, often referred to as hydrological sensitivity, has been found consistently larger in AMIP uniform warming experiments than that in CMIP experiments. Therefore, global mean precipitation in CMIP experiments would be over-predicted if the hydrological sensitivity from AMIP uniform warming were used. The mechanism behind this gap, however, is not clear. Here we demonstrate that the gap is mainly caused by latitudinal heterogeneity in surface temperature change in CMIP experiments and the sensitivity of global precipitation to local surface temperature: surface temperature changes mildly over the tropics where surface temperature change is much more efficient in inducing global precipitation anomaly while the temperature change is amplified over extratropical areas where global precipitation is less sensitive to local surface temperature change. Based on these findings, we propose a new definition of hydrological sensitivity as the sensitivity of global mean precipitation to the change of tropical mean sea surface temperature (TMSST) instead of global mean surface temperature (GMST), which is much more consistent between AMIP uniform warming and CMIP experiments. Our results suggest that TMSST might be a better choice than GMST to represent the state of the climate system and be used to normalize climate variables, e.g., in climate sensitivity and feedback studies.