A Multi-Kernel Analysis of Radiative Feedbacks and Forcings on the Hydrological Cycle

All models predict precipitation to increase in response to a warming climate, but uncertainty in the magnitude of this change exists.  Global-mean precipitation is constrained by the atmospheric energy budget, and therefore, inter-model spread in the sensitivity of the hydrological cycle is determined by uncertainty in atmospheric radiative feedbacks and forcings.  To better understand the specific sources of inter-model spread, we use a combined radiative kernel-regression technique to compute radiative feedbacks, rapid adjustments and direct forcing at the top of atmosphere, surface and within in the atmosphere (top of atmosphere – surface) in a suite of CMIP5 global climate models. We also conduct a thorough comparison of radiative kernels developed from multiple modeling groups, to investigate the extent that differences in radiative transfer modeling contributes to the inter-model spread in precipitation change. In contrast to previous studies, we find that differences in the temperature and longwave water vapor radiative kernels are the leading contributors to the inter-kernel spread in the surface radiative feedbacks.  We further note that the majority of radiative kernels analyzed in this study contain an unexpected change in downwelling longwave radiation when the surface is warmed in isolation.  Our findings suggest that differences in model radiation schemes may be contributing more to the inter-model spread in hydrological cycle changes than previously reported.