An enormous amount of attention has been paid to the diversity of responses in the CMIP and other multi-model ensembles to changes in the state of the atmosphere that introduce a radiative forcing: a change in the net radiant energy at the top of the atmosphere. Model response is normally interpreted as expressing the model’s sensitivity to radiative forcing, but the forcings themselves vary among models through a mixture of error, deliberate choices, and model-specific behavior. However this diversity arises, the interpretation of climate model simulations has been made more difficult because model-dependent radiative forcing is not well-known.
Responding to this challenge, the investigators participating in this proposal are among those organizing a Radiative Forcing Model Intercomparison Project (RFMIP) to be associated with the next round of CMIP experiments. RFMIP will assess sources of error and spread in the radiative forcing calculated by state-of-the-art climate models for historical and future climate studies. RFMIP consists of three linked components: one assessing the forcing by greenhouse gases, one the forcing by natural and anthropogenic aerosols, and one linking these with estimates of effective forcing inferred from global model integrations via careful diagnosis.
Each component relies on calculations and analysis beyond the capabilities of most modeling centers which are only now enabled by the supercomputing Leadership Class Facilities developed by the DOE. These include very large-scale reference calculations for forcing by greenhouse gases and aerosols using computationally-intensive line-by-line radiative transfer models, careful and consistent diagnosis of aerosol radiative properties, and the systematic construction of a set of radiative kernels useful for diagnosing forcing, including one for each model and one based on observations. This proposal seeks funding for these lynchpin activities.
The results of these calculations would provide benchmarks for the two largest sources of direct forcing against which model calculations can be compared, forming the basis for metrics of climate model performance in this crucial area.