Evaluating Interannual Variability of the TOA Energy Budget in CMIP5

Understanding Earth's energy budget is fundamental for studying the climate system, since a change in climate is controlled by (1) the budget of energy fluxes entering and leaving at top-of-atmosphere (TOA) and (2) the distribution of the energy remaining in the system. Global climate models serve as our most sophisticated tools for simulating present-day climate and predicting future changes; a model's ability to accurately recreate Earth's energy budget is often used as a benchmark for model quality. Recent high-quality measurements of radiative fluxes using NASA's Clouds and the Earth's Radiant Energy System (CERES) satellite instrument provide an excellent 14-year data record for model evaluation. This study focuses on the evaluation of the widely-used Coupled Model Intercomparison Project Phase 5 (CMIP5) ensemble of climate models with performance metrics related to the interannual variance of key radiation budget quantities. Variance within the climate system is generated by physical processes—such as clouds and surface conditions—and is therefore a measure of model response to a forcing; variance-related performance metrics are used in this study to evaluate the fundamental physical processes that underpin model projections. It is found that most CMIP5 models reproduce interannual variance and probability distributions of TOA radiation reasonably well compared with CERES observations, and models from the same modeling center tend to behave similarly across the tested performance metrics. However, all models fail at reproducing certain observed conditions, such as the regression between TOA longwave radiation fluxes and surface temperature; this metric represents a fundamental measure of atmospheric column energy processes. We explore the various related radiative processes to identify those which models are least able to recapture, highlighting possible avenues for improvement in the next generation of climate models.