Diagnosing drivers of tropical precipitation biases in coupled climate model simulations

In this study, we analyse the large-scale biases in tropical precipitation climatology of two different types of coupled ocean–atmosphere general circulation models (GCMs): the coupled model intercomparison project (CMIP) models and ICON-Sapphire, a global storm-resolving model. We employ the simple globally resolved energy balance (GREB) diagnostic precipitation model to evaluate four drivers of the precipitation biases in the simulations: surface specific humidity, surface relative humidity, tropospheric mean and variability in the vertical motion. The tropical precipitation biases in the CMIP and ICON-Sapphire simulations are surprisingly similar in their patterns and also in the elements forcing them. The results of our analysis using the GREB model show that the precipitation biases result from both biases in the sensitivity to the four forcing fields and biases in the simulated forcing fields themselves. The most significant bias for both, the CMIP and ICON-Sapphire simulations, is a too high sensitivity to the mean vertical circulation ( \({\omega }_{\text{mean}}\) ${\omega }_{\text{mean}}$ ) and bias in the \({\omega }_{\text{mean}}\) ${\omega }_{\text{mean}}$ climatology itself. This also holds for specific long-standing biases, such as the double intertropical convergence zone problem. Meanwhile, biases in the climatology of specific and relative humidity play only a minor role but contribute to an overall small increase in precipitation in CMIP models that may be related to the “drizzling” bias. These results can give insights to the modelling community regarding model development, and illustrate that the GREB diagnostic precipitation model applied is a good tool for evaluating the drivers of large-scale tropical precipitation.