Does Increasing Horizontal Resolution Improve the Simulation of Intense Tropical Rainfall in GFDL's AM4 Model?

We examine tropical rainfall from the Geophysical Fluid Dynamics Laboratory's Atmosphere Model version 4 (GFDL AM4) at three horizontal resolutions of 100 km, 50 km, and 25 km. The model produces more intense rainfall at finer resolutions, but a large discrepancy still exists between the simulated and the observed frequency distribution. We use a theoretical precipitation scaling diagnostic to examine the frequency distribution of the simulated rainfall. The scaling accurately produces the frequency distribution at moderate‐to‐high intensity (≥10 mm day−1). Intense tropical rainfall at finer resolutions is produced primarily from the increased contribution of resolved precipitation and enhanced updrafts. The model becomes more sensitive to the grid‐scale updrafts than local thermodynamics at high rain rates as the contribution from the resolved precipitation increases. Plain Language Summary: State of the art global scale climate models have horizontal resolutions of the order of tens of kilometers. However, these resolutions are much lower than the scales required to resolve tropical convection. This study investigates whether a resolution increase from 100 km to 25 km results in notable improvements in simulating tropical rainfall. Higher resolution simulations capture more intense rainfall events that are missed by their coarser counterparts. However, they struggle to capture the accurate frequency distribution of intense rainfall events. Results reported in this study underscore the importance of scrutinizing and carefully interpreting the outcomes of high‐resolution climate model simulations. Key Points: Tropical rainfall distribution is analyzed using simulations from the Geophysical Fluid Dynamics Laboratory's Atmosphere Model version 4 at three horizontal resolutions: 100 km, 50 km, and 25 kmIncreasing horizontal resolution yields more intense tropical rainfall but not the accurate frequency distributionTheoretical precipitation scaling accurately captures the frequency distribution of the simulated precipitation at moderate‐to‐high intensity [ABSTRACT FROM AUTHOR]