Assessing Biases and Climate Implications of the Diurnal Precipitation Cycle in Climate Models

Abstract

The diurnal cycle is a common benchmark for evaluating the performance of weather and climate models on short timescales. For decades, capturing the timing of peak precipitation during the day has remained a challenge for climate models. In this study, the phase and amplitude of the diurnal precipitation cycle in Coupled Model Intercomparison Project (CMIP) models are compared to satellite data. While some improvements align CMIP6 models closer to satellite observations, significant biases in the timing of peak precipitation remain, especially over land. Notably, precipitation over land in CMIP6 models still occurs ∼5.4 h too early; the diurnal cycle amplitude is ∼0.81 mm day⁻¹ too small over the oceans. Further, the diurnal phase of oceanic precipitation correlates weakly with the equilibrium climate sensitivity in CMIP6 models: models with a later precipitation peak over oceans tend to exhibit a higher climate sensitivity. However, it is unclear whether this relationship is robust.

Plain Language Summary

Rainfall on Earth is not uniform in time and occurs preferentially at certain times of the day. Surface and space-based observations have been extensively used in previous research to characterize the diurnal cycle of precipitation, which peaks in the late afternoon and early evening over land and in early morning over oceans. Atmospheric simulations tend to produce precipitation too early in the day even when mean fields are in line with observations, indicating fundamental physical aspects of the atmosphere are improperly represented. In this work, outstanding diurnal cycle biases in the latest generation of Coupled Model Intercomparison Project (CMIP) models (CMIP5 and CMIP6) are documented and quantified. While marginal improvements are made, CMIP6 models still precipitate too early over both land and ocean.