Hidden Warming Effects of Cloud Cycle Errors in Climate Models

Clouds' efficiency at reflecting solar radiation and trapping the terrestrial one is strongly modulated by their diurnal cycle. Much attention has been paid to mean cloud properties due to their critical role in climate projections; however, less research has been devoted to their diurnal cycle. Here, we quantify the mean, amplitude, and phase of the cloud cycles in current climate models and compare them with satellite observations and reanalysis data. While the mean values appear to be reliable, the amplitude and phase of the diurnal cycles of clouds show marked inconsistencies, especially over land. We show that, to compensate for the increased net radiation input implied by such errors, an overestimation of the cloud liquid water path may be introduced during calibration of climate models to keep temperatures in line with observations. These discrepancies are likely to be related to cloud parametrization problems in relation to atmospheric convection.