Evaluation of Shallow‐Cumulus Entrainment Rate Retrievals Using Large‐Eddy Simulation

AbstractA complete and quantitative understanding of cumulus entrainment remains elusive, in part due to the difficulty of directly observing cloud entrainment rates. Multiple approaches to ground‐based observational retrieval of bulk fractional entrainment rates (ε) within cumuli have been developed, such as the parcel model by Jensen and Del Genio (JDG, 2006, https://doi.org/10.1175/JCLI3722.1) and Entrainment Rate In Cumulus Algorithm (ERICA) by Wagner et al. (2013, https://doi.org/10.1175/JTECH-D-12-00187.1). In this paper, a new cumulus entrainment retrieval based on a turbulent kinetic energy (TKE) similarity theory is presented. This method estimates ε based on only the environmental and subcloud conditions. By conducting large‐eddy simulations of a range of continental and maritime shallow cumulus convection cases as Observing System Simulations Experiments, the first numerical verification of the three retrieval methods is produced. These simulations consider a broad range of shallow cumulus environments along with variations of the numerical configuration. The diagnosed ε from these simulations is found to be robustly larger in cumuli over the ocean than in cumuli over land. For continental cumuli, the experiments also reveal a diurnal cycle with increasing ε in the late afternoon. These diagnosed ε serve as the “truth” against which the pseudo‐retrieved entrainment rates from several different implementations of each retrieval are verified. Overall, the simpler JDG and TKE retrievals outperform the more sophisticated ERICA method and better capture the sensitivity to continentality. Only the TKE method reproduces the diurnal variations in ε within continental cumuli. The mean error in the ε retrievals are between 20% and 30% for the TKE and JDG methods, but 50% for ERICA.