Low-level clouds such as shallow cumulus and stratocumulus, prevalent over the trade-wind regions and the cool upwelling waters off the west coast of continents, play a key role in cloud-climate feedbacks. Low cloud feedbacks are known to be a key source of uncertainty in climate prediction, and the ones showing the largest inter-model spread of all feedbacks in climate models. This is primarily due to major challenges in the correct representation of turbulence, convection and cloud processes in climate models combined with multiple low cloud feedbacks that may offset each other. In this project, we aim to contribute to a more realistic physical representation of low-level clouds in the Simple Cloud-Resolving E3SM Atmosphere Model (SCREAM).

SCREAM is a global convection-permitting model based on the E3SMv1 model with a target global resolution of 3.25 km, allowing for a more accurate representation of complex mesoscale deep convective dynamics. Small-scale planetary boundary layer turbulence and shallow convection still need to be parameterized, which in SCREAM is accomplished through the Simplified Higher-Order Closure (SHOC). SHOC is based on the assumed probability density function (PDF) method but follows a simplified approach as the second-order moments needed to construct the PDF are diagnosed instead of prognosed. Global SCREAM simulations show a more realistic representation of certain aspects of low-level clouds and the cloudy PBL in general. However, several issues still remain to be solved even at these horizontal resolutions (e.g., accurate representation of shallow cumulus). Some of these issues are likely related to limitations of assumed PDF closures, such as SHOC, in representing exceptionally skewed distributions of vertical velocity and thermodynamic quantities typical of cumulus clouds.

In the last few years, the lead PI’s group has been developing a new framework designed to address some of the issues of assumed PDF closures. In this context, the stochastic moist multi-plume MF scheme has been coupled to SHOC (SHOC+MF) in SCREAM and promising results from single-column model (SCM) simulations for benchmark cases of shallow convection suggest that the multi-plume MF approach offers a physics-based and cost-effective complement to SHOC due to its ability to represent extremely skewed values of vertical velocity and thermodynamic properties not well captured by the assumed PDFs.

In this project, we will implement SHOC+MF into the full three-dimensional version of SCREAM. We will first evaluate SHOC+MF using the Doubly Periodic SCREAM version for various cloud regimes (forced by intensive observation period files and/or reanalysis) versus ARM observations of low-level clouds, reference large-eddy simulation (LES) output, and relevant satellite data. We will improve and calibrate specific aspects of SHOC+MF to realistically represent the full range of convective regimes in the cloudy boundary layer—with a special focus on the representation of shallow cumulus and the transition from stratocumulus to cumulus. For the validation of the full three-dimensional SCREAM version with the SHOC+MF parameterization, we will compare SCREAM simulations with satellite observations for a variety of critical variables related to low-level clouds.