Analysis of subgrid cloud variability in a year-long CRM simulation over the ARM SGP

Poster PDF

Description

General circulation models (GCMs) predict cloud cover fractions and hydrometeor concentrations only in discrete vertical layers where clouds are assumed to be horizontally homogeneous in a coarse grid. They do not explicitly specify vertical geometric associations or horizontal optical variations of clouds. Subsequently, clouds within a GCM grid are simulated as a single effective volume that impacts radiation using various vertical overlap assumptions. The parameterization of cloud vertical overlap and horizontal inhomogeneity in the radiation schemes of GCMs has been a long-standing challenge. The inclusion of subgrid cloud variability in the radiation calculation for GCMs requires the knowledge of cloud distribution under different climate regimes, which is not yet available from observations. The year-long cloud-resolving model (CRM) simulation forced with the ARM large-scale forcing provides a unique data set to document the characteristics of cloud horizontal inhomogeneity and vertical overlap and to evaluate and represent their effects on the radiative fluxes and heating rates over a GCM grid. The analysis of inhomogeneity parameter defined as the ratio of the logarithmic and linear average of cloud liquid and ice path distribution shows that relatively larger inhomogeneous clouds occur in summer and spring than in fall and winter. The inhomogeneity parameter also shows a large variation in the vertical. Significant radiative effects of cloud inhomogeneity are quantified by the diagnostic radiation calculation with horizontally homogeneous clouds in comparison with the CRM simulations. To account for the cloud inhomogeneity effect, the approach used in some GCMs is by scaling the cloud water paths with a constant reduction factor. The evaluation of this approach using the CRM simulations suggests that the parameterization with the vertically varied reduction factors represents the radiative effects of cloud inhomogeneity better than a constant reduction factor. Three cloud overlap assumptions (i.e., maximum, minimum, and random overlap assumptions) are currently used in GCMs. The analysis using the vertical profile of CRM cloud fractions indicates that none of these assumptions is able to reproduce the total cloud fraction. The maximum overlap assumption systematically underestimates the total cloud fraction, while the random and minimum overlap assumptions systematically overestimate the total cloud fraction.