Investigation of precipitation processes with RAMS and observations

Poster PDF

Description

Fundamental understanding of cloud microphysical processes is critical for improving model parameterizations and remote sensing observations, as well as observing global trends in precipitation. Observations often lack the continuity to be able to directly pinpoint and diagnose atmospheric processes. Similarly, although cloud resolving models can provide diagnostics of microphysical processes, they are difficult to validate directly with observations, which are often spatially and temporally irregular. Toward the goal of relating surface raindrop size distributions to storm processes, we have developed a new framework based on principal component analysis (PCA) which captures the primary modes of variability of surface precipitation. Using this framework, we have analyzed a large catalogue of simulations from the Colorado State University (CSU) Regional Atmospheric Modeling System (RAMS) for comparison with an extensive database of disdrometer and radar observations. Simulations have been carefully subsampled by cloud morphology and structure (convective, stratiform, shallow, deep) to reflect similar distributions to those in the disdrometer dataset. From this comparison, we find that although RAMS is able to reproduce the bulk DSD variability seen in the observations, critical differences point to shortcomings in some microphysical parameterizations. For example, the first two principal components from observations are uncorrelated, but are highly correlated in RAMS. This is found to be related to an abundance of mean drop sizes at around 1 mm, which is not pronounced in the disdrometer observations. Nonetheless, we utilize the microphysical process rates from the simulations to contextualize the DSD groups found in observations toward the ultimate goal of inferring storm processes from long-standing surface observations.