TITLE: Variability of low clouds and drizzle: a tale of two frameworks
Modern microphysics schemes in climate models typically account for subgrid variability by assuming probability distribution functions (PDFs) of process rate inputs such as cloud and rain water mixing ratios (qc and qr, respectively). These PDFs can be obtained from a number of sources: in situ observations, ground- or space-based remote sensing, or fine scale numerical simulations such as large eddy simulation (LES). It is appealing to use model output to generate these distributions because they generate large sample sizes, they can simulate a variety of cloud regimes/case studies, and they are subject to neither the measurement ambiguities of remote sensing products nor the sampling uncertainties of in situ observations. Despite the appeal of using such simulations for parameterization development, it has never been shown that LES can satisfactorily reproduce the observed spatial structure of the cloud and rain fields.
In this talk, I will compare the structure of observed and modeled qc and qr of a drizzling stratocumulus field using multifractal analysis, an approach that quantifies variability across spatial scales. The statistics of simulated qc closely match in situ and ground-based remote sensing estimates of the multifractal parameters H1 and C1, which measure the smoothness and intermittency of a field, respectively. There are major discrepancies between observed and simulated qr properties, though, with simulated qr consistently displaying the multifractal properties of observed clouds (smooth, minimally intermittent). This result is robust across a range of domain sizes, cellular organization (i.e. open vs. closed cells), the choice of model numerics and different bulk scheme formulations. These results suggest a fundamental limitation of bulk schemes to produce sufficiently intermittent and rough precipitation, which in turn implies that simulations of warm rain with bulk microphysics are likely unable to reproduce the observed heavy-tailed nature of precipitation.
The intermittency of precipitation can be enhanced by allowing more degrees of freedom in the evolution of the drop size distribution (DSD) with a bin microphysics scheme. Yet even with the enhanced complexity of explicitly resolving the DSD, drizzle remains overly smooth, which may point to a systematic limitation of the Eulerian approach to microphysics. I end by briefly discussing the advantages of using multifractal