What controls drizzle initiation? Insights from a comparison of large-eddy simulations with observations

Drizzle occurs frequently in shallow, warm boundary layer clouds. For example, in stratocumulus it occurs approximately 1/3 of the time in full cloud cover conditions (Wood 2012). Drizzle affects moisture and energy budgets, and cloud albedo, morphology and lifetime. At the cloud scale, processes that control drizzle formation include turbulence production via radiative cooling and/or shear, entrainment, and surface moisture fluxes. At the micro-scale, collision-coalescence is the primary process relevant to warm drizzle formation. Differential gravitational sedimentation and turbulent air motions cause cloud droplets to collide, creating drops much larger than can be formed by condensation alone. Other factors, such as preferential concentration and entrainment mixing may also be relevant. The process is typically subdivided into three regimes: autoconversion (small drops self-collide), accretion (large drops collect small drops), and hydrometeor self-collection (large drops self-collide). Of these regimes, autoconversion is the rate-limiting step in existing analytical representations.

This study (i) evaluates whether our best theoretical understanding of collision-coalescence in the autoconversion regime can replicate observations, with a broader goal of (ii) exploring which cloud-scale factors are most important for drizzle initiation. A state-of-the-art turbulent collisional growth model is applied to a bin microphysics scheme within a large-eddy simulation such that the full range of cloud drop growth mechanisms are represented (i.e. CCN activation, condensation, collision-coalescence, mixing, etc.) at realistic atmospheric conditions. We compare cloud drop spectra produced by the LES with observations to assess the quality and limits of our theoretical knowledge. The comparison will be performed over a range of observational cases that span a range of drizzle rates. These cases differ in their radiative cooling rates, shear, cloud-top temperature and moisture jumps, and entrainment rates. Using these diverse cases, we will begin to tease apart the cloud-scale factors governing drizzle rates. Initial results for question (i) suggest that in some cases enhancements of 1 to 2 orders of magnitude over predicted collision rates are necessary to reproduce observations.

Back to: Warm Boundary Layer Clouds and Climate Change From the Cloud- to the Global Scale II