Examining rain formation via drop coalescence in super-droplet models
Figure 1. Vertical cross sections through the cloud center from a three-dimensional large-eddy simulation of a precipitating cumulus congestus cloud using particle-based “super-droplet” microphysics. Left panels show cloud water content and right panel show rain water content (color contours). Black lines show vertical air velocity with contour lines starting at 3 m/s and increasing every 3 m/s. Three different times are shown in the top, middle, and lower panels as indicated.
Image by Hugh Morrison, NSF NCAR
Figure 2. Timeseries (mean +/- 1 st. dev.) of domain total a) cloud mass, b) rain volume, c) rain mass, and d) cloud base rain flux from large-eddy simulations of a cumulus congestus cloud using a stochastic coalescence method (SDM) and two deterministic methods that do not consider stochastic fluctuations in drop collisions (LSdSDM and mLSdSDM). Results show that rain initiation is substantially delayed using the deterministic LSdSDM and mLSdSDM schemes compared to the stochastic SDM.
Image by Hugh Morrison, NSF NCAR
An emerging tool in cloud microphysics modeling represents clouds and raindrops by computational particles called “super-droplets” that evolve in the modeled flow. Recent studies have documented that different super-droplet models produce widely varying predictions of rain. Understanding these differences is an important step in the wider adoption of these models by the community and their use for developing bulk microphysics schemes in Earth system models.
This study examined different methods for representing how drops collide and coalescence in super-droplet models, which is the main mechanism for rain initiation in warm clouds. It also investigated the sensitivity of rain formation to the number of computational particles (super-droplets) in the model. This study highlights that stochastic coalescence, leading to the generation of “lucky drops” that collide with other drops much more frequently than the average, is critical for overall rain production in warm clouds.
Various methods for representing drop collision-coalescence in a “super droplet” model were examined in this study in the context of simulating rain initiation in warm clouds. To separate the impacts of stochastic coalescence from variability in the modeled flow, the “piggybacking” method was employed, which uses a single dynamical flow realization from a large-eddy simulation of a cumulus congestus cloud in all microphysics tests. It was shown that the timing of rain initiation was insensitive to the number of super-droplets in three-dimensional cloud simulations. However, rain formation was substantially delayed using a coalescence method that limits random variability in droplet collisions. It was also shown that excessive coalescence variability using the super-droplet model is not important in practice, given that microphysical variability is overwhelmed by flow variability when the microphysics is allowed to feed back to the dynamics (i.e., in runs without piggybacking).