Coupling microphysics parameterizations to cloud parameterizations

It is difficult for numerical modelers to accurately interface microphysical processes, such as drizzle formation, with cloud properties, such as cloud water content. Because both drizzle and cloud water vary on small scales, both must be parameterized. Furthermore, drizzle and cloud water interact, and therefore information about cloud water must be fed into the drizzle parameterization, and vice-versa. Simply inputting the grid box average cloud water is insufficient, since drizzle formation depends nonlinearly on cloud water. Rather, one must input information about the distribution of cloud water within the grid box. Some authors have speculated that the failure to do so has led to the need to re-tune microphysics parameterizations when they are implemented in separate large-scale models of different grid box size.

We have coupled a boundary layer cloud parameterization to a drizzle parameterization in a way that accounts for sub-grid variability in cloud fields. The cloud parameterization is based on the ``assumed PDF" method. In this method, the shape of a smoothly varying family of PDFs is assumed. (We assume a mixture of Gaussians.) A particular PDF is selected from the family for each grid box and timestep. This provides the information on subgrid variability that we need to input into the drizzle parameterization. We use the drizzle parameterization of Khairoutdinov and Kogan. We integrate their formulas analytically over the PDF in order to compute grid-box-average drizzle processes.

The combined cloud/drizzle parameterization has been used in single-column mode to simulate the DYCOMS-II RF02 marine stratocumulus case. We find that we can emulate the drizzle profiles of a 3D LES that also uses the Khairoutdinov-Kogan drizzle parameterization.