9.4 SILHS: A Monte Carlo interface between clouds and microphysics

Larson, Vincent, Carsten Harlass, and Jan Hoft, University of Wisconsin

In regional simulations with coarse grid spacings, it may be beneficial to account for subgrid variability in microphysical calculations. In contrast, most microphysics schemes in WRF are fed grid-mean inputs by WRF and implicitly assume that grid boxes are microphysically uniform. Subgrid variability information, such as cloud fraction, could be embedded directly into each of WRF’s microphysics schemes. However, including subgrid information in all of WRF’s microphysics schemes would require substantial code revisions.

An alternative approach is to develop a general interface that feeds information on subgrid variability from WRF’s cloud parameterizations into any of WRF’s microphysics schemes. To this end, we have developed the Subgrid Importance Latin Hypercube Sampler (SILHS). It is designed to work with parameterizations of subgrid probability density functions (PDFs). SILHS draws multiple sample points from a subgrid PDF, feeds those sample points into a microphysics scheme, averages the resulting microphysical tendencies, and feeds the averaged tendency back into WRF’s dynamical core. A disadvantage of SILHS is that it is computationally expensive. An advantage is that it is non-intrusive; that is, it accounts for subgrid variability without requiring modifications to the microphysics code.

We have simulated the VOCA case of marine stratocumulus clouds off the coast of South America. In this simulation, the use of SILHS increases liquid water path. We find that monthly averaged quantities are fairly insensitive to the number of sample points chosen per grid box and time step.