Evaluating cloud microphysics schemes in nested NMMB forecasts

As part of the ongoing development at NCEP/EMC of the Nonhydrostatic Multiscale Model on the B grid (NMMB), which is planned for implementation into the operational North American Mesoscale (NAM) model for the National Weather Service (NWS) next year (see Rogers et al., this conference), this paper will focus on assessing the impacts of grid-scale microphysics changes on nested NMMB forecasts, and in particular quantitative precipitation forecasts (QPF) from the higher-resolution nests. Comparisons were made between runs using the Ferrier and WSM6 microphysics based on 4-km CONUS forecasts nested within a larger 12-km domain roughly half the size of the current operational NAM. The WSM6 microphysics was selected because of its use in high-resolution NSSL WRF runs during recent SPC Spring Programs, the relatively small QPF bias seen during several years of runs at NSSL, and its computational efficiency compared to other WRF microphysical packages that include graupel. Preliminary assessments of changes to both microphysics packages were made using single-column test runs with prescribed vertical velocity forcing (B. Shipway, UKMO), from which runs that produced more promising results were then tested in a series of experimental NMMB forecasts using initial and boundary conditions from the NAM. While the forecast skill, as measured from various objective verification statistics (including QPF), was similar between both schemes, the Ferrier microphysics runs were ~30% faster. Encouraging results were seen when modifying the Ferrier microphysics to account for larger rain drop sizes, increasing the assumed number concentrations of cloud droplets and precipitation ice particles, increasing the fall speeds for rimed ice (especially as the density of the particles approach heavily rimed graupel and frozen drops), and using a new formulation for the autoconversion of cloud droplets to rain (Liu et al., 2006). Although the forecast improvements from these microphysics changes have been relatively small, additional experiments will be conducted with the most promising configuration to be incorporated into real-time parallel NMMB runs prior to final testing later this year.