Improvement of quantitative precipitation forecasts using a new microphysical parameterization

Accurate prediction of the amount and spatial distribution of precipitation remains a difficult forecasting problem, particularly in regions of complex terrain. Substantial errors in precipitation forecasts result in part from inadequate representation of microphysical processes in mesoscale atmospheric models. Better formulation of the microphysical processes is especially important for predicting precipitation in winter mountain storms, where errors in riming rates, mean terminal velocities, and trajectories of the falling snow can result in serious over- and under-prediction of precipitation amounts on windward and leeward mountain slopes, respectively.

As a first step in developing a new microphysical parameterization for mesoscale models, we have implemented a new scheme in a cloud-resolving model EULAG. Existing double-moment ice microphysical schemes for mesoscale models generally treat ice crystal nucleation explicitly whereas the ice crystal size is inferred based on the information on nucleation and ice water content (IWC). These schemes also contain prognostic equations for IWC for various categories of hydrometeors such as cloud ice, snow and graupel. The two moment microphysical scheme employed in this work differs in both of these aspects.

Instead of treating nucleation explicitly, which introduces large uncertainties into predicted microphysical fields, the new scheme employs a prognostic equation for the slope parameter of the size distribution, which is a measure of the ice particle size. From the known size distribution and predicted IWC, the ice particle number concentration is computed using mass conservation. The scheme is based on Mitchell's snow growth model (SGM) which has realistically predicted evolution of ice size spectra in nine case studies of cirrus and frontal clouds. This suggests that using a prognostic equation for the slope parameter, rather then ice particle number concentration, can greatly reduce uncertainties arising from assumptions about the nucleation process. Our approach also eliminates the need for ice hydrometeor categories, since a single size distribution is evolved through all growth processes (vapor diffusion, aggregation and riming). Since the ice particle number concentration is determined through mass conservation, various cloud ice categories are not needed for determining precipitation. Due to its improved treatment of riming and fall speeds, the new scheme implemented in mesoscale models should improve the forecast accuracy for the amount and spatial distribution of precipitation as well as demonstrate the sensitivity of snowfall rates to changes in droplet size spectra (which depend on aerosol concentration).

Snowfall rates obtained employing the new microphysical parameterization are compared to the observed rates and the rates predicted using the standard microphysical schemes in simulations of the Sierra Nevada wintertime orographic precipitation for selected winter storms documented during the Sierra Cooperative Pilot Project(SCPP).