Development and Testing of a Stochastic Representation of Ice Microphysical Properties in WRF

Poster PDF

Description

Ice microphysics strongly impact cloud radiative forcing and precipitation generation, and interact with the cloud dynamics through latent heating and condensate loading. However, their representation in models is uncertain, particularly because of the wide variety of ice particle shapes and characteristics. Most current microphysical schemes represent the range of ice properties using a few different categories that have typical ice particle characteristics of a given ice type (e.g., small ice, snow, graupel). Important physical properties like the mass-size (m-D), projected area-size (A-D), and fallspeed-size (V-D) relationships are assumed to follow power laws that are held fixed in time and space for each category. However, observations show large natural variability of these parameters that is not clearly correlated with local microphysical or environmental conditions (e.g., ice water content, relative humidity, temperature) that are predicted by models. To incorporate this parameter variability into models and assess its impacts, a new stochastic parameterization is being developed. The initial focus is on stochastically varying the m-D “a” and “b” parameters (m = aD^b) in the Predicted Particle Properties (P3) microphysics scheme. Two key issues addressed are the co-variability of the "a" and "b" parameters, and their spatiotemporal autocorrelation. Initial tests using the idealized two-dimensional squall line case in the Weather Research and Forecasting model (WRF) will be presented, including the impact of varying the spatiotemporal autocorrelation scale (note that if the scale is equal to or larger than the domain size, then the parameters are fixed in time and space as in traditional schemes). To assess the impacts in a less idealized setting with observational validation, WRF simulations of the 20 May, 2011 Mid-latitude Continental Convective Clouds Experiment (MC3E) squall line using P3 with both fixed and stochastically varying "a" and "b" parameters are performed and compared with one another and radar observations. This initial development and testing sets the stage for more detailed development later that will constrain the stochastic modeling approach using in situ and remote-sensing microphysical observations.