Evaluation of Cloud Microphysical Schemes in a Simulated Warm Frontal Snowband During the GPM Cold Season Precipitation Experiment (GCPEx)

This study evaluates the ability of newer, more advanced bulk microphysical parameterizations (BMPs) within the NASA-Unified Weather Research and Forecasting (NU-WRF) modeling system in simulating the evolution of a warm frontal band on 18 February 2012 during the Global Precipitation Mission Cold-season Precipitation Experiment (GCPEx) in Ontario, Canada.  Field measurements highlighted the unique nature of this band where large snow aggregates rapidly transitioned to small graupel particles as a layer of supercooled cloud water developed within the band.  Model forecasts vary greatly for these mixed-phase precipitation events due to the different assumptions and parameterizations utilized by the microphysical schemes. The GCPEx measurements help identify weaknesses and biases in these assumptions and parameterizations, which ultimately can lead to overall improvements in the BMPs and model forecasts.  

For this study, four different NU-WRF simulations are conducted using the same model setup and configuration with the exception of the cloud microphysics scheme which varies between the following: single-moment NASA Goddard 4ICE, single-moment Stony Brook (SBU-YLIN), double-moment Morrison, and the double-moment Predicted Particle Properties (P3).  We use a triple nested domain configuration down to 1-km grid spacing centered over the field study site. Model output for the inner most nest of each simulation is evaluated against in situ aircraft and ground-based measurements during GCPEx.  The P3 scheme is able to simulate the abrupt transition from large snow aggregates to small graupel-like particles that was shown in the field measurements.  Conversely, the Morrison scheme continues to predict large snow aggregates while generating minimal graupel amounts during this transition period.  The single-moment schemes, Goddard 4ICE and SBU-YLIN, predict much higher total ice mass concentrations in the post-frontal region due to the assumptions required in diagnosing the size distribution intercept parameter, which leads to development of precipitating ice-phase particles and the release of instability well behind the developing frontal band.  Thus, less instability is available for band development in the single-moment simulations, which helps lead to a weaker, poorly organized band.  Our sensitivity simulations reveal that a major reason for the stronger, more organized band in the double-moment P3 simulation is related to the development of heavily rimed, dense particles (e.g., graupel) that undergo limited melting and evaporational processes along the warm frontal band.  Conversely, the other schemes predict much higher snow concentrations along the frontal band with slower fall speeds that allow for considerable melting to occur.  The P3 also predicts limited evaporational cooling due to the explicit calculation of droplet condensation and evaporation.  The other schemes use a saturation adjustment parameterization that can lead to droplet evaporation artifacts near cloud edges.