The January 2017 event featured a large band of snowfall exceeding 10” across Nebraska, South Dakota, and Iowa. WRF was used to simulate the event at 4 km grid spacing, with an output frequency of 10 minutes. 10 microphysics schemes were evaluated, with all other physics options held constant. The selected microphysics schemes calculate reflectivity within each scheme, ensuring consistency between the microphysics in the model and the reflectivity, in contrast with post-processing approaches which may introduce additional ambiguity.
Though the microphysics schemes may produce similar amounts and distributions of precipitation over the course of an event, differences in the way each scheme simulates hydrometeor evolution and calculates reflectivity will produce differences in reflectivity at any given time. For example, the median of the domain-integrated 1 km simulated reflectivity over 24 hours was 16% higher when the WDM6 scheme was used in comparison with the Thompson scheme, while the domain-integrated accumulated precipitation with the WDM6 scheme employed was only 4% higher than the precipitation produced when the Thompson scheme was used. This work further documents the reasons for the differences between reflectivity and precipitation for all microphysics schemes employed. Results suggest that particular care should be used when comparing reflectivities from models run using different microphysics schemes, such as within NCEP’s High Resolution Ensemble Forecast system, as the relationship between simulated reflectivity and precipitation will not be the same for each scheme.