Predictability of Snow Multi-Bands in the Cyclone Comma Head Using a 40-Member WRF Ensemble

Extratropical cyclones can exhibit multiple fine-scale precipitation bands in the comma head, referred to as multi-bands; however, their prediction in high resolution NWP models has not been quantified. The first goal of this study is to determine how accurately the Weather Research and Forecasting (WRF) model can simulate multi-bands. The second objective is to determine what processes may be missing in the model if the band formation is poor.

For this study, a 40-member WRF ensemble was constructed with an outermost domain (18-km grid spacing) centered on the Mid-Atlantic region, and nested subdomains with 6-km and 2-km grid spacing. The initial (ICs) and lateral boundary conditions (BCs) for the outermost domain are from the first five perturbed members of the GFS Ensemble Reforecast (GEFSR). These members are combined with different planetary boundary layer (PBL) schemes (MYNN2 and YSU) and microphysical schemes (Thompson and Morrison) for 20 members, while the other 20 members applied a stochastic kinetic energy backscatter scheme (SKEBS) and stochastic perturbation to physical tendencies scheme (SPPT) while retaining the MYNN2 and Thompson schemes. The ensembles for all cases were initialized at 0000 UTC, and the model verification is for the time window of 1200 UTC to 0600 UTC the following day (forecast hours 12-30), except for one case for which multi-bands persist to forecast hour 36.

A feature-based verification is applied to the WRF ensemble for four snow events over the Northeast U.S. from 2014 through 2017 using hourly WSR-88D radar reflectivity from the KBOX, KGYX, and KOKX sites at the scan time closest to the top of the hour. The radar reflectivity was interpolated to a horizontal plane of 2-km height above sea level. This same level is used for the simulated reflectivity. The Method for Object-Based Diagnostic Evaluation (MODE) tool within the NCAR Model Evaluation Tools (MET) software package is used for identifying bands. Multi-bands are defined as two or more bands that are 5-20 km in width and that also exhibit a >2:1 aspect ratio. The RAP analysis and upper-air sounding observations from corresponding sites (KCHH used for KBOX) are used for verification.

Preliminary results suggest that WRF underpredicts the number of multi-bands, with probability of detection (POD) between 0.19 and 0.49 and false alarm ratio (FAR) between 0.04 and 0.24. Despite the generally poor scores, WRF shows some ability to simulate the shape, orientation angle, and location of multi-bands. In order to investigate the processes with these bands, WRF members were then grouped into time periods of 2 hours and 1 hour before genesis, genesis, 1 hour and 2 hours after genesis, dissipation, and times with no multi-band activity within an hour before or after. For each of these subsets, spatial plots are produced to show what proportion of WRF member-hours in that subset contain 1) potential instability, 2) conditional instability, 3) inertial instability, 4) potential symmetric instability, and 5) conditional symmetric instability. Various thresholds of mid-level frontogenesis are also checked. These results will be discussed during the presentation.