Evaluation of WRF in Simulating Multi-Band Precipitation Structures Within Northeast U.S. Winter Storms

Northeast U.S. winter storms can exhibit several different precipitation structures within the cyclone comma head: (1) a relatively large primary band, (2) multi-band (no primary band), (3) both single and multi-bands simultaneously, (4) cellular convection, and (5) stratiform only (no bands). It has been shown that mesoscale models when run at high resolution (< 5 km grid spacing) can realistically simulate primary bands. However, an evaluation of a mesoscale model to simulate finer-scale multi-bands is lacking. This study determines the ability of the Weather Research and Forecasting (WRF) model to resolve multi-bands or a combination of single and multi-bands. The bands must have a minimum radar reflectivity threshold of 30 dBZ and persist for at least two hours to be included in the dataset. Approximately 15-20 winter storm cases that meet these criteria were selected from the cool seasons of 2012 to 2016. Radar reflectivity from the WSR-88D radar network at six sites along the Northeast U.S. coast was used to select the cases as well as verify the WRF precipitation structures. Additional thermodynamic and kinematic verification was obtained from several sounding locations and the Climate Forecast System Reanalysis (CFSR). The WRF was nested down to 1.33-km grid spacing within two outer (12-km and 4-km) domains covering the eastern and the northeast U.S., respectively. Select cases are run down to 400 m grid spacing. The WRF was run using 45 vertical levels and used the same physics as the operational HRRR so the the results are consistent with that operational model. The WRF physics include the Thompson microphysics, MYNN2 planetary boundary layer scheme, and the Grell-Freitas convective parameterization only in the 12-km domain. The WRF was initialized using the Rapid Refresh (RAP) analyses every 3-hours. Preliminary results suggest that the WRF often struggles to predict the presence of multi-bands. For all events the probability of detection for predicting these bands is quantified as well as an evaluation of the various band structures (size, number, and longevity).  Cases with relatively large versus small errors are separated and compared with soundings and the CFSR analyses for various band ingredients (frontogenesis, stability, and moisture amount/depth) in order to determine which factor(s) may be limiting some of the predictability of these bands.