Differences between Well-Forecast and Poorly-Forecast Bow Echo Events in the WRF

Convection-Allowing Models (CAMs) allow us to simulate small-scale structures of convective systems very similar to those observed, such as outflows, updrafts, and bowing segments. However, the timing, location, and details of mode often have some errors. In particular, prior studies have found that bow echoes may be simulated more poorly than other modes and are in need of further study.

To gain further insight into the causes for poor forecasts of bow echoes, cases have been simulated using the WRF with 3 km horizontal grid spacing and two different microphysics schemes (Thompson and Morrison 2-moment), over a domain of 1800 x 1800 km. Some cases were chosen where the forecasts were very poor, some where the model simulated the bow echo well, and some where runs using one microphysics scheme did well but runs using the other performed poorly. The focus of the study is to understand and determine causes for problems in the model's ability to simulate the bowing such as the synoptic environment and how well it is initialized, deficiencies in simulation of stratiform regions, and shortcomings in simulation of mesoscale circulations associated with the convection.

Multiple environmental parameters are being analyzed and compared to the 20 km RUC and RAP observational datasets. These include different low level and deep layer vertical wind shears, MUCAPE, wind speeds, and ambient moisture. In addition, cold pool characteristics will be compared among the configurations and cases. The Method for Object-Based Diagnostic Evaluation (MODE) also will be used to compare forecasts to the observations and compute different system attributes.

Preliminary tests have examined 0-6 km and 0-3 km vertical wind shear, MUCAPE, 500 mb wind speeds and have not found noticeable differences between well predicted and poorly predicted environments. In addition, these parameters do not appear to differ more from observations prior to bow echo formations in poorly predicted events than in well predicted events. Work is ongoing to perform more quantitative evaluation of any differences along with smaller-scale comparisons focused on the near-storm circulations, making use of Doppler velocity data.