A comparison of WRF-simulated radar information to observations for a well-forecast derecho event

A damaging derecho wind event occurred across portions of central Iowa during the night of July 18-19, 2010, resulting in a long swath of surface winds of 30-35 m/s. A WRF simulation using 4 km horizontal grid spacing initialized at 12 UTC July 18 and using the Thompson microphysics scheme performed surprisingly well at showing a rather small-scale but intense bowing convective system moving south-southeastward across central Iowa roughly 18 hours into the model forecast. Near-ground winds in the simulation exceeded 30 m/s with reflectivities above 70 dBZ peaking in intensity near the melting level. The strongest simulated winds were not immediately behind the bowing echo but instead roughly 20-30 km further behind, just southwest of a comma head feature. Radar observations from the National Weather Service WSR88D in Johnston, IA, depicted somewhat similar features with a bowing system, although the model was too fast with the south-southeastward movement of the system. Model displacement errors in key radar and wind features were around 100 km over a several hour period when the system was most intense. Because of the rapid movement of the system, timing errors at the locations experiencing the worst damage were just over one hour, with the model too soon to bring in the system. Near-ground winds (less than 1000 ft above ground) observed by DMX radar exceeded 35 m/s, matching many of the reports in Storm Data at the surface. The strongest winds, however, occurred along and just behind the bowing echo, a somewhat different result from the simulations. This poster presentation will focus on a comparison of the simulated radar quantities to those observed, and also on some sensitivities of the simulation to grid spacing. This is a unique case to examine because some of the features simulated by WRF that would have the most serious impacts on the public were forecast very well. The study will examine if simulated radar characteristics also resembled observations well.