Reforecasts of a 2004 Elevated Convection Event Misforecast by the Eta Model

During the period 24-25 July 2004, elevated convection produced rain totaling up to 5 inches across east-central Missouri. No rain was forecast by the 12 km Eta model over this area. The presence of two concurrent mesoscale convective systems rendered this event a particularly challenging one for numerical weather prediction. An analysis of the physical differences between the model forecast and the observations is initially presented. Deficiencies in the model initial condition and problems inherent in the parameterization of convection are highlighted. Over the past 6 years, advances in numerical weather prediction and computing power have enabled the development of more sophisticated numerical weather prediction forecasts. Therefore, numerous locally-run forecasts have been generated in an attempt to determine if the original Eta forecast can be improved upon. First, three forecasts conducted on the same grid as the Eta are analyzed to understand the effects of changing the model core and the convective precipitation scheme. The results do not offer substantial betterment over the 2004 Eta forecast, but some improvement is noted. Next, two runs at the convection-allowing grid spacing of 4 km have been generated, with the microphysics scheme varying. Both runs result in more substantial forecast improvement, although some deficiencies remain. Finally, a run with grid spacing of 2 km has been generated. Results indicate that the more realistic simulation of smaller-scale features does not lead to quantitative precipitation forecasts that are notably better than those from the much less computationally expensive 4 km runs. Corresponding Author: Martin A. Baxter Dept. of Geology and Meteorology, Central Michigan University 314 Brooks Hall, Mt. Pleasant, MI 48859 E-mail: [email protected] REFORECASTS OF A 2004 ELEVATED CONVECTION EVENT MISFORECAST BY THE ETA MODEL