Forecasting warm season precipitation systems remains a difficult problem for numerical weather forecasts. It is hoped that advances in observation systems and in analysis techniques will improve model initial conditions and subsequent forecasts. In this paper, we examine the impact of variational data assimilation on MM5 forecasts of mesoscale convective systems. Comparisons of initial fields produced by variational and standard objective analyses will be presented. Additionally, the impact of data sources will also be examined. These data include wind profilers, high-resolution GOES winds, GOES temperature and moisture soundings, and GPS precipitable water data.

The first case chosen for study is an MCS that occurred over eastern Nebraska early on 30 May 2000 that produced damaging winds and heavy rain and which was forecasted poorly by the NCAR/MMM real-time forecast system. Preliminary results show that given identical input data, the variational initialization results in a slightly improved forecast over a forecast using the conventional objective analysis.