During the late winter of 2003, FSL produced a set of six mesoscale model runs four times daily to support a project sponsored by the Federal Highways Administration. This project, called the Maintenance Decision Support System, is intended to provide a tool to snow removal equipment garage supervisors to assist in determining when and where snowplows should deployed, and what chemical treatments should be applied. The FSL model runs were transmitted in real time to NCAR/RAP, where they were ingested into the Road Weather Forecast System. RWFS produces point forecasts of pavement temperature and chemical concentration along roadways, and then applies encoded rules of practice to suggest plow and treatment plans. The system was demonstrated in cooperation with the Iowa Department of Transportation, specifically three garages in Ames and Des Moines.

The ensemble consisted of three different mesoscale models (MM5, RAMS, and WRF) using two different lateral bounds models (Eta and AVN), for a total of six members. The models were configured with identical grid size, resolution, and geometry, centered on Iowa. The four-per-day execution schedule was driven by the update frequency of the lateral bounds models as provided by NCEP.

This plan reflects the requirements as put forth by the client to focus on the 12-24 hr forecast, for which it was expected that lateral bounds would be an important controller of forecast quality. However, experience gained during the 2003 Demo indicated that greater value from the forecasting system could be gained by concentrating on the earlier hours of the forecast, the 3-12 h range. Thus, a different modeling strategy will be used for the 2004 Demo.

One of the goals of this project is to package the software components of MDSS for release to public and private entities which provide weather services to state Departments of Transportation. Practical aspects of the ensemble modeling component such as affordability, portability, and reliability will be addressed.