Accurate forest fire forecasts are important for protecting the lives and property of residents and fire fighters over the western United States, as well as other parts of the country and world. Mesoscale models can play an important role in fire-weather prediction because they provide forecasts at high spatial and temporal resolution, which is particuarly important over regions of fine-scale topography. At the NOAA Cooperative Institute for Regional Prediction (CIRP) at the University of Utah, we are using a real-time version of the Weather Research and Forecast (WRF) model that provides high resolution forecasts at 12.5 km grid spacing for the western United States. This effort seeks to improve WRF forecasts, including those relevant to fire-weather prediction, prior to its implementation as an operational model at the National Centers for Environmental Prediction (NCEP) in 2005.

The poster will present an analysis of WRF performance during the summer 2003 fire-weather season. The evaluation will be based on observations provided by the Mesowest cooperative network (http://www.met.utah.edu/mesowest), which collects data from over 3000 stations over the western United States, including Remote Automated Weather Stations (RAWS) that are positioned to support fire-weather monitoring and forecasting. In addition to an evaluation of the accuracy of standard surface sensible weather parameters, the ability of WRF to predict events that are critical for fire-weather forecasting will be examined. For example, how well does WRF predict conditions that contribute to red flag warnings, such as 2-m relative humidity below 35% for 4 or more hours, or 2-m relative humidity below 35% for any duration and 10-m wind speed greater than 15 mph? Factors that contribute to poor model forecasts will also be evaluated and discussed.