Development of an Objective Spot Forecasting Tool Using a Mesoscale Model

The state of Florida has led the nation with respect to wildland fire management for many years. In order to more accurately deal with the threat from wildfire, as well as manage the largest open burning program in the country, Florida has designed a GIS based wildland fire management system (Florida Fire Management Information System, FFMIS). One component of this system is the use of a mesoscale model in providing forecast fire danger parameters. In addition, the state of Florida has developed a tool to provide spot weather forecasts from the model output. This aspect of the project hopes to fill the void in weather information available to the prescribed fire community due to the National Weather Service's lack of spot weather support for nonfederal prescribed fires.

Spot weather forecasting provides some special challenges for numerical weather prediction. The daily variability in the locations requiring spot forecasts limits the application of traditional MOS techniques as reliable statistics cannot be compiled for every site in the model domain. To resolve the small scale processes that influence local weather conditions requires high resolution which must be balanced against operational time constraints in forecast dissemination. The level of detail required in a spot forecast is much greater than that of a general forecast; how the temperature varies throughout the day is as important as the expected high temperature.

Hourly model output is compared to observations for seven sites around Florida for a 20-day period in May of 1999. A 7-day running mean is used to remove the mean hourly bias from the model output for each variable (temperature, relative humidity, wind speed and direction). Removal of an hourly bias rather than a daily mean bias helps improve the forecast's diurnal variation.