Further Oklahoma field evaluations of PB-Piedmont, a mesoscale model for nighttime smoke dispersion

Prescribed fire is becoming an increasingly used tool in Oklahoma for fuels management, with one to two million acres annually being burned. For purposes of smoke management, it is especially important to know where smoke may go during the nighttime. If a reliable forecast tool could be developed, fire managers could, for example, forgo a burn on a given day if it looked like sensitive areas would be smoked out overnight. Besides adverse health effects, poor visibility is also a concern and can contribute to motor vehicle accidents.

PB-Piedmont is a mesoscale numerical model that was originally developed to simulate and predict nighttime smoke movement near the ground over terrain characteristic of the Piedmont of the southeastern United States. The current model simulates flows (e.g., drainage, synoptic) over complex terrain and utilizes digital elevation data to 30-m resolution. It is initialized with hourly synoptic weather observations and can run with either hourly synoptic data (in a historic, post-burn mode or “nowcast” mode) or with hourly output from the MM5 model (in a forecast mode). In the Oklahoma version being tested, PB-Piedmont produces image files of smoke location every 15 minutes through the period the model is run; screen images are updated every 10 minutes.

Through a research agreement between Oklahoma State University and the Southern Research Station of the USDA Forest Service, PB-Piedmont is now being tested in Oklahoma to see if it is applicable to local landscapes outside the Piedmont region, where it is currently being used operationally. Much of Oklahoma, with the exception of the plains of the western areas and the mountains in the southeast, has terrain similar to that of the Piedmont.

A number of field evaluation studies will be presented from prescribed burns in Oklahoma during the period 2007 through 2009. With the aid of a GPS unit, “smoke” or “no-smoke” observations were taken by the first author during the nighttime hours by driving local roads surrounding and within the prescribed burn areas. PB-Piedmont was later run in the post-burn, historical mode with hourly synoptic weather data. Comparisons of smoke locations predicted by the model were then made with the actual smoke observations from a given location and time. In this presentation, animated maps of modeled smoke distribution will be shown as well as comparisons with field observations at specific times. Results from these and former field studies indicate that while PB-Piedmont under some conditions performs well in Oklahoma, there are also multiple instances where smoke is predicted and none is observed, and vice-versa. Probable causes for these discrepancies, some of which have to do with model design limitations, will be discussed.