648 Wildfire Duration Model for Air Quality Forecast System

Wednesday, 9 January 2013
Exhibit Hall 3 (Austin Convention Center)
Jong-Jae Lee, Pusan National University, Busan, Korea, Republic of (South); and H. C. Kim, C. H. Kim, F. Ngan, A. F. Stein, and P. Lee

Handout (990.6 kB)

Wildfire duration model for Air Quality Forecast, using satellite-based fire detection and assimilated meteorological condition measurement, has developed. In order to forecast the impact of wildfire emissions to regional air quality, accurate and prompt detection of wildfire events with respect to both timing and location is crucial. Satellite detection of active wildfires is one of the most efficient ways of retrieving wildfire information, and is widely used for fire impact forecast systems. Such systems, however, have inherent weaknesses since satellite products have finite latency. Even the fastest product (e.g. MODIS Rapid Response Fire Product) does not provide turely realtime information, and does not postulate information into the future.. To overcome this incapability, many systems apply natural decaying rates for the longevity of fire events due to climatology, but such methods cannot deal with rapid changes of meteorological conditions. We have utilized a multi-year satellite-detected-fire data set (e.g. Hazard Mapping System (HMS) and Fire Radiative Power (FRP) data from MODIS and GOES) and meteorological information from North American Mesoscale Model (NAM), to develop a wildfire duration model to predict how long current fire events, detected by satellite, can last for a given meteorological condition and land type information. Based on previous studies on Fire-Weather correlation, such as the Haines Index (HI) and Canadian Fire Weather Index (FWI), we improved model's performance by adding more information from satellite and land type information. The goal of this study is to build a practical methodology to predict (1) wildfire occurrence probability, and (2) duration of wildfire events under given meteorological and geographical conditions. The former has been studied by modelers for a long time, yet much uncertainty remains. The latter is a more robust and straightforward problem since we know the occurrence and circumstantial conditions of the wildfire events. and is more useful information for air quality forecast systems. Moreover, hind-cast and process analyses of occurred fires can effectively identify deficiencies of the model and allow focused effort for improvement. Applications of the model will be tested and verified for various cases.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner