A NEW METHOD FOR FIRE DETECTION BASE ON SOIL MOISTURE INDEX

Fire plays a major role in destroy of forest. If fire was discovered in initial hours, can be taken under control and a large area of natural resource can be protected. But fires are not largely monitored and enough detection facilities are not available. Traditional fire detection algorithms mainly rely on hot spot detection. They use thermal infrared (TIR) channels with fixed or contextual threshold that Based upon the Temperature Sensitivity of 4 and 11 micron. This work presents a new method based on combination of Giglio (2003) method and use of tow sequence MODIS images. The principle of the approach is to detect fires by comparing the observed TOA brightness temperature of the pixel with the expected temperature during fire conditions. The expected temperature is modeled by series of observations and the end product is a thresholding algorithm which is masked by Soil Moisture Index. The satellite detected fire pixels were validated with ground data collected by Iran’s The Forests, Range and Watershed Management Organization (FRWMO). The result illustrated that in this algorithm the sensitivity to detect fires increased .The proposed method was applied on fifty case studies. Result demonstrates that this algorithm can detect fires in initial hours of burning and all of fires detected correctly and in some cases we had one or two false alarms.