Analysing Threshold Value in Fire Detection Algorithm Using MODIS Data

Abstract

MODIS instruments have been designed to include special channels for fire monitoring by adding more spectral thermal band detector on them. The basic understanding of remote sensing fire detection should be kept in mind to be able to improve the algorithm for regional scale detection purposes. It still gives many chances for more exploration. This paper describes the principle of fire investigation applied on MODIS data. The main used algorithm in this research is contextual algorithm which has been developed by NASA scientist team. By varying applied threshold of T4 value in the range of 320-360K it shows that detected fire is significantly changed. While significant difference of detected FHS by changing ΔT threshold value is occurred in the range of 15-35K. Improve and adjustment of fire detection algorithm is needed to get the best accuracy result proper to local or regional conditions. MOD14 algorithm is applied threshold values of 325K for T4 and 20K for ΔT. Validation has been done from the algorithm result of MODIS dataset over Indonesia and South Africa. The accuracy of MODIS fire detection by MOD14 algorithm is 73.2% and 91.7% on MODIS data over Sumatra-Borneo and South Africa respectively.