Discussion on the Method of Removing Abnormal Hot Point in Forest Fire Remote Sensing Monitoring Based on Vegetation Index

Abstract

Based on the MODIS NDVI from 2001 to 2015 and slope data, the surface classification of Chongqing was extracted which is divided into forest land, grassland, cultivated land, mixed land (agriculture, forestry and grass), construction land. There include two key fire prevention areas, one is forest land, grassland, the other is forest land, grassland, mixing land(woodland, grassland, agriculture).The method of distinguishing non-fire hot points was established based on the surface classification and historical RS hot points. Superimpose the hot points with the surface classification, and remove the points fall outside the two key fire zones, extract the points falling in key fire zones. The results showed hot points in the area of key fire prevention was significantly less than that original monitoring points. 5.1% and 15.88% hot points by MODIS fell in the first and second key fire prevention area respectively. 58%, and 68.1% hot points by FY3 fell in two fire prevention area respectively. 53.7% and 73.37% hot points by NOAA fell in fire prevention area respectively. Using surface classification data, hot points in non-forest fire prevention areas can be effectively excluded.