Forest Fire Detection With NOAA/AVHRR Data

Forest fire is always detected using NOAA/AVHRR and MODIS data. The main principle of fire detection is based on the Plank radiation theory that the object with high temperature radiates more energy in mid-infrared band than in thermal infrared band, i.e., the mid-infrared band is more sensitive to high temperature than thermal infrared band. Currently, three major forest fires detection algorithms were proposed, they are the fixed threshold method, the analysis of neighborhood pixels and the brightness temperature combined Normal Difference Vegetation Index (NDVI).This thesis tries to propose a method of the fixed threshold combined NDVI to detect some forest fire which happened in recent years in China. We also analyses the fire points, the smoke, the scar area and so on. The results show that the proposed method is feasible, and the characteristics of this method were also summarized.The contents of each chapter are arranged as follows:Chapter 1 presents the development of forest fire and fire detection by remote sensing in China and abroad, and gives the outline of this thesis.Chapter 2 gives the summary of principles and methods of fire detection by using NOAA/AVHRR data, and introduces the fire detection method of fixed threshold combined NDVI.Chapter 3 uses the fire detection method of fixed threshold combined NDVI to detect some forest fire which happened in recent years in China, and discusses the feasible of this method and its characteristics.Chapter 4 gives the summary and discusses the related problems should be solved in future.