Small-area fires detection and evaluation using multi-spectral NOAA data

摘要

The paper discusses the experience of creating a geoinformation technology for the use of AVHRR data with spatial resolution of 1.1 km for detection and evaluation of characteristics for fires the linear dimensions of which are by several orders lower than a pixel of image. The technology is based on the effective algorithm of detecting small area brightness anomalies, which in some statistical sense differ from the surrounding background. The process of making a decision on the presence or absence of a signal is based on the methods of the statistical estimation theory. The Neumann-Pirson criterion is used as the initial detection principle, which provides the detection with the constant frequency of false alarm. The evaluation of the area of the fire zone is carried out on the basis of complex interpretation of mathematical modelling results for the field of the image brightness, optimal evaluation of the signal amplitude and geodata. The developed information technology allows detecting fire zones the area of which is 10~(-4) of the area of image pixel.