A fast algorithm for the automatic recognition of heat sources in satellite images

Abstract

A fast algorithm has been implemented for the automatic recognition of heat sources using data from the Advanced Very High Resolution Radiometer (AVHRR) of the TIROS-N series of meteorological satellites. Channel 3 of the AVHRR measures radiation in the mid-infrared part of the spectrum (3.8 µm wavelength) and is sensitive to high temperature heat sources even when these occupy a very small part of the instantaneous field of view. The use of channel 3 of the AVHRR for the detection of heat sources is well documented but to date these heat sources have been identified by operator interpretation of hard copy or other display media. The algorithm described obviates the need for manual interpretation or display facilities. Attention is given to the avoidance of spurious results arising out of the problem of pixel dropout and other non-systematic noise.

The algorithm has important implications in that it facilitates the detection of wildfires and agricultural burning and the monitoring of gas disposal operations on a routine and cost-effective basis. Known heat sources, such as oil and gas production sites and refineries, are able to be monitored continually, while maverick heat sources can be reported in near-real time. This is particularly useful where fire reporting services are poorly developed.