An automatic algorithm to detect vegetation fires globally from NOAA-AVHRR data
Abstract
A contextual algorithm for fire detection using NOAA-AVHRR was developed at the Natural Resources Institute (NRI). Unlike ‘traditional’ fire detection algorithms applied to NOAA data (e.g. multi-channel thresholds), the decision to record a fire is made by comparing a potential fire pixel with the pixels in its immediate neighbourhood. The procedure automatically extracts values in channels 3 and 4 neighbourhood pixels, which are then compared to those of the potential fire pixel. The algorithm is self-adaptive and therefore is a very consistent method over large areas as well as through seasons, without the need to change the thresholds. The algorithm was successfully applied in several areas of the world. This paper describes the approach chosen, compares it with traditional’ techniques’, and analyses advantages and drawbacks in light of some examples.
Citation
Ceccato, P.; Flasse, S.; Downey, I. An automatic algorithm to detect vegetation fires globally from NOAA-AVHRR data. EARSeL Advances in Remote Sensing (1996) 4 (4) 84-89.
Links
An automatic algorithm to detect vegetation fires globally from NOAA-AVHRR data