Evaluation of Contextual AVHRR Hotspot Detection Algorithms for Australian Conditions
Geoscience Australia currently operates and maintains the MODIS-based Sentinel bushfire monitoring system used on a daily basis by fire managers to plan their activities. Geoscience Australia plans to improve the service through the addition of hotspot information from the Advanced Very High Resolution Radiometer (AVHRR) sensors onboard the National Oceanic and Atmospheric Administration (NOAA) satellites. This will provide hotspot information on a greater frequency than is currently available by using MODIS data only and will provide redundancy for the MODIS instruments. This paper evaluates the performance in Australian conditions of three contextual algorithms for the detection of hotspots from AVHRR data. The algorithms we implemented were from the papers by Justice et al. (1993), Flasse and Ceccato (1996) and Giglio et al. (1999). After initial tests, Giglio algorithm has been selected for further development and validation using suitable AVHRR data from October 2005 for a study area that covered parts of Queensland, the Northern Territory and Western Australia. Results from this study showed that Giglio algorithm captured on average 85% of the hotspots. This algorithm is currently being integrated with the Sentinel system to provide hotspots from the AVHRR sensors to the user community by November 2006.