Super typhoon Haiyan makes landfall
Super typhoon Haiyan
Subscribe here to receive special images


Lake Maggiore - Italy, 3-6 April 2000

Drawbacks and advantages of the VEGETATION and AVHRR instruments for burnt area detection in Northern Australia.

D. Stroppiana1, M. Maggi1, D. Graetz2, S. Campbell2, I. Balzer2, J-M. Grégoire1 and J.M.C Pereira3.
1 JRC-Space Applications Institute, Ispra, Italy
2 CSIRO-Earth Observation Centre, Canberra, Australia
3 Instituto Superior de Agronomia, Lisbon, Portugal

Until recently, the detection and mapping of burnt areas from low resolution satellite data, has been done exclusively with the imagery provided by the ATSR and AVHRR instruments. The VEGETATION instrument is now opening new perspectives for the detection of burnt areas.

In order to assess these perspectives and to investigate the specific advantages of the VEGETATION instrument in a multi-sensors approach, burnt area maps were derived from NOAA-AVHRR and SPOT-VEGETATION images for a study site in Northern Australia for the dry season in 1999. Daily data acquired from both sensors were composed into 10 day images using a two steps compositing criterion for AVHRR (minimum channel two and maximum channel four) and a single step criterion for VGT (minimum NIR). A burnt area detection algorithm for each sensor was then derived using a Classification Trees approach. The algorithm was directly applied to the VGT composites to classify them into burnt/unburnt. For what concerns the AVHRR data set, active fire maps produced by the World Fire Web were used as inputs in a seed-growing approach. The burnt area detection algorithm derived for AVHRR is applied starting from the fire points and let them grow untill no neighbouring pixels satisfy the burnt conditions.

Finally Landsat TM and ATSR images were used for a first evaluation of the results.

Preliminary conclusions are: 1)the lack of a thermal channel on the VEGETATION instrument is certainly a drawback, specially during the compositing process 2)the excellent geometry of the VEGETATION imagery, compared to the AVHRR one, allows to improve the performance of a burnt area algorithm based on a change detection approach 3)the MIR channel improves the detectability of the burnt areas 4)the "seed-growing" method, which combines active fire locations and burnt area algorithm, seems very promising and could be the basis of a combine VEGETATION-AVHRR approach to the detection and mapping of burnt areas at global level.