Mapping and inventory of forest fires from digital processing of tm data

Abstract

The application of space‐borne sensors to forest fire mapping and inventory was evaluated. Digital image processing of Thematic Mapper data was used to study a big forest fire on the Mediterranean coast of Spain. The results showed that image processing techniques cannot discriminate perfectly the area affected by the fire. The main problem was spectral overlapping between burned and unburned vegetation especially caused by the sparseness of shrub and confusion with other cover types such as urban/rural villages. Most of these problems can be solved by using visual analysis for masking the affected area, although this strategy is not completely reliable on studies of small forest fires. An inventory of the damaged area was performed in three levels: total area affected by the fire, area by previous vegetation species, and identification of levels of damaged on the burned vegetation.