Extraction of Active Fire Line and Map Using Aviris Imagery

Fire managers who respond to wildland fire outbreak will need intelligent information regarding the state of the fire such as location, direction and active fire perimeter for better fire management. The Infrared Interpreters, who visually examine the infrared imagery to provide intelligent information, usually undertake the task of drawing the active fire map by hand. Vegetation indices like Normalized Difference Vegetation Index (NDVI) and Normalized Difference Burn Ratio (NDBR) have only been used to calculate and map vegetation characteristics, ascertain portions of an area that have been burned thereby providing burn severity classification. These indices have the potentiality of being extended to extract active fire line and fire map but have not been exploited. In this study, we propose a technique of extending their use to extract active fire line as well as active fire map. Image processing tools like multiband image gradient and image dilation were developed to extract active fire line and fire map using AVIRIS imagery. The extracted fire line was refined by B-spline curve fitting and dominant fire pixels obtained from the multi-band image gradient. The smooth curve may be used as an input for a model that estimates direction of propagation of fire and also for visualization of the evolution of active fire front. We show results of extracted active fire line and fire map with San Bernardino Mountain Fire, AVIRIS imagery of 1996.