The development of an automated algorithm to map fire severity from satellite imagery: tropical savannas northern Australia.

Fire severity is the post-fire effect of fire on the vegetation. The fire severity mapping algorithm developed in this study correlated helicopter-based spectra collected over a site using a hand held spectrometer and ground data describing the fire severity within the spectrometer field of view. The differenced Normalized Burn Ratio (∆NBR) quite clearly distinguished between severe and not-severe fires (r 2 = 0.94). However, further discrimination into three or more classes required the development of other indices incorporating the region of the spectrum represented by MODIS band 6 (1628-1652 nm). This poses problems operationally as band 6 on Aqua is dysfunctional thus halving the available data.