Remote sensing of burn severity and the interaction between
burn severity, topography, and vegetation in interior Alaska

Justin Epting

M. S. Thesis
July 2004
University of Alaska Fairbanks

ABSTRACT

A variety of single-band, band ratio, vegetation index, and multivariate algorithms were evaluated for mapping burn severity using Landsat TM and ETM+ imagery across four burns in interior Alaska. The Normalized Burn Ratio (NBR) outperformed all algorithms, both when tested as a single post-fire value and when tested as a differenced (prefire postfire) value. The NBR was then used to map burn severity at a historical burn near Yukon-Charley Rivers National Preserve and a time-series of images from 1986 to 2002 was analyzed to investigate interactions between vegetation, burn severity, and topography. Strong interactions existed between vegetation and burn severity, but the only topographic variable that had a significant relationship with burn severity was evelation, presumably due to the strong control of elevation on forest cover. The highest burn severity occurred in spruce forest, while the lowest occurred in broadleaf forest. Areas with high burn severity experienced disproportionately more shifts toward spruce woodland and shrub classes, while areas with low to moderate severity were less likely to change vegetation type. Finally, vegetation recovery, estimated using a remotely-sensed vegetation index, peaked between 8-14 years post-fire, and recovery was highest for areas with the highest burn severity.