Remote Sensing of Survey Line Burn

Justin Epting

The Normalized Burn Ratio (NBR) is an index developed for burn severity mapping using Landsat satellite imagery. This algorithm was originally developed in Glacier National Park, Montana but has not been thoroughly tested in Alaska. The algorithm is useful for detecting changes in the landscape due to the loss of vegetation and moisture. These changes are also associated with wildland fire, thereby making the algorithm useful for mapping burned areas. We are currently testing the accuracy of this algorithm on a fire that burned in the summer of 2001. The fire, known as the Survey Line Fire, occurred southwest of the Bonanza Creek Experimental Forest, and affected more than 45,000 hectares of land. The goal of this project is to test the results of the NBR algorithm against ground control data collected in the summer of 2002. The ground control data include estimates of vegetation killed and consumed, the loss of the litter and duff layers, the amount of charring on trees, and the amount of regeneration following the fire. These ground data are serving as the reference points by which to validate the NBR algorithm. Using Landsat ETM+ images from June 16, 2001(pre-fire) and July 21, 2002 (post-fire) we have computed the NBR change image. The change image was then categorized into burn severity classes based on previously published threshold values. Based on preliminary analyses, the burn severity classes appear to correspond well with post-burn ground data. The NBR algorithm has currently been compared against one other algorithm, the Normalized Difference Vegetation Index, and has produced better results. We are currently performing a quantitative accuracy assessment, whereby we will compare the NBR to several other common algorithms used in burned area mapping.