Evaluating the potential of the normalized burn ratio and other spectral indices for assessment of fire severity in Alaskan black spruce forests

This thesis presents an assessment of the potential of the differenced Normalized Burn Ratio (dNBR) and other spectral indices for mapping fire severity in Alaskan black. Using simple linear regression, the dNBR derived from Landsat TM and ETM+ data was correlated with ground measures of fire severity including the Composite Burn Index (CBI), depth of the organic soil remaining after the fire, reduction in the depth of the organic layer, and Canopy Fire Severity Index; these being measures of fire severity used to assess the ecological effects of fire. Regression analyses yielded weak correlations: the highest R2 for a comparison between the dNBR and CBI was 0.52, p<0.0001. However, the mid-infrared ratio showed higher potential than other spectral indices in many comparisons. Overall, these results indicate (1) validation of the dNBR is needed and (2) burn severity mapping schemes which are more comprehensive than the dNBR should be developed.