U.S. Department of Agriculture: Forest Service -- National Agroforestry Center


Date of this Version



American Society for Photogrammetry and Remote Sensing Annual Conference Proceedings (2004) CD-ROM. p. 96-104.


Preliminary results are presented from ongoing research on spatial variability of fire effects on soils and vegetation from the Black Mountain Two and Cooney Ridge wildfires, which burned in western Montana during the 2003 fire season. Extensive field fractional cover data were sampled to assess the efficacy of quantitative satellite image-derived indicators of burn severity. The objective of this study was to compare the field burn severity measures to the digital numbers used to produce Burned Area Reflectance Classification (BARC) maps. Canopy density was the field variable most highly correlated to BARC data derived from either SPOT Multispectral (XS) or Landsat Thematic Mapper (TM) imagery. Among the other field variables, old litter depth and duff depth correlated better with the satellite data than did old litter cover. Ash cover correlated most poorly. Old litter cover correlated better with the satellite data than did exposed mineral soil or rock cover, but combining the mineral soil and rock cover fractions into a single inorganic cover fraction improved the correlation to a comparable level. Most field variables, with the notable exception of ash, tended to vary more at low and moderate severity sites than at high severity sites. Semivariograms of the field variables revealed spatial autocorrelation across the spatial scales sampled (2 – 130 m), which the 20 m or 30 m resolution satellite imagery only weakly detected. Future analyses will be broadened to quantify burn severity characteristics in other forest types and to consider erosion processes, such as soil water infiltration following fire.