GIS Modeling of Forest Cover Distributions in a Southern Appalachian Watershed

This study develops a method to model forest cover distributions in the Little Tennessee River Watershed in Macon County, North Carolina using a geographic information system (GIS). The model employs common environmental variables, e.g. elevation, aspect, and slope, and a newly developed metric called a topographic index. The topographic index quantifies landscape form to reflect the degree of site convexity or concavity. To generate index values, the GIS sums relative aspect differences between all adjacent cells and the center cell in a neighborhood analysis. To determine predominant community types and their environmental ranges in the study area, I surveyed canopy trees, small trees, and saplings on 216 10 x 10 m plots located throughout the watershed.

TWINSPAN identified eight community types: Birch-Hemlock, Basswood-Birch, Tuliptree, Red Oak-Hickory, White Oak-Black Oak, Chestnut Oak, Chestnut Oak-Red Oak, and Chestnut Oak-Black Oak. All communities showed strong association with the topographic index. Most showed strong association with elevation and a general preference for either N/NE/E or S/SW/W aspects. No overall relationship with slope was apparent. Field and GIS values for elevation and aspect correlated well, but slope data differed markedly. A discriminant analysis identified topographic index, elevation, and aspect as significant predictive variables of forest cover types. Classification success depended on the number of forest cover types in the analysis. Two cover types produced a success rate of 84%; four produced 64.8%, six produced 48.8%, and eight produced 40.8%. Several factors could account for the decrease in successful classification including inadequacy of the topographic index, spatial resolution of the GIS data, the clusters produced by TWINSPAN, the need for additional significant predictor variables, or the failure to account for disturbance regimes. Overall the topographic index provides a quantitative, objective, and reliable measurement of landscape form. Further research should increase its usefulness and applicability to other studies.