Testing ecoregions in Kentucky and Tennessee with satellite imagery and Forest Inventory dataThis article is part of a larger document. View the larger document here.
Ecoregions are large mapped areas of hypothesized ecological uniformity that are delineated subjectively based on multiple physical and biological components. Ecoregion maps are seldom evaluated because suitable data sets are often lacking. Landsat imagery is a readily available, low-cost source of archived data that can be used to calculate the normalized difference vegetation index (NDVI), which is associated with seasonal and annual vegetation conditions. This paper reports on a study designed to test the use of NDVI as an integrating variable for detecting differences among mapped ecoregions and determine if NDVI was associated with biological components within ecoregions using USDA Forest Service Forest Inventory and Analysis (FIA) field data. Published 1.1 km resolution georeferenced NDVI imagery was obtained at the beginning (June 22) and end (September 23) of the summer growing season for 11 years (1989-1999). Public domain GIS software (WinDisp4) was used to determine NDVI values at 5,399 georeferenced, forested plot locations in Kentucky and Tennessee that are periodically inventoried by FIA. Plots were grouped by ecoregions of various scale and tested for significant differences. Analysis of variance revealed significant (P< 0.001) differences in mean NDVI between the two macro-scale ecoregions (divisions) and among most of the next lower ecological units (provinces). Regression analysis indicated that NDVI was associated (P< 0.01) with season of sampling, elevation, and forest stand basal area of the inventory plots. At the ecoregion analysis scale, NDVI values consistently increased with higher elevation and forest basal area at the plot locations. We concluded that NDVI determined at FIA plot locations has potential for testing differences among ecoregions. Satellite imagery has often been used to determine classification accuracy within small mapped ecosystems; results from our study suggest that imagery may also be used to test for hypothesized overall differences between large mapped ecosystems.