Research that incorporates GIS and remotely sensed imagery has become increasingly popular and important for large-scale environmental applications, such as generating land use and land cover maps. One of the critical aspects of land cover analyses is assigning a land use and land cover classification scheme. This research evaluated two classification schemes, the 2002 Natural Heritage Classification and the 1976 James Anderson System in a land cover analysis of the Canandaigua Lake Watershed using Landsat imagery. It was hypothesized that the Landsat imagery could be used to identify unique ecological communities such as those delineated by the Natural Heritage Classification. A composite image, created from an August 15, 2003 Landsat image using bands 1, 3 and 5, was used for the fine cluster analysis, which produced 38 unique clusters. Using the Canandaigua Lake Watershed Council's land use and land cover map as a truth image (26 single NHC classes and 14 mixed NHC classes), the clustered Landsat image was used in an unsupervised classification analysis that resulted in generalized land use and land cover maps using the Natural Heritage and James Anderson Classification schemes (5 and 6 dominant land covers respectively). Because many clusters were associated with several land cover classes, two to three training sites were identified for each land use and land cover in the truth image and a supervised classification analysis was performed. The revised maps produced eight land use and land covers in both the Natural Heritage and James Anderson Classification Systems. In both approaches, dominant land covers, such as residential, deciduous forest and cropland, were identified but spatially small ecological communities were absorbed, such as wetlands, shale talus slope woodlands, and shrublands. The analysis did, however, suggest that Landsat imagery could be used to help identify ecological communities when prior knowledge of the area is available (e.g. Appalachian Oak Hickory ridgelines), which may help focus ground-truthing efforts. The cluster image may also suggest areas where delineations may need revisions due to the complex mixture of clusters within delineated polygons. Suggestions for future research include incorporating elevation, terrain and seasonality as additional variables (surrogate bands) as part of the supervised and unsupervised classification analyses.
Library of Congress Subject Headings
New York Natural Heritage Program; Land use--Classification; Ecology--Classification; Watershed ecology--New York (State)--Canandaigua Lake Watershed; Land use surveys--New York (State)--Canandaigua Lake Watershed
Department, Program, or Center
Thomas H. Gosnell School of Life Sciences (COS)
Money, Travis, "A land use and land cover accuracy assessment based on Landsat 7 imagery within The Canandaigua watershed: Natural Heritage Program and The James Anderson Classification System" (2004). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus