Scholarly
    Communications Project


Document Type:Master's Thesis
Name:David Dean Morton
Email address:dmorton@vt.edu (dave_morton@usgs.gov)
URN:1998/00821
Title:LAND COVER OF VIRGINIA FROM LANDSAT THEMATIC MAPPER IMAGERY
Degree:Master of Science
Department:Fisheries and Wildlife Sciences
Committee Chair: Robert H. Giles, Jr.
Chair's email:rogiles@vt.edu
Committee Members:James Fraser
James B. Campbell
Keywords:satellite remote sensing, wildlife habitat, 1992 land cover of Virginia
Date of defense:June 12, 1998
Availability:Release the entire work immediately worldwide.

Abstract:

Knowledge of land cover is important in a variety of natural resources applications. This knowledge becomes more powerful within the spatial analysis capabilities of a geographic information system (GIS). This thesis presents a digital land cover map of Virginia, produced through interpretation of 14 Landsat Thematic Mapper (TM) scenes, circa 1991-1993. The land cover map, which has a 30m pixel size, was produced entirely with personal computers. Hypercluster aggregation, an unsupervised classification method, was used when hazy and mountainous conditions were not present. A haze correction procedure by Lavreau (1991) was used, followed by a supervised classification on coastal areas. An enhanced supervised classification, focusing on topographic shading, was performed in the mountains. Color infrared photographs, digital maplets, expert knowledge, and other maps were used as training data. Aerial videography transects were flown to acquire reference data. Due to the spatial inaccuracies inherent in the videography reference data, only homogeneous land cover areas were used in the accuracy assessment. The results of the overall accuracy for each scene determined the ordering of scenes within the statewide land cover mosaic (i.e., scenes with higher accuracy had a higher proportion of area represented). An accuracy assessment was then performed on the statewide land cover mosaic. An overall accuracy of 81.8% and a Kappa statistic of 0.81 resulted. A discussion of potential reasons for land cover class confusion and suggestions for classification improvements are presented. Overall deciduous forest was the most common land cover in Virginia. Herbaceous areas accounted for 20% of the land area, which was the second largest. Mixed forest and coastal wetlands were the cover types with the least area, each under 3%.

List of Attached Files

fig25.jpg part1.pdf part2_a.pdf
part2_b.pdf part3_a.pdf part3_b.pdf
part3_c.pdf part3_d.pdf part3_e.pdf
part3_f.pdf part3_g.pdf part3_h.pdf
part3_i.pdf part3_j.pdf part3_k.pdf
part3_l.pdf part3_m.pdf part3_n.pdf
part3_o.pdf part4.pdf


The author grants to Virginia Tech or its agents the right to archive and display their thesis or dissertation in whole or in part in the University Libraries in all forms of media, now or hereafter known. The author retains all proprietary rights, such as patent rights. The author also retains the right to use in future works (such as articles or books) all or part of this thesis or dissertation.