West Virginia University
Predicting species composition in an eastern hardwood forest with the use of digitally derived terrain variables
Abstract
dc:description.abstractThis thesis addresses the need for improved classification of remotely sensed imagery in the complex hardwood forests of West Virginia. A geographic information system (GIS) was used in conjunction with forest plot data to develop a model to predict species composition in the eastern hardwood forest of West Virginia. The study area was located on the West Virginia University Research Forest (WVURF) in northern West Virginia. Terrain variables including aspect, curvature and slope change drastically at a local scale within the forest to greatly influence species composition. Light Detection and Ranging (LiDAR) data was collected for the entire WVURF, which produced an extremely detailed digital elevation model (DEM), with 1 m spatial resolution. Individual tree crown polygons were created from the LiDAR data so that individual trees could be co-registered to the DEM eliminating the bias of misplaced inventory points. Forest-plot data was collected and each individual tree crown polygon that was created from the LiDAR was assigned actual ground data. Terrain variable values were then sampled for each plot. The data was analyzed using a classification and regression tree (CART) to produce a binomial decision tree that was used within GIS to create a prediction grid of species distribution based on terrain variables. With the decreasing price of data acquisition and with new technology this method is likely to become more widespread and useful to various management agencies.
Degree
thesis:*- Name thesis:degree_name
- MS
- Level thesis:degree_level
- Thesis
- Discipline thesis:degree_discipline
- Wood Science and Technology
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Flanigan, Richard D.
Subjects
dc:subject × 4Identifiers
dc:identifier.*- Identifier
- https://researchrepository.wvu.edu/etd/2957