Date of Award

Winter 2008

Project Type

Thesis

Program or Major

Natural Resources

Degree Name

Master of Science

First Advisor

Russell G Congalton

Abstract

Benthic habitats are some of the most difficult habitats to map using remote sensing. In a study of six bays in the Texas Gulf Coast, maps were created from digital aerial imagery. Using an object-based image analysis (OBIA) approach, the image was classified with the Classification and Regression Tree (CART) technique to produce a draft map. The draft map was then extensively manually edited to produce a contractor map. Accuracy assessments of both maps revealed that the two were not significantly different. The objective of this study was to determine why the editing did not improve the draft map. Our analyses indicate that the small segmentation scale parameter chosen for the map over-segmented the image and reduced the effectiveness of the classification technique and the manual editing. When compared to a similar map with a larger scale parameter, the smaller initial polygons proved more difficult to accurately classify.

Share

COinS