| This thesis explores the viability of using multispectral scanner data to detect and delineate potential seepage sites along the Erie Canal. Thermal and near infrared data were collected with a multispectral scanner in 29 flight lines, over three major sections covering 145 miles of the canal. Automated image classification techniques were used to separate potential seepage sites, indicated by wet areas on the ground, from other types of ground cover. The resulting classified images were registered to New York State Department of Transportation planimetric maps and loaded into a geographic information system database along with canal centerline station locations. Hardcopy maps were printed for field use. Error and accuracy matrices were computed for three of the 29 flight lines, one for each of the major sections, and the results indicated the average accuracy of the three flight lines' classifications was 92.62%. |
