A Comparison of Satellite and Gis Classification Techniques for Delineating Forested Wetlands
- Ahl, Douglas E.
University of Maine
Wetland delineation has become an important issue for wildlife managers, municipal planners, and various state and federal agencies. Wetlands perform a number of valuable services for humans including: moderation of downstream flooding, maintenance of good water quality, and pollution control. However, the United States Department of the Interior estimates that only 4Wk of the nation's original wetlands remain intact. Several federal agencies are now debating. how wetlands should be defined and protected. The National Wetland Inventory Project (NWI) was initiated in 1975 to inventory the wetlands of the United States using aerial photographs. NWI interpreters found that forested wetlands were the most difficult to identify. Previous studies have suggested the integration of remote sensing and geographic information systems (GIS) technologies to improve wetland mapping.This investigation utilized remote sensing and GIS technologies to compare different methods for delineating forested wetlands in a Maine study area. An unsupervised classification using Landsat Themati Mapper (TM ) data was implemented followed by a supervised or hybrid approach . A Tasseled Cap transformation was employed to reduce the dimensionality of the Landsat data to Brightness, Greenness, and Wetness features. The Tasseled Cap features were then used in an unsupervised classification. Finally, an additive GIS model was developed incorporating Landsat TM, NWI, soils, topographic, and hydrography data to identify potential forested wetland environments. Four classes were identified in each of the classification methods: forested wetland, other wetland, forested upland, and other upland. Classification results were compared to photo interpreted reference data.
You must be logged in to post a comment.