
| HOME SCHEDULE AUTHOR INDEX SUBJECT INDEX |
|
80 Comparison and accuracy assessment of wetland land cover classification systems in Oregon. Lett, Christina1, Santelmann, Mary1, 1 Oregon State University, Corvallis, OR ABSTRACT- The purpose of this study was to better understand the level of wetland classification consistency among three different land-cover data layers created using Geographical Information Systems (GIS) and remote sensor technology. Where inconsistencies existed, the goal was to identify and characterize the inconsistency. We focused on wetland land-cover classes used for habitat suitability studies. A GIS was used to manage and spatially combine data from three sources: 1) the Oregon Department of Fisheries and Wildlife (ODFW) 2) the National Wetlands Inventory (NWI) and 3) the Pacific Northwest Ecosystem Research Consortium (ERC). In addition to the GIS, digital orthophotos and site visits were used to verify the accuracy of a random sample of the wetlands (1000 wetlands sampled from orthophotos and 100 wetlands sampled by site visits) identified by each system. From this verification process, we were able to compare the strengths and weaknesses of each classification system. The NWI identified more wetland area in the watershed than either the ODFW or ERC land cover data, and were most often correct in assignment of the site to a field verified wetland class (69% of sites assessed) compared to only 25% accuracy for the ERC and 64% for the ODFW. The ERC system was unable to distinguish forested wetlands from hardwood forest or mixed forest and thus underestimated this wetland type. The ODFW system tended to under-identify seasonal wetlands and wet shrub wetlands relative to the other two systems. However, the NWI did not identify small streams in the area, which both of the other systems included as a wetland type, and was less likely to identify the wet shrub wetland type relative to the ERC system. Understanding the limitations of each data coverage will help in developing a better characterization of wetland land cover for use in watershed restoration planning. KEY WORDS: wetlands, Geographic Information Systems |