Mapping and monitoring wetlands using airborne multispectral imagery

Wetland areas in the semiand west of the United States play an important eco-hydrological role. The size, location and function of wetlands are affected by numerous factors such as variations in water availability, water quality, geomorphic characteristics and anthropogenic factors such as runoff from irrigation systems, or discharge of urban effluents. Typically, wetland vegetation occurs in patches of variable size and plant species composition, requiring high-resolution imagery to accurately identify distribution and extent of the different wetland habitat types when mapping these complex systems. Airborne multispectral digital imaging offers several advantages, including cost-effectiveness and ability to resolve wetland features ranging in size from a few meters to hundreds of hectares. This paper describes the mapping of wetland habitats within the Great Salt Lake Ecosystem using high resolution multispectral imagery (1-m pixels). Image classification was based on supervised signature extraction supported by comprehensive ground truth. An error analysis was conducted using independent ground truth data, and the resulting classification accuracy for the final wetland resource map was 92%.