Regional approach to map floating and emergent aquatic vegetation in coastal wetlands of eastern Georgian Bay, Lake Huron, Canada.
Expansion and contraction of floating and emer- gent vegetation due to fluctuating water levels has a direct impact on the amount of critical fish habitat in the coastal marshes of Georgian Bay, Lake Huron (Canada). Traditional mapping approaches developed for site-specific studies are too expensive to quantify such changes at the scale of Georgian Bay. Here, we use IKONOS images to develop a classification method (process-tree classification (PTC)), an automated, object-based, image-analysis approach that can produce regional maps of wetland habitat for south-eastern Georgian Bay (1466.7 Km). PTC discriminated among six wetland habitat classes (emergent, high-density floating, low- density floating, meadow, water, and rock) in four IKONOS satellite images with a mean accuracy of 87.4%. The PTC was then applied without modification to 17 other IKONOS images collected concurrently in 2002. Based on analysis of 50 randomly chosen wetlands in these images, we estimate that at 2002 water levels, at least 25% of an average wetland (6.5 ha) contains potential fish habitat. Although the PTC developed is specific to the 21 IKONOS images used in this study, the framework is transferable to satellite images acquired in other regions of Georgian Bay, and the approach itself could be applied to other large lakes.