Mapping northern wetlands with high resolution satellite images and LiDAR

Abstract

Most wetland maps must be updated for much of North America. High-resolution satellite images and LiDAR-based elevation data have been developed in recent years, and these may prove useful in wetland mapping. We compared the accuracy of the Wisconsin Wetland Inventory maps (WWI) to maps derived from IKONOS high resolution satellite data and LiDAR data for a 63.4 km2 study area in north-central Wisconsin. One-meter, resolution merged multispectral and panchromatic IKONOS data were used with a 1-m resolution LiDAR-based digital elevation model in a manual interpretation of wetland types. IKONOS/LiDAR data were significantly more accurate (74.5 classification accuracy) than WWI data (56) when wetlands were categorized into WWI classes. Improved accuracies were largely due to less confusion between upland and wetland classes, and generally better distinction among the various wetland classes. LiDAR data improved upland/wetland distinction, based on relative terrain heights and derived terrain-shape indices. Confusion in all classifications was particularly common among lowland coniferous species, and among evergreen shrub and moss classes. © 2009 The Society of Wetland Scientists.