WORLDVIEW-2, QUICKBIRD AND IKONOS SATELLITE MAPPING AND CHANGE ASSESSMENT OF COASTAL ENVIRONMENTS (Poster)
Academic Article
Overview
Overview
Abstract
Coastal environments require high spatial resolution data to map habitats to the level of precision that management officials need. The primary goals of this project were to identify the usefulness of various types of satellite imagery for coastal habitat mapping, standardize the mapping process by testing several image processing software classification algorithms, measure the spatial accuracy and precision of the resulting maps, and compute habitat change in the study area between 2002 and 2010. WorldView-2, QuickBird and IKONOS satellite sensors were used to classify the study area using unsupervised and supervised methods, a variety of spectral band combinations, LiDAR elevation and texture data, unsharpened and pan-sharpened images, and spatial filtering. In total, 168 maps were generated to determine the best combination of data and methods. Results indicated that WorldView-2 images produce the most accurate maps using supervised classification and all eight bands from the sensor. WorldView-2 maps were consistently more accurate than QuickBird and IKONOS (73 out of 80 or 91%). Supervised maps tended to be more accurate than unsupervised (61 out of 76 or 80%). Pan-sharpening improved 35 out of 72 maps (49%), and majority filtering to smooth the maps improved 55 out of 84 maps (65%). During this eight-year period 20% of overall study area changed and 30% of the barrier island alone changed. Intertidal marsh experienced the most change, but smaller habitat classes changed substantially as well, including 84% of upland scrub-shrub, which occurs primarily along the sparsely-researched spoil islands. These results shed light on the dynamic nature of this coastal area, validate the use of relatively new satellite sensors, and may be used to guide coastal habitat mapping and monitoring at other locations.
