| Abstract Detail
Ecology Hackett, Rachel A [1], Heumann, Benjamin W [2], Monfils, Anna K [3]. A remote sensing approach to delineation of floristic zones within wetland communities. Remote sensing has been used to map land cover data for several decades. As the technology advances, we are looking to not only distinguish trees from buildings and roads but to identify different floristic zones within larger natural systems. Mapping of floristic zones within a larger natural system can be a useful tool for researchers and managers searching for distinct or unique habitats, creating floristic zone based management areas, studying within wetland landscape ecology, monitoring management treatments, or monitoring change over time. Some of the obstacles faced by researchers and managers wanting to incorporate remote sensing into their work has been attributed to 1) cost of imagery, 2) access to imagery of appropriate resolution(s), and 3) a knowledge and communication gap between experts. To bridge this gap, we aimed to evaluate the accuracy and accessibly to map floristic zones within prairie fen wetlands among three imagery sources: unmanned aerial system (i.e., drone), National Agricultural Imagery Program (i.e., manned aircraft), and Worldview-2 satellite. These image sources each had different spectral resolution, spatial resolution, and accessibility. The imagery was classified using eCognition 8.9.1 software with object-based image analysis and supervised classifying algorithms. Each classification map was evaluated for accuracy using ground-truthed floristic zone classification data. The results illustrated which type of imagery was most accurate overall in classifying the floristic zones within a wetland, which was most accurate in identifying individual zones, and which was most effective given the accessibly and cost of the different sources. With this research, researchers and managers can better determine their investment needs into acquiring the best imagery for their objectives for classifying floristic zones within a high diversity wetland. Log in to add this item to your schedule
1 - Central Michigan University, Earth and Ecosystem Science, 1455 Calumet Court , Bioscience Building 2100, Mount Pleasant, MI, 48859, USA 2 - Central Michigan University, Institute for Great Lakes Research, Department of Geography, Dow Science Complex 279, Mount Pleasant, MI, 48859, USA 3 - Central Michigan University, Biology, 2401 Biosciences, Mt. Pleasant, MI, 48859
Keywords: wetland remote sensing delineation classification imagery community.
Presentation Type: Oral Paper Session: 3, Ecology Section - Community Processes and Delineation Location: Sundance 5/Omni Hotel Date: Monday, June 26th, 2017 Time: 8:45 AM Number: 3003 Abstract ID:77 Candidate for Awards:Ecological Section Best Graduate Student Paper |