B53E-0229:
Developing Remote Sensing Products for Monitoring and Modeling Great Lakes Coastal Wetland Vulnerability to Climate Change and Land Use

Friday, 19 December 2014
Laura L Bourgeau-Chavez1, Mary Ellen Miller1, Michael Battaglia1, Elizabeth Banda1, Sarah Endres1, William S. Currie2, Kenneth J. Elgersma3, Nancy H F French1, Deborah E. Goldberg4 and David W Hyndman5, (1)Michigan Tech University, Ann Arbor, MI, United States, (2)University of Michigan Ann Arbor, Ann Arbor, MI, United States, (3)University of Northern Iowa, Department of Biology, Cedar Falls, IA, United States, (4)University of Michigan Ann Arbor, Ecology and Evolutionary Biology, Ann Arbor, MI, United States, (5)Michigan State University, East Lansing, MI, United States
Abstract:
Spread of invasive plant species in the coastal wetlands of the Great Lakes is degrading wetland habitat, decreasing biodiversity, and decreasing ecosystem services. An understanding of the mechanisms of invasion is crucial to gaining control of this growing threat. To better understand the effects of land use and climatic drivers on the vulnerability of coastal zones to invasion, as well as to develop an understanding of the mechanisms of invasion, research is being conducted that integrates field studies, process-based ecosystem and hydrological models, and remote sensing.

Spatial data from remote sensing is needed to parameterize the hydrological model and to test the outputs of the linked models. We will present several new remote sensing products that are providing important physiological, biochemical, and landscape information to parameterize and verify models. This includes a novel hybrid radar-optical technique to delineate stands of invasives, as well as natural wetland cover types; using radar to map seasonally inundated areas not hydrologically connected; and developing new algorithms to estimate leaf area index (LAI) using Landsat. A coastal map delineating wetland types including monocultures of the invaders (Typha spp. and Phragmites austrailis) was created using satellite radar (ALOS PALSAR, 20 m resolution) and optical data (Landsat 5, 30 m resolution) fusion from multiple dates in a Random Forests classifier. These maps provide verification of the integrated model showing areas at high risk of invasion. For parameterizing the hydrological model, maps of seasonal wetness are being developed using spring (wet) imagery and differencing that with summer (dry) imagery to detect the seasonally wet areas. Finally, development of LAI remote sensing high resolution algorithms for uplands and wetlands is underway. LAI algorithms for wetlands have not been previously developed due to the difficulty of a water background. These products are being used to improve the hydrological model through higher resolution products and parameterization of variables that have previously been largely unknown.