HR: 1340h
AN: B33E-1076    [Abstracts]
TI: Use of JERS Satellite Imagery Mosaics for Boreal Wetlands Mapping
AU: Whitcomb, J B
EM: jane.whitcomb@sbcglobal.edu
AF: University of Michigan, 3238 EECS 1301 Beal Ave, Ann Arbor, MI 48105 United States
AU: * Moghaddam, M
EM: mmoghadd@umich.edu
AF: University of Michigan, 3238 EECS 1301 Beal Ave, Ann Arbor, MI 48105 United States
AU: Kellndorfer, J
EM: josefk@whrc.org
AF: Woods Hole Research Center, PO Box 296, Woods Hole, MA 02543 United States
AU: McDonald, K
EM: kyle.mcdonald@jpl.nasa.gov
AF: Jet Propulsion Laboratory, Mail Stop 300-233 California Institute of Technology 4800 Oak Grove Drive, Pasadena, CA 91109-8099 United States
AU: Podest, E
EM: erika.podest@jpl.nasa.gov
AF: Jet Propulsion Laboratory, Mail Stop 300-233 California Institute of Technology 4800 Oak Grove Drive, Pasadena, CA 91109-8099 United States
AB: Wetlands play a critical role, not only in the health of boreal ecosystems, but also as significant carbon sinks and sources whose quantification may be key in balancing the global carbon budget. An accurate assessment of the spatial and temporal distribution of wetlands could thus be used to significantly improve estimates of the global net carbon exchange. The locations, types, and extents of wetlands are still uncertain, however, partly because it is difficult to identify and classify wetlands on a global scale using widely available optical remote sensing data. Low-frequency synthetic aperture radar (SAR) is well suited to the task of identifying and classifying wetlands. Its ability to penetrate the forest canopy has been used to advantage in characterizing vegetation structure, biomass, and moisture content. It is especially sensitive to scattering from flooded forest swamplands, due to its ability to penetrate vegetation and reflect back from standing water under vegetation. We have used multi-temporal L-band JERS-1 SAR imagery in order to produce a thematic map of wetlands in the North American boreal zone. The map will identify four land cover classes based on their distinct scattering characteristics: Open water, Herbacious wetlands (e.g., marshes, fens, bogs), Woody wetlands (e.g., swamps), and Non-wetlands. Tasks involved in generating the map included the following: 1) training sites for each class within each ecoregion are identified, 2) two seasons of JERS imagery are geographically co-registered with a digital elevation model (DEM), a slope model, and an open water mask, 3) the slope model is used to mask out areas that cannot be wetlands and the open water mask is applied to distinguish herbacious wetlands from open water, 4) the spectral characteristics of each wetlands class as a function of imagery acquisition date and boreal ecoregion are identified, and 5) classification is performed by a combination of the spectral/eco-regional knowledgebase with a minimum distance classifier. The performance of the algorithm is validated using ground truth reference data from multiple wetland validation sites of known characteristics.
DE: 0497 Wetlands (1890)
SC: Biogeosciences [B]
MN: Fall Meeting 2005