INVENTORY OF VEGETATION SPECTRAL PROPERTIES IN THE SOUTH BAY SALT PONDS: A DATABASE FOR ENHANCING DECISION SUPPORT AND RESTORATION MAPPING

摘要

In the past century, more than 85percent of the historical marshlands in the San Francisco Bay were converted to salt ponds or filled for urban development, resulting in a loss of biodiversity. The municipalities along the southern margin of the San Francisco Bay are in the process of one of the most extensive tidal wetland restoration projects ever undertaken, the South Bay Salt Pond Restoration Project (SBSPRP). The goal of this project was to perform an analysis of the spectral variation between different salt pond vegetation types and to track the changes in vegetation distribution from 2000 to 2010. These data will be shared with the SBSPRP partners to aid in their three-year classification of vegetation. This project has identified the spectral characteristics of dominant salt marsh vegetation through the use of in-situ spectral measurements and classification of remotely sensed imagery from EO-1 Hyperion and Landsat TM 5. Fieldwork included the use of a handheld spectroradiometer to gather spectral curves for analysis as well as obtaining point vegetation information for image classification. Comparison of the spectral signatures of the dominant vegetation showed little distinction among vegetation species. Field data and IKONOS imagery were used to identify presence of vegetation throughout the study area to aid in the classification of Landsat imagery, and to track the yearly changes in vegetation colonization for the region between 2000-2010. The spectral angle mapper classification algorithm was applied to a July 2010 Hyperion scene to classify pickleweed in the Alviso area. For this study area, it was determined that Landsat is better suited at detecting overall changes in vegetation. Additional field data could improve the classification of Hyperion imagery.