부산대학교 도서관

Shoreline wetlands in the Great Lakes basin are susceptible to frequent changes in vegetation composition due to fluctuations in climate and water level. Although water-level changes occur naturally and are essential to maintain productivity, the magnitude and rate of these changes can have a significant effect on the wetland ecosystems. From a management and scientific viewpoint, it is important to be able to map and monitor these long-term changes. Using Long Point on Lake Erie as a test site, the goal of this research study is to refine methods for using multiple dates of Landsat imagery to map and monitor wetlands over a relatively long period of time. Landsat images covering the period from 1976 to 1999 are used to produce multitemporal normalized difference vegetation index (NDVI) images. Two change-detection methods, postclassification comparison and multitemporal data clustering, were selected to determine patterns of change in the Long Point wetlands over the 23 year period. These patterns are compared with lake water levels and Palmer drought severity index (PDSI) data recorded over the same time period. Results show that large sections of the shallow marshes of Long Point have experienced an increase in the amount of emergent vegetation over the period of study. This occurred simultaneously with downward trends in lake water level and PDSI values. Unlike the postclassification comparison approach, the multitemporal data clustering technique provides a method to observe fluctuations in NDVI over the entire time period.