Assessing Land Cover Change and Anthropogenic Disturbance in Wetlands Using Vegetation Fractions Derived from Landsat 5 TM Imagery (1984-2010)

Anthropogenic disturbance of wetlands in Canada is extensive. In wetland monitoring programs disturbance assessment often relies on single date field and geo-spatial data, thereby rendering detection of the nature, timing and magnitude of disturbance events and trends difficult. The Landsat temporal archive provides potential for more comprehensive temporal change analysis. However, its 30 m pixel size may result in omission of small disturbances or lack of spatial precision near boundaries where change often occurs. Spectral mixture analysis (SMA) is a subpixel technique that has been used to assess change in a variety of land cover types, but rarely for non-coastal wetlands. This research utilized SMA vegetation, bare, and moisture fractions derived from 26 Landsat 5 Thematic Mapper (TM) scenes from 1984 to 2010 in two-date comparisons and time series analysis to assess disturbance in two wetland complexes in eastern Ontario. The 1984-2010 two-date analysis showed overall declines in vegetation and moisture fractions and increases in the bare fraction, while time series analysis over the 26 year period showed more variable inter-annual changes including years of sudden change, cyclic trends and more gradual trends. Anthropogenic disturbances that were identified included a lake created for recreational purposes, logged/cultivated areas, and wetland degradation due to encroaching development.