Harold Morrison

Quantifying changes in land cover within cities is important for understanding the dynamic ecology of urban landscapes. In this study, supervised, maximum-likelihood classifications were conducted to identify six unique land cover features in the City of Toronto, Canada using Landsat 5 Thematic Mapper (TM5) images obtained in 1985, 1994, and 2005. Overall classification accuracies averaged 0.94 Kappa for the three years. Land cover class statistics were calculated and presented at a city-wide level. Post-classification change detection was conducted to investigate how the land cover classes changed from 1985-1994, 1994-2005, and 1985-2005, and results were presented at both a city-wide and municipal ward level. In 2005, impervious surfaces accounted for over 34 percent of Toronto's land cover, which is a 20 percent overall increase in this cover class from 28.4 percent in 1985. Contiguous treed areas accounted for 5.8 percent of Toronto's total land cover in 2005, down from 6.8 percent in 1985; these findings indicate an overall loss of approximately 15 percent total contiguous tree cover in two decades. Findings for municipal wards vary widely, but appear to show temporal correlation with dominant construction periods. The study confirms the usefulness of Landsat data to quantify general land cover changes within urban areas.