Urban tree cover change in Detroit and Atlanta, USA, 1951-2010

We assessed tree cover using random points and polygons distributed within the administrative boundaries of Detroit, MI and Atlanta, GA. Two approaches were tested, a point-based approach using 1000 randomly located sample points, and polygon-based approach using 250 circular areas, 200 m in radius (12.56 ha). In the case of Atlanta, both approaches arrived at similar estimates of tree cover (50-53%) for both time periods, yet they show that roughly one-third of the tree-covered land area in 1951 was also tree-covered in 2010 and about 30-31% of the sampled land area lacked tree cover during both assessment periods. In the case of Detroit, the two approaches resulted in different estimates of tree cover (19.6% vs. 30.8% in 2010), yet similar levels of transitions over time. The only similarities between the two cities were that about 15-20% of each city's land area was covered with trees in 1951, yet lacked tree cover in 2010. While the polygon-based approach to estimating tree cover may result in a product that more explicitly represents covered areas, the point-based approach is recommended due to the time and effort involved with the polygon-based approach and potential error introduced through topographic displacement of trees and shadows. Overall, canopy cover over time remained stable while distribution varied greatly. However, while multi-decade change in aggregate is undetectable at the scale of a city, there seems to be substantial shifts in the spatial arrangement of the tree canopy.