Mapping private gardens in urban areas using object-oriented techniques and very high-resolution satellite imagery

Although private gardens represent the largest component of urban green spaces, they are the least studied, mainly because of difficulties in accessing private property and a lack of methodology for classifying and analyzing garden data. Newly available high-resolution aerial imagery may make it possible to gain important ecological data from gardens without having to have physical access. Here, Ikonos imagery of 4-m resolution was subjected to object-oriented classification in which contiguous pixels are grouped iteratively into meaningful objects. The goal was to isolate private residential gardens and classify them into three categories based on the proportion of trees and shrubs vs lawn (< 30%, 30-70%, > 70% trees and shrubs). The method proved 90% accurate at identifying private gardens. Distinguishing among the three garden types was less successful but offered guidance for future endeavors.