by C Wu, Q Xiao, EG McPherson
Urban Forestry & Urban Greening 7(2):65-762008
In support of Los Angeles’s Million Tree Initiative, a method was devised to determine if there were sufficient potential tree planting sites for the project. Remote sensing technology and QuickBird satellite imagery were used to classify land cover into four tree-relevant categories: impervious surface, trees, irrigated grass, bare soil/dry grass. An algorithm was developed to find suitable planting spaces (land cover of grass, bare soil, or dry grass; tree trunks a specified distance from impervious surface; minimum square feet of pervious surface available; no crown overlap with existing trees). The algorithm featured two modules, the first to identify planting sites and the second to “plant trees” within those areas. It was designed to be run repeatedly, with each iteration adding trees to appropriate areas that remain open. Ground-truthing was performed to assess accuracy. A total of 2.2 million suitable planting sites were found.