by M Alonzo, K Roth, D Roberts

Remote Sensing Letters 4(5):513-521

2013

Urban trees must be identified at the species level for management purposes and to make use of tools that quantify ecosystem services, but assessing species on a tree-by-tree basis on the ground is time consuming and expensive. Remote sensing is increasingly able to distinguish individual trees from aerial imagery and to determine crown size, but identifying species from imagery remains challenging. In this study, 15 common species were classified in downtown Santa Barbara, CA, using AVIRIS imagery and canonical discriminant analysis (CDA). An overall accuracy of 86% was achieved, with the largest, most densely crowned trees being the easiest to identify.

Region: Santa Barbara, California
Publication Type: Journal article
Keywords: aerial and satellite imagery, remote sensing, species composition, and urban forestry