Prototyping Global Industrial Forest Mapping, a Landsat Spatio-Temporal Approach

The objective of this proposal is to prototype a method to automatically map and monitor global industrial forests from Landsat imagery. The prototype will be applied to wall-to-wall WELD data of the conterminous United States, and tested over planation forests in Malaysia and Brazil to assess the feasibility of global mapping using Landsat data. The development methodology will follow a three stage process: 1. Per-pixel multitemporal analysis of forest cover change, followed by 2. Object-oriented analysis of shapes, textures, and spatial relationships of the areas of deforestation/afforestation 3. Validation using existing extensive field data Through including methods commonly employed in computer vision, but rarely used in remote sensing, we will implement methodology that closely simulates the analysis performed by the human interpreter.