Object-based Forest Classification to Facilitate Landscape-Scale Conservation

In the Mississippi Alluvial Valley (MAV), conservation partners have implemented strategic habitat conservation for wildlife via a landscape-scale approach to bottomland hardwood forest conservation and restoration (USFWS, 2008).  This landscape conservation design prioritizes areas for reforestation that are contingent upon an accurate classification of forest cover.  To characterize the forest landscape, we developed a repeatable forest classification system utilizing high resolution imagery, satellite imagery (Landsat 5 TM), and object based image analysis with a random forest classifier.  Using this innovative methodology allowed us to perform a large-scale forest assessment quickly and accurately as compared to traditional pixel based approaches.  This landscape-scale forest classification was in turn used to refine a spatially explicit decision support tool that defined reforestation priorities for forest breeding birds (Twedt et al. 1999, 2006). By capitalizing on new analytical tools and satellite imagery, we were able to create a forest classification ruleset that produces accurate and repeatable results and can be applied to large landscapes without introducing interpretation error.  The forest classification ruleset will provide the conservation community with a new tool to facilitate landscape conservation and assessments in a repeatable and transparent manner.