ABSTRACT

Timely and accurate acquisition of information on the status and structural change of forest composition is crucial to develop strategies for sustainable management of natural resources and ecosystem modeling (Pu and Landry 2012). In practice, field surveys and aerial photograph interpretation are two traditional ways to obtain the information about tree canopy and species composition. Both methods are time-consuming, expensive, and usually cannot provide complete coverage of large areas. Satellite remote sensing has an advantage of being able to obtain data for large areas simultaneously. However, previous studies have demonstrated that accurately mapping individual tree species and tree canopy using moderate-resolution satellite imagery is difficult or impossible (Katoh 1988; Congalton et al. 1991; Brockhaus and Khorram 1992; Franklin 1994; Carreiras et al. 2006).