An Automated Image Selection Approach for Annual Assessment of US Forest Dynamics
Abstract
North American forests are thought to be a long-term sink for atmospheric carbon, with much of the sink attributed to either forest regrowth from past agricultural clearing or to woody encroachment. However, the magnitude of the North American forest sink is uncertain, because disturbance and regrowth dynamics are not well characterized or understood. The North American Forest Dynamics (NAFD) team from the University of Maryland, US Forest Service and NASA has been working to develop a sound understanding of forest disturbance patterns in North America as a contribution to the North American Carbon Program since 2003. We have found that spatial and temporal sampling of the Landsat data record result in substantial residual errors remaining in our US national estimates for disturbance rates. Conducting a comprehensive annual, wall-to-wall analysis of US disturbance history over the 1985-2010 (Landsat Thematic Mapper) time period will overcome these limitations. The first phase of production includes developing a successful automated image selection approach for selecting the 10,000-15,000 Landsat images (depending on cloud compositing needs) necessary for an annual assessment (1984-2010) of CONUS forest dynamics. A major goal of this approach is to minimize the negative impact of clouds, phenology, and mis-registration by selecting the images necessary for producing annual, clear views of the land surface using a compositing algorithm. This approach has the potential of being adapted for implementation outside the CONUS.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2011
- Bibcode:
- 2011AGUFM.B51N0617S
- Keywords:
-
- 0428 BIOGEOSCIENCES / Carbon cycling;
- 0480 BIOGEOSCIENCES / Remote sensing