Reconstruct forest disturbance history using Landsat data: methodology development and preliminary results
Chenquan
Huang, University of Maryland, cqhuang@umd.edu
(Presenting)
Samuel
N.
Goward, University of Maryland, sgoward@umd.edu
Jeffrey
G.
Masek, NASA/GSFC, Jeffrey.G.Masek@nasa.gov
Nancy
Thomas, University of Maryland, nthomas1@umd.edu
Scott
Powell, USDA Forest Service, Pacific Northwest Research Station, Scott.Powell@oregonstate.edu
Karen
Schleeweis, University of Maryland, ska1@umd.edu
Forest disturbance and regrowth are significant forces moderating North American carbon balance. Quantifying the carbon fluxes of forest changes requires the changes be assessed with appropriate spatial and temporal details. The Landsat imagery archive accumulated since 1972 provides a unique data source for characterizing forest changes over the last 30+ years. Through the “North American Forest Disturbance and Regrowth since 1972” project, we have developed a highly automatic method for reconstructing the chronicle sequence of forest disturbance history using dense time series of Landsat observations. Specifically, the method was designed to analyze a stack of images consisting of one Landsat acquisition every one or two years. It produces products that reveal where and when each disturbance occurred, and provide indicators of the magnitude of each disturbance. Knowing when each disturbance occurred allows estimation of the stand age of post-disturbance recovering forest. This method has been tested in several areas in eastern and western U.S. The derived results were assessed by visually interpreting the input Landsat images and were found highly reliable. Field plot data collected through the Forest Inventory and Analysis (FIA) program were used to validate the age of post-disturbance forest estimated based on the derived disturbance products. The Landsat derived age of post-disturbance forest was found highly correlated with FIA field measurements.