Close Window

Automated characterization of disturbance year, intensity, and recovery rates in a two-decade stack of yearly Landsat Thematic Mapper imagery

Robert E Kennedy, USDA Forest Service PNW Research Station, robert.kennedy@oregonstate.edu (Presenting)
Todd Schroeder, USDA Forest Service PNW Research Station, todd.schroeder@oregonstate.edu
Warren B Cohen, USDA Forest Service PNW Research Station, warren.cohen@oregonstate.edu

To relate landscape-level trends in forest disturbance and recovery to the environmental and economic factors that affect them, cross-ownership maps are needed of the timing and severity of forest disturbance as well as the rate of revegetation. The historical archive of Landsat imagery is a potentially rich source for developing such maps, but extraction of information from Landsat image stacks requires novel change detection approaches. Here, we describe results from a method for extracting continuous-variable estimates of forest disturbance and recovery properties from a stack of near-yearly Landsat images for a large (180 by 180km) region of western Oregon, U. S. A., from the year 1984 to 2004. The method utilizes a robust non-linear least-squares fitting algorithm to match idealized disturbance or recovery trajectories to temporal traces of normalized reflectance for every pixel in the image. Mapped disturbance parameters include year of disturbance, magnitude of reflectance change during disturbance (an estimate of intensity of disturbance), and the exponent of an exponential curve fit to the recovery trajectory of reflectance (an estimate of the recovery rate). This latter parameter is also used to describe recovery rate of areas disturbed prior to the change interval that are still recovering vegetative cover. Unlike other landscape disturbance maps, the approach requires no pre-stratification or classification, and determines thresholds for change from simple f-statistics describing the goodness of fit of the disturbance or recovery model.

Presentation Type:  Poster

Abstract ID: 111

Close Window