Mortality Estimates of Eastern Forest Insect Outbreaks from Landsat Imagery

M. Joseph Hughes, University of Tennessee, Knoxville, jhughes@utk.edu (Presenter)
Daniel Hayes, Oak Ridge National Laboratory, hayesdj@ornl.gov

Large scale climate models require spatially-explicit estimates of energy, water, and carbon flux from forest disturbance. Despite the important role that insect outbreaks are considered to play in regional-to-continental scale carbon budgets, comprehensive and consistent data sets on mortality currently do not exist, and few modeling frameworks have incorporated their effects. In Eastern Forests, invasive insects such as the hemlock wooly adelgid and southern pine beetle play a significant role in forest degradation by inhibiting tree function, eventually resulting in tree death and increased fire risk. We evaluated the ability to use a 28-year time-series of Thematic Mapper imagery from Landsat 4 and 5 to quantify the severity of known insect outbreaks in the Southern Appalachians. Affected forest regions near the Cherokee National Forest were first identified from the US Forest Service Aerial Detection Survey geodatabase, a spatially-explicit, nominally yearly survey of forest disturbances in the United States derived from sketch-maps. The change detection approach is based on the 'LandTrendr' concept, where cloud-free summer averages of vegetation indices were then calculated from Landsat TM imagery over the area for each year, and piecewise linear regressions fit over time to each pixel. The linear regressions provide a start and end date for discrete forest-change events, as well as a severity of the event from the regression slope. These severities were then normalized using known tree mortality data from the Forest Inventory and Analysis database. The result is estimates of tree mortality for each disturbed region over time, including years in which aerial surveys did not occur, which are then summarized into disturbance within each quarter-degree grid cell. The gridded insect mortality data is designed to be combined with other disturbance information as driver data input into the cohort framework of the Terrestrial Ecosystem Model (TEM) for assessing carbon cycle impacts.

Presentation Type:  Poster

Session:  Poster Session 2-A   (Wed 11:00 AM)

Associated Project(s): 

• Related Activity or Previously Funded TE Activity

Poster Location ID: 47