Dramatic variability in Katrina-induced mortality in the Pearl River basin.  Cypress-tupelo swamp forests in swales and other low-lying sites experienced little damage whereas adjacent bottomland hardwood forest suffered extremely high mortality (high resolution aerial imagery from LSU’s GIS Information Clearinghouse).  Initial results from this project will be published in the 16 Nov 2007 issue of Science.

High tree mortality in a bottomland hardwood forest of the Pearl River basin caused by hurricane Katrina.  Invasive Chinese tallow trees, visible in the background, are rapidly colonizing these disturbed areas.

The primary goals of this project are to: (i) Develop empirical relationships between wind-field parameters and species-specific tree mortality rates for Major (Saffir-Simpson Category 3-5) hurricanes striking the Southeastern Region, (ii) Modify an ecosystem simulation model (Ecosystem Demography – ED) to allow simulation of catastrophic hurricane-driven tree mortality dynamics; (iii) Develop a set of hurricane scenarios based on historical records and possible future trends driven by natural cycles and anthropogenic climate change, and employ ED to predict forest carbon balance impacts from these scenarios; and (iv) Utilize hurricane events that are likely to strike the Southeastern Region during the course of this study to further expand and test empirical relationships and model predictions.

The overall working hypothesis is that disturbance related to hurricane activity in the Southeastern Region has significant negative impacts on forest carbon balance and the regional carbon sink, and that expected future changes in the hurricane disturbance regime will further degrade the Southeastern Region carbon sink.

This project will employ a synthetic approach combining field-based tree mortality and damage investigations, remote sensing image analyses, and an ecosystem simulation model.  Forest inventory analysis methods will be employed to estimate biomass loss from damage and tree mortality.  Spatial variability in mortality rates will be correlated to hurricane wind-field parameters using regression methods.  Remote sensing investigations using spectral mixture analysis on non-photosynthetic vegetation endmembers will provide maps of disturbance variability within wind-field classes, and allow further development of tree mortality probability distribution functions.  A modified ED model will allow integration of results, and be used to explore historical and future hurricane scenarios on Southeastern Region forest carbon balance.

Expected deliverables include: (i) A network of forest inventory plots for Major hurricanes striking the Southeastern Region. (ii) Atmospherically corrected remote sensing images (e.g. ALI, Hyperion, Landsat) and relative abundance maps of disturbed areas within footprints of Major hurricanes.  (iii) A modified ED model allowing simulation of past hurricane events and future scenarios.