|
Track: Forestry
Anantha Prasad
USDA Forest Service
359 Main Road
Delaware, OH 43015
Telephone: 614-368-0103
Fax: 614-368-0152
E-mail: prasad@neusfs4153.gov
Louis Iverson
Modeling Tree Distributions in Eastern United States Using ArcInfo GIS and Splus Statistical Package
Defining Issue: ArcInfo GIS and SPlus statistical package were used in an effort to map the current and potential future distributions of 104 tree species in the eastern U.S under two climate change scenarios. Methodology: The modeling effort involved a dynamic exchange of information between the ArcInfo GIS and Splus statistical environments to assess current and potential future tree distributions in the eastern United States. The database effort involved the use of various Unix tools (eg., shell scripts, awk, perl) as well as ArcInfo software's AML and Splus functions to construct the database and automate various processes. Data for over 100,000 plots and nearly 3 million trees east of the 100th meridian were aggregated to a county level. Species importance values for over 2,100 counties were estimated. These tree importance values were combined with county-level data on climate, soils, landuse, spatial pattern and socioeconomics. Several types of prediction modeling were used to predict the
potential future tree distributions under a doubling of carbon dioxide as predicted by the global circulation models (GCMs). These included regression tree analysis (RTA), spatial autocorrelation/regression (SAR), and generalized additive model (GAM). The resulting maps show the distribution of various tree species as well as the variables that control the geographic distribution of the species. Graphical outputs from RTA, SAR, and GAM, combined with the predicted tree species distribution maps in GIS, provide a powerful means of understanding the relationships among various factors associated with current and potential future tree species distributions. Software: ArcInfo modules (mainly: ARCGRID, ARCPLOT, and AML), Splus Statistical Package (including Spatial Statistics) UNIX tools, and Perl.
|
|
