Spatial Logistic Models of Insect Outbreaks in Southern Forests
Investigators:
Thomas P. Holmes,
Southern Research Station
Marcia Gumpertz, North Carolina State University
John M. Pye, Southern Research Station
Description:
Common indicators of forest health are represented as binary variables such as the presence or absence of insect outbreaks. Logistic regression is a useful method for quantifying the relationship between the probability of a binary outcome and a set of explanatory variables. These models can help determine which environmental, weather and land use variables influence outbreak status.
Methods for fitting logistic models to independent data are readily available. However, spatial and temporal autocorrelation of insect outbreaks make this a nonstandard statistical problem. This research applies the method of generalized estimating equations for fitting spatially correlated binary data to a set of covariates. Results to date indicate that, for county level data on the incidence of southern pine beetle outbreaks, spatial autocorrelation occurs over a range of about 150 miles and that temporal and spatial autocorrelation strongly influence model parameter estimates and conclusions.
Problem Area(s):
Quantitative impacts, Aggregate impacts
Status:
Work in process.
Products:
Gumpertz, M. L., Wu, C.-T., and Pye, J. M. 2000. Logistic regression for southern pine beetle outbreaks with spatial and temporal autocorrelation. Forest Science 46:95-107.
|
modified:
11-JUL-2000
|
|
USDA FS SRS
|