Southern Pine Beetle Infestation Probability Mapping Using Weights of Evidence Analysis

Jason B. Grogan,

Stephen F. Austin State University - Arthur Temple College of Forestry & Agriculture 

Satellite imagery and other forms of remote sensing currently are more easily available, affordable and increasing in resolution and quality.  However, most hazard rating systems were developed long before these data were readily available.  As new data and technologies have become increasingly available most risk/hazard assessment researchers have attempted to extract data needed for existing hazard rating systems, rather than developing new methods capturing the full potential of the data.  Research was conducted to explore ways of using these data and technologies in innovative new systems, rather than attempting to mold data into old methods.  Weights of Evidence (WofE) spatial analysis was used to predict the probability of Southern Pine Beetle (SPB) (Dendroctonus frontalis Zimmermann) infestation in Angelina, Nacogdoches, San Augustine and Shelby Counties, Texas.  Thematic spatial data derived from LandSat imagery (1974-2002 LandSat 1-7) were used in the analysis.  Predicted infestation probabilities were significantly higher at actual infestation locations, (p < 0.0001) and more infestations occurred in the higher probability areas (p = 0.002).  A significant correlation was found between WofE probability and traditional SPB hazard rating, calculated from forest inventory data.  Probability maps were classified to produce current SPB three and five-class hazard rating maps.  Results indicate the Weights of Evidence geospatial analysis was successful at predicting the probability (hazard) of SPB infestation and showed the analysis benefited from data not used in past systems, such as forest patch size and other spatially-related variables.

corresponding author:

Jason B. Grogan
Columbia Regional Geospatial Service Center
Arthur Temple College of Forestry & Agriculture
Stephen F. Austin State University
jgrogan@sfasu.edu