Date of Award

4-2018

Degree Name

Master of Science

Department

Geography

First Advisor

Dr. Charles Emerson

Second Advisor

Lisa DeChano-Cook, Ph.D.

Third Advisor

Gregory Veeck, Ph.D.

Keywords

Geographic Information Systems (GIS), SANET, deer-vehicle collisions, ordinal logistic regression, network kernel density estimation

Access Setting

Masters Thesis-Open Access

Abstract

Kalamazoo County ranked 15th among 83 Michigan counties in 2015 with 917 deer-vehicle collisions (DVCs). With the white-tailed deer population on the rise, the chances of collisions are also more likely to increase. Most predictive models developed for wildlife-vehicle collisions are used for specific areas and have localized characteristics that are hard to apply to different locations and animal species. By analyzing Kalamazoo County’s specific land cover, weather, traffic, and time variables with locations of deer-vehicle collisions throughout the county, characteristics are identified distinguishing between areas of differing risk. Geographic Information Systems (GIS) can locate areas with a high rate of collisions using network kernel density estimations. These estimations were the response variable for environmental, weather, traffic, and time ordinal logistic regression models. The creation of statistical models and the use of variable reduction techniques identify variables associated with changes in DVC risk. Within Kalamazoo County, only two variables, wet roads and visibility, were found to be consistently statistically significant in terms of spatial variations in DVC risk levels.

Download

Share

COinS