Identifying High Risk Locations of Animal-Vehicle Collisions on Washington State Highways

Animal-vehicle collisions (AVCs) have been increasing with increases in both animal populations and motor vehicle miles of travel and have become a major safety concern nationwide. Most previous AVC risk studies have not considered factors related to human behavior or the spatial distribution of animal populations in depth because of missing datasets or the poor quality of data. The two common sources of data—the Collision Report (CRpt) and Carcass Removal (CR) datasets—are often found to be significantly different. To address these data issues, two approaches were followed in this research. In the first approach, a fuzzy logic-based data mapping algorithm was developed to obtain a more complete AVC dataset from the CRpt and CR data. In comparison to the original CR dataset, the combined dataset increased the number of AVC records by 13~22 percent. This combined dataset was used to develop and calibrate amicroscopic probability (MP) model that can explicitly consider drivers’ behaviors and the spatial distributions of animal populations. In the second approach, a Diagonal Inflated Bivariate Poisson (DIBP) regression model was developed to fit the two datasets simultaneously. The DIBP model can effectively identify the overlapping parts of the two datasets and quantify the impacts of road and environmental factors on AVCs.

Both proposed models used the CRpt and CR data collected from ten selected study routes in Washington state. The MP model results showed that variables including number of lanes and animal habitat areas are significantly associated with the probability of animals crossing the highway. Two factors, speed limit and truck percentage, have impacts on the probability of a driver’s ineffective response. A wider median may decrease the probability of an animal failing to avoid a collision. The DIBP results showed that speed limit, restrictive access control, and roadway segment length have an increasingrelationship with AVCs. Furthermore, hotspots (high risk roadway segments) were identified for all the study routes on the basis of the modeling and data analysis results. These quantitative results will help WSDOT develop countermeasures to AVCs.

Publication Date: 
Sunday, October 31, 2010
Publication Number: 
WA-RD 752.1
Last modified: 
10/12/2016 - 15:42
Yinhai Wang, Yunteng Lao, Yao-Jan Wu, Jonathan Corey.
Transportation Northwest Regional Center X (TransNow) and Washington State Transportation Center (TRAC)
Number of Pages: 
Accident prone locations, Collisions, Animals, State highways, Accident data, Animal behavior, Drivers, Correlation analysis, Fuzzy algorithms, Regression analysis, Accident risk forecasting.