6. Summary and Conclusions
The following chapter relates the conclusions reached as a result of this research effort. The first section provides a brief description of the reconstruction site candidate selection, as well as the projects that were ultimately selected. The second part of this chapter gives the general trends in crash rate observed for study sections. The third section gives the roadway design variables studied for significance, as well as the overall fit of an estimated model. The fourth part is an analysis of what can be said about the individual reconstruction sections. Finally, two recommendations for further study are proposed.
6.1 Project Selection
The first aspect of this research effort was the selection of candidates for study. ArcGIS proved to be a powerful tool in locating sections of roadway that experience either a high frequency or rate of reported wild animal crashes. Statewide maps showing the animal-vehicle crash rates and crash frequencies were created using this tool. With the aid of the State Transportation Improvement Programs (STIPs) created over the past decade, it was a straight-forward process to identify seven recent reconstruction projects on segments that experience a higher than normal number of wildlife crashes. The seven reconstruction projects were as follows:
- WY 130 Centennial East Section – between Centennial and Laramie from milepost 21.32 to 27.431. Reconstruction was started in November of 1996.
- US 14/16/20 Hanging Rock Section – between Yellowstone National Park and Cody from milepost 19.4 to 27.6. Reconstruction was started in June of 1998.
- US 189 Round Mountain Section – between Kemmerer and LaBarge from milepost 45.78 to 59.02. Reconstruction was started in April of 1999.
- US 14/16 Clearmont North Section – between Sheridan and Gillette from milepost 38.61 to 45.96. Reconstruction was started in November of 1999.
- WY 34 Morton Pass Section – between Bosler Junction and Wheatland from milepost 9.69 to 16.53. Reconstruction was started in March of 2001.
- US 89 Astoria Section – between Alpine Junction and Jackson from milepost 136.65 to 140.69. Reconstruction was started in March of 2000.
- US 26/85 Torrington West Section – between Torrington and Lingle from milepost 94.60 to 102.93. Construction was started in October of 1997.
6.2 Overall Trends
Using an analysis that compared the changes in crash rates for each of the seven sections, several trends were identified as to the changes in risk following the reconstruction.
6.2.1 Wildlife Crash Rate
The crash rate involving the animal-vehicle crashes is observed to increase. An increase in the animal-vehicle crash rate is observed in all seven of the reconstruction projects studied. The level of confidence for this claim is in excess of 97%. When changes in the size of the wildlife population are considered, the level of confidence is somewhat lower but still 96%.
6.2.2 Other Crash Rate
The crash rate for all crashes not involving wild animals (all crashes except animal-vehicle crashes) is observed to decrease. Five of the seven studied projects experienced this trend. The two that did not were the Hanging Rock section, located west of Cody, and the Torrington West Section, a roadway that was widened from two lanes to four lanes and connects Torrington to Lingle. The downward trend in the other crashes is confident to the 95% level.6.2.3 Total Crash Rate
The overall crash rate (all crashes including animal-vehicle crashes) for the seven sections as a whole is observed to decrease. While this trend is observed on the whole, it is only seen on four of the seven sections. In addition to the two sections that experienced an increase in the non-wildlife related crash rate (see section 6.2.2), the Astoria Section south of Jackson can be included in this case. The confidence level that the overall crash rate decreased is 87%. The lower confidence in this statement may be due, in part, to the conflicting trends between the wildlife-related crash rate and the rate of all other crashes.
6.3 Roadway Attribute Analysis
An analysis was performed to gauge the effect of several roadway design variables. These included design speed, the design speed with shoulder and lane width speed reductions, lane width, shoulder width, and overall pavement width. The design speed for the reconstructed projects was taken from the construction documents, while for the previous sections it was estimated using the horizontal and vertical curvature of the roadway. An additional variable, animal density, was used to account for changes in animal population.
Through three different tests, the only variables deemed to have a statistically significant factor in the rate of animals being hit are animal population density and design speed. When the two variables are modeled using linear regression, the adjusted R2 value (a measure of actual data points fitting the model, as well as accounting for multiple variables) is 0.45. As the maximum value of this R2 value is 1.0, only 45% of the variation of the crash data are accounted for in these two variables. While more localized animal populations, as well as a more accurate measure of true driver speed, may account for more of the variation, it is likely that other variables are important to wild animal crash rates.
6.4 Individual Analysis of Sections
An attempt to quantify the changes in crash rates for each of the individual sections was made. This was performed using a variation of the Poisson distribution. The Poisson distribution is used in situations where the events, in this case, animal-vehicle crashes, occur independently. This assumption appears to be reasonable for this data.
Few of the study sections contain crash frequencies on their own high enough to state with confidence a noticeable trend. Concerning wild animal crash rates, only the Astoria section demonstrates a high probability (98.3%) that the crash rate increase is not due to chance. In the rate of all other crashes, the Morton Pass section (99.9%), the Clearmont North section (95.5%) and the Round Mountain section (approaching 100%) demonstrate high likelihoods the decrease in rate is not due to chance. Finally, only the Morton Pass section (99.6%) and the Round Mountain (approaching 100%) section show that the decrease in total crash rate is not due to chance.
6.5 Summary of Conclusions
- ArcGIS proves valuable for the analysis and selection of high animal-vehicle crash areas and selecting potential study sections.
- Animal-vehicle crash rates are observed to increase.
- Non-wild animal-vehicle crash rates are observed to decrease.
- The total crash rates are observed to decrease.
- Animal population density and roadway design speed are significant variables in affecting animal-vehicle crash rates.
- When studying individual sections independently there is less statistical confidence in the results as opposed to looking at all seven sections in aggregate.
6.6 Recommendations for Further Study
While this study indicates that design speed is the most important variable, only seven locations are considered as a part of this research effort. It is possible the additional conclusions could be drawn regarding the impact of roadway features with a larger sample size. Two directions have been identified that may help to further clarify the situation. The first would be a continuation of this research effort, increasing the number of reconstruction projects examined. The second would be to focus on the hypothesis that vehicle speed is the primary roadway factor contributing to the rate of wild animals being hit.
The most obvious direction to take in the continuation of this research effort would be to expand upon the data already collected. The addition of more reconstruction sections would allow for more certainty in the conclusions reached, and allow for the testing of further roadway attributes.
If this direction is taken, the selection of reconstruction sections might be better served by using wild animal crash frequency, rather than rate, as the primary selection factor. This may allow for more to be concluded about specific sections of highway. While the use of crash rate as the primary choice factor allowed for the sections selected in this effort to have a wide geographical distribution, the low volume nature of many Wyoming roads lowers the usefulness of these sections in a statistical sense. This is due to the fact that it may only take a change of only a few crashes to significantly raise or lower the rate of a specific section.
If the primary focus for further examination is deemed to be confirming the role that speed plays in the wild animal crash rate, a study should be developed that eliminates other variables from consideration. A possible direction in this fashion would be to look at changes in posted speed limit. As with design speed, posted speed limit does not directly quantify the actual speed of drivers, but may function as a suitable surrogate to gauge the effect of speed on the number of wild animal crashes.
While localized changes in speed limit are rare, one case presents itself as a possible opportunity to examine large portions of the roadway system in a reasonably controlled situation. Following the removal of a nationally mandated speed limit, Wyoming raised the speed limit of the majority of its rural highway system; rural Interstate Highway speed limits were raised from 65 mph to 75 mph, while most two-lane highway speed limits were raised from 55 mph to 65 mph. In the case of Wyoming, this occurred at a very specific time: December 8, 1995. For those sections that did not experience an increase in the posted speed limit, it may be possible to use these sections as a control of the study.
While this situation presents an opportunity to examine the effect of increased speed on the number of wild animal related crashes, several difficulties must be addressed to make this study practical in nature.
- The before-after nature of this study would require data from the years before and after the change in posted speed limit. During the course of this research effort, attempts to obtain crash records from the period prior to 1995 were unsuccessful. It would be beneficial in determining trends to gain access to this data, especially the years of 1993 and 1994.
- Posting a roadway at a given speed limit does not ensure compliance with that speed. It would be beneficial to examine before and after recorded vehicular speed in several locations containing speed limit increases, to determine the actual changes in speed. The current research effort was able to find actual speeds prior to construction for only two of the seven reconstructed sections, raising concerns as to the availability of this data.
- Roadway sections that experienced reconstruction or other improvements to geometry would introduce additional variables into the exploration. Sections that had roadway improvement during either the before or after periods should be identified and removed from consideration.
If the aforementioned challenges are addressed, this investigation may provide an opportunity to determine, with some degree of certainty, what the effects of speed on wild animal-vehicle crash rates are.
References
- American Association of State Highway and Transportation Officials (AASHTO), A Policy on Geometric Design of Highways and Streets. Washington, D.C.: AASHTO, 2004
- Allen, R.E., and D.R. McCullough. 1976. "Deer-Car Accidents in Southern Michigan" Journal of Wildlife Management, Vol. 40 pp. 317-325.
- Barnum, Sarah. 2003. Identifying the Best Locations Along Highways to Provide Safe Crossing Opportunities for Wildlife. CDOT No. CDOT-DTD-UCD-2003-9 Accessed online, <http://ttap.colostate.edu/Library/CDOT/CDOT-DTD-UCD-2003-9.pdf> 15 May 2006.
- Bellis, E.D., and H.B. Graves. 1971. "Collisions of Vehicles With Deer Studied on Pennsylvania Interstate Road Section." Highway Research News, No. 43, 13-17.
- Bertwistle, J. 1999. "The Effects of Reduced Speed Zones on Reducing Bighorn Sheep and Elk Collisions with Vehicles on the Yellowhead Highway in Jasper National Park." In the Proceedings of the International Conference on Wildlife Ecology and Transportation. Missoula, MT, September 13-16, 1999, pp. 727-735.
- Bissonette J., and M. Hammer. 2001. Effectiveness of Earthen Escape Ramps in Reducing Big Game Mortality in Utah. Thesis, Utah State University. Accessed Online <http://www.cnr.usu.edu/faculty2/jbissonette/documents/MHammerthesis.pdf>
- Braden, A., R. Lopez, and N. Silvy. 2005. Effectiveness of Fencing, Underpasses and Deer Guards in Reducing Key Deer Mortality on the US 1 Corridor, Big Pine Key, Florida. FDOT No.BD-477 Accessed online, <http://www.dot.state.fl.us/research-center/Completed_Proj/Summary_EMO/FDOT_BD477_rpt.pdf> 8 July 2006.
- Curtis, P., and J. Hedland. 2005. Reducing Deer-Vehicle Crashes. Cornell Cooperative Extension, Ithaca, N.Y. Accessed online, <http://wildlifecontrol.info/ccewdmp/Publications/Deer-Vehicle_factsheet1.pdf> June 2006.
- Feeney, D., G. Beauvais, R. Coupal, S. Lanning, S. Lieske, N. Nibbelink, and K. Nordyke. 2004. "Big Game Migration Corridors in Wyoming." Open Spaces. No. B-1155. Accessed Online, <http://www.uwyo.edu/openspaces/docs/MigrationCorridors.pdf> 5 June 2006.
- Gordon, K.M., S.H. Anderson, B. Gribble, M., and Johnson. 2001. Evaluation of the FLASH (Flashing Light Animal Sensing Host) System in Nugget Canyon, Wyoming. Report No. FHWA-WY-01/03F. University of Wyoming, Wyoming Cooperative Fish and Wildlife Research Unit, Laramie, WY.
- Gordon, K. and S. Anderson. 2003. Evaluation of an Underpass Installed in US Highway 30 at Nugget Canyon, Wyoming, For Migrating Mule Deer. Report No. FHWA-WY-03/01F. University of Wyoming, Wyoming Cooperative Fish and Wildlife Research Unit, Laramie, WY.
- Gunther, K.A., M.J. Biel, and H, L. Robison. 1998. "Factors Influencing the Frequency of Road-killed Wildlife in Yellowstone National Park." In the Proceedings of the International Conference on Wildlife Ecology and Transportation. Fort Myers, FL, February 9-12, 1998, pp. 395-405.
- Hartmann, M. 2003. "Evaluation of Wildlife Crossing Structures: Their Use and Effectiveness." Wildlands CPR Website. Accessed Online, <http://www.wildlandscpr.org/resourcelibrary/reports/EvaluationByMaureenHartmann.htm>, 4 April 2007.
- Hauer, Ezra. Observational Before-After Studies in Road Safety. Kidlington, Oxford, UK: Elsevier Science, 2002.
- Insurance Institute For Highway Safety. "Many Deaths in Vehicle-Animal Collisions Are Avoidable." Insurance Institute for Highway Safety. Nov. 2004. 26 Sept. 2006. Accessed Online, <http://www.iihs.org/news/2004/iihs_news_111804.pdf> 26 September 2006.
- Jaren, V., R. Andersen, M. Ulleberg, P.H. Pedersen, and B. Wiseth. "Moose-Train Collisions: The Effects of Vegetation Removal with a Cost-Benefit Analysis." Alces, Vol. 27, 1991, pp. 93-99.
- Knapp, Keith. 2004. Deer-Vehicle Crash Countermeasure Toolbox: A Decision and Choice Resource Dept. of Civil and Envro. Engr. Univ. of Wisc., Madison, Wisconsin. Accessed Online, <http://deercrash.com/toolbox/finalreport.pdf> 10 June 2006
- Knapp, Keith. 2005. "Roadside Design Decisions and Animal-Vehicle Crashes." 3rd International Symposium on Highway Geometric Design: Compendium of Papers.
- Marcoux, A., S. Riley, G. Hickling, and S. Winterstein. 2005. Attitudes, and Knowledge Toward Deer-Vehicle Collisions in Southeastern Michigan. Presentation made on September 29, 2005. Accessed Online, <http://www.fw.msu.edu/people/riley/TWS_Final_Presentation_91.2.05.pdf> 8 June 2006.
- Olsen, Rick. 1992. Mule Deer Requirements and Management in Wyoming B-965. Department of Renewable Resources. University of Wyoming. Laramie, WY. Accessed Online, < http://ces.uwyo.edu/PUBS/B965R.pdf>
- Perrin, Joseph. 2003. Animal-Vehicle Accident Analysis. Utah Department of Transportation Research Report UT-03.31. University of Utah, Salt Lake City, Utah. Accessed Online, <http://www.udot.utah.gov/download.php/tid=296/UT-03.31.pdf> 9 June 2006.
- Pojar, T.M., D.F. Reed, and T.C. Reseigh. 1972. "Deer Crossing Signs May Prove Valuable in Reducing Accidents and Animal Deaths." Highway Research News, Vol. 46, pp. 20-23.
- Pojar, T.M., D.F. Reed, and T.C. Reseigh. 1975. "Effectiveness of a Lighted, Animated Deer Crossing Sign." Journal of Wildlife Management, Vol. 39, No. 1, pp. 87-91.
- Rea, Roy, Roy Rea, Sr. 2005. "Of Moose and Mud." Public Roads, Vol. 69, No. 2.
- Reed, D.F., T.M. Pojar, and T.N. Woodward. 1974. "Mule Deer Response to Deer Guards." Journal of Range Management No. 10, pp. 349-354.
- Reed, D.F. and T.N. Woodard. 1981. "Effectiveness of Highway Lighting in Reducing Deer-Vehicle Accidents." Journal of Wildlife Management, Vol. 45, No. 3, pp. 721-726.
- Reeve, A.F., and S.H. Anderson. 1993. "Ineffectiveness of Swareflex Reflectors at Reducing Deer-Vehicle Collisions." Wildlife Society Bulletin, Vol. 21, pp. 127-132.
- Reilly, R.E., and H.E. Green. 1974. "Deer Mortality on a Michigan Interstate Highway." Journal of Wildlife Management, Vol. 38, pp. 16-19.
- Romin, L.A., and L.B. Dalton. 1992. "Lack of Response by Mule Deer to Wildlife Warning Whistles." Wildlife Society Bulletin, Vol. 20, No. 4, pp. 382.
- Sawyer, H., and B. Rudd. 2005. Pronghorn Crossings: A Review of Available Information and Potential Options. Accessed Online <http://www.westinc.com/reports/big_game/Sawyer%20and%20Rudd%202005.pdf> 20 July 2006
- Sebesta, J.D. 2000. Design and Evaluation of Deer Guards for Florida Key Deer. Thesis, Texas A&M University.
- Utah State GIS Laboratory. Mule Deer Mapping Project. Accessed Online June, 2006 <http://www.gis.usu.edu/current_proj/muledeer.html> June 2006.
- Ward, A.L. 1982. "Mule Deer Behavior in Relation to Fencing and Underpasses on Interstate 80 in Wyoming. Transportation Research Record, Vol. 859, pp. 8-13.
- Wilson, R., and R. Karhu. 1995. Rev. 2004. "Fencing Guidelines for Wildlife." Habitat Extension Bulletin No. 53. Accessed Online <http://gf.state.wy.us/downloads/pdf/habitat/Bulletin%20No.%2053.pdf> Oct 2006.
- WYDOT (Wyoming Department of Transportation) Transportation Planning Program. Vehicle Miles 2004. Cheyenne: WYDOT, 2004.