Improving train detectability to reduce collisions with wildlife

• Author / Creator

  Backs, Jonathan A.J.

• Collisions of motorized transport with wildlife impact a wide range of species and can cause injuries and economic losses to people. On roads, vehicle collisions with animals have been studied extensively, resulting in mitigation measures that reduce collisions by segregating animals and vehicles, warning drivers about animals, or encouraging animals to be more wary of vehicles. Wildlife–train collisions have received less attention despite documented impacts to species of conservation concern. In addition, characteristics of railways may limit the feasibility of using mitigation measures designed for roads. As part of a larger initiative studying grizzly bear (Ursus arctos) mortality from train collisions in Banff and Yoho National Parks (Alberta and British Columbia, Canada), the aim of this dissertation was to determine how collisions might be reduced, particularly for grizzly bears and at locations where collisions have occurred in the past.

  I considered the problem of wildlife–train collisions as a complex systems failure at the interface of the animal, train, and railway environment systems. Within a hierarchy of causal mechanisms, I identified potential for reducing collisions by improving animal awareness of approaching trains. I approached this problem with three specific objectives: (1) to understand better the availability of acoustic signals that indicate train approach, (2) to design a warning system to alert wildlife about approaching trains, and (3) to test whether that system causes animals to leave the track earlier when a train approaches.

  For the first objective, I measured the audibility of approaching trains along sections of railway track to determine if train audibility could be predicted from features of the track environment, and I tested if poor audibility was associated with the density of recorded animal collisions. I showed that raised topography within track curves might reduce train audibility around curves. Differences in train speed and in the sound power emitted by locomotives contributed more consistently to differences in audibility within sites, while background noise from adjacent roads and rivers appeared to create differences among sites. Where the audibility of trains was lower, collisions occurred at higher densities on average. Using a physics-based model to predict train audibility along the entire railway, I found that clusters of collisions only sometimes coincided with locations of low predicted audibility; this result suggested that low train audibility is not a necessary condition for the occurrence of collision clusters.

  For the second objective, I tested multiple low-cost sensors in two configurations for their ability to detect trains, leading to the invention of an electronic system to promote wildlife avoidance of trains via associative learning. I showed that magnetic and vibration sensors could reliably detect trains as they passed, enabling my design of a warning system in which warning signals are triggered wirelessly by distant train detectors.

  For the third objective, I built working prototypes of this warning system, and I used remote cameras triggered by train approach to measure the responses of wildlife to trains where warning signals were and were not provided. I demonstrated that animals that were provided with warning signals left the track earlier than those that were not: on average, 62% earlier for larger animals (coyotes, Canis latrans, and larger) and 29% earlier for smaller animals.

  Together, my results suggest that the risk of wildlife–train collisions may be high where trains are difficult for animals to hear, that this risk could be mitigated with a train-triggered warning system, and that such a system increases the time interval between wildlife leaving the track and a train arriving at their location. With the increasing overlap between vulnerable populations of wildlife and frequent, fast-moving trains, this approach to reducing wildlife–train collisions could help to protect diverse species in locales around the world.

• Subjects / Keywords

  • Mortality
  • Acoustics
  • Sensors
  • Hearing
  • Collisions
  • Warning system
  • Flight initiation
  • Railroad
  • Vibration
  • Train detection
  • Behaviour
  • Animals
  • Technology
  • Magnetism
  • Detectability
  • Learning
  • Deterrents
  • Strikes
  • Machine learning
  • Railway
  • Associative learning
  • Wildlife
  • Audibility
  • Visibility
  • Vehicles
  • Infrared
  • Detection

• Graduation date

  Spring 2020

• Type of Item

  Thesis

• Degree

  Doctor of Philosophy

• DOI

  https://doi.org/10.7939/r3-4nee-v393

• License

  Permission is hereby granted to the University of Alberta Libraries to reproduce single copies of this thesis and to lend or sell such copies for private, scholarly or scientific research purposes only. Where the thesis is converted to, or otherwise made available in digital form, the University of Alberta will advise potential users of the thesis of these terms. The author reserves all other publication and other rights in association with the copyright in the thesis and, except as herein before provided, neither the thesis nor any substantial portion thereof may be printed or otherwise reproduced in any material form whatsoever without the author's prior written permission.