To reduce the occurrence of motor-vehicle crashes, professionals in education, enforcement, and engineering are continually tasked with implementing safety solutions. Identifying locations of high rates of crashes allows safety solutions to more adequately target their intended audience. This research examines advances in identifying hot spots of motor-vehicle crashes. These advancements come from improving: 1) the calculation of spatial autocorrelation and interpolation, 2) the identification of spatiotemporal patterns, and 3) the influence of geographical patterns on the spatial distribution of crashes. Overall, by improving the hot spot analysis, concerned professionals may be better prepared and lower the number of alcohol-related crashes. The location of hot spots is important in the implementation of enforcement campaigns. A lapse in accuracy may allow a vehicle operator suspected of disobeying traffic laws from being properly disciplined. Improvements in the calculation of spatial autocorrelation and interpolation result from the use of network distances instead of Euclidean based distances. Network based distances increase the accuracy of resulting hot spots. With the accuracy of hot spots improved, the optimal times to implement safety campaigns in their identified areas become important. Many hot spots purely analyze crashes as if they all occurred at the same time. By investigating crashes in this manner, some key influences may be lost and the efficiency of the implemented campaign may be reduced. Spatio-temporal hot spots are examined and show that as time progresses, clusters of crashes occur and disappear throughout space. By moving campaign sites as the location of crashes move, the overall efficiency of campaign tactics would benefit. Hot spots of crashes have continually been scrutinized for their focus on areas of large populations. In an effort to rectify this belief, the normalization of hot spot is examined in relation to population density. It is found that the strict use of population density provides unfavorable results. Instead, the identification of hot spots through either the frequency or societal crash costs varies the resulting hot spot location. Using crash frequency allows for high visibility/mass target campaigns to best be realized. Meanwhile, the use of societal costs best targets high valued crash occurrences. The use of hot spots may be beneficial in improving campaigns to reduce alcohol-related crashes. Once the hot spot maps are created, this research uses the results to develop a new method of patrolling for intoxicated drivers. The hot spot maps are broken down into local indicators of spatial association, which show statistically significant locations where intoxicated drivers are likely to be present. Route optimization models are then used to guide officers to these locations. These models are compared with traditional methods of corridor patrolling through a series of performance metrics. Failure probability models are then created to further compare the two methods of patrolling, as well as aiding captains of jurisdictions in decision-making processes. By utilizing location-based hot spots, new methodologies of patrolling may be developed in order to reduce the amount of alcohol-related crashes. This new method of patrolling will guide officers to statistically significant locations, allowing them to be more accurate while patrolling for intoxicated drivers. Additionally, this method proves to pass through more alcohol-related crash locations per minute and mile, indicating it may be more efficient than current practices of patrolling. By improving how officers patrol, people may more accurately be deterred from driving intoxicated and alcoholrelated crashes may be ultimately reduced. (Author/publisher)
Samenvatting