Accident analyses in Hong Kong : accident blackspot identification, casualty injury severity and before-after analysis

Abstract

Road safety is a major concern because of the social and economic costs resulting from traffic crashes. Numerous studies investigating traffic collisions and the resulting costs have been conducted. One important research topic in these studies is accident blackspot identification. The measures of accident frequency and accident rate are commonly adopted as identification criteria. However, identifying accident blackspots based solely on accident frequency or accident rate has been found, in practice, to be inefficient as neither can correctly reveal the extent of accident consequences. For example, high accident frequency does not necessarily result in a large number of fatalities or serious injuries. To supplement the knowledge gained by previous researchers, a new method to rank accident blackspots is proposed in this study. In the proposed blackspot identification method, instead of accident frequency, accident consequences (in terms of injury or accident costs) are considered when identifying accident blackspots. The merit of the proposed method is that it takes not only number of injuries (or accident frequency) but also injury severity (or accident severity) into the consideration. To illustrate the proposed method, a case study was carried out using Hong Kong traffic accident data. The results indicate that adopting accident consequences, such as injury costs, can identify accident blackspots with higher injury costs but the methods of using accident frequency only or the Hong Kong Transport Department's blackspot definition may not be guaranteed. In view of the importance of casualty injury costs as regards identification of accident blackspots, this feature is further investigated in this study. Focus is on an analysis of the effects of various contributory factors (categorized by environmental, site, and vehicle factors) on the injury severity of driver, passenger and pedestrian casualties and on accident costs in Hong Kong. Binary logistic regression model is adopted to quantify the associations between injury severity of casualties and the contributory factors, while linear regression model is used for modeling the effects of contributory factors on accident costs. A Hong Kong traffic accident dataset for the whole territory from 2007 to 2009 is used in this study to estimate the coefficients of the regression models. The results of the regression models reveal that accident time, rain conditions, speed limits, traffic congestion levels, road types, and vehicle types significantly affect casualty injury severity and accident costs. Each contributory factor has a different effect or a different degree of effect on driver/passenger and pedestrian casualties. A Before-After analysis is also used for investigation of accident effects on traffic speed in this study. Factors affecting the accident effects on traffic speed are firstly identified. Regression models are calibrated with empirical data to quantify the influences of factors on the degree of accident effects. A case study is carried out in which there is a total of 313 accidents occurred on a local urban area of Hong Kong during the study period: from September 2009 to December 2010 together with the corresponding speed data before and after the occurrence of these accidents. From the results of the case study, three factors, namely accident severity, accident time and accident location, were found to significantly affect the degree of accident effects on traffic speed.