Spatio-temporal modeling of the impact of climate change on road accidents: A Case Study of New Brunswick

The objective of this research is to study the impact of climate change on the hazardous weather-related road accidents. The New Brunswick province of Canada is considered as a case study. The study uses road accident data collected from police accident reports for the period of 1997-2007. The climate change modeling uses thirty-year weather records of seven climate zones of New Brunswick, National Centers for Environmental Prediction (NCEP) re-analysis dataset, and large-scale simulation data from the Canadian Global Circulation Model, General Climate Model, and Coupled Global Climate Model (CGCM3). The large-scale simulation data from CGCM under SRES-A2 scenario during 21st century are used to model the climate in the future. This study develops an Exposure to Weather-Accident Severity (EWAS) index and estimate the relationship between EWAS index and weather-related explanatory variables of road accidents by applying negative binomial regression and Poisson regression models. The regression models find out that surface-weather condition, weather-driver’s gender, weather-driver’s age, weather-driver’s experience and weather-vehicle’s age have strong positive correlation with EWAS index. The surface-road alignment and surface-road characteristics have negative relationship with EWAS index. The spatial pattern of EWAS index with respect to weather-related explanatory variables is examined for the fifteen census divisions of New Brunswick province, which derives similar relationships. The climate change modeling estimates that the number of rainy days may increase in all climate zones and the number of snowy days and freezing days may decrease or stay the same in most of the climate zones during three 30-year periods of 21st century (i.e. 2011-2040, 2041-2070, 2071-2100). The findings of this study imply that more hazardous weather in future will result in increased accident severity.