Spatio-temporal Modeling of Traffic Risk Mapping on Urban Road Networks

Abstract

Over the past few years, traffic collisions have been one of the serious issues all over the world. Global status report on road safety, reveals an increasing number of fatalities due to traffic accidents, especially on urban roads. The present research work is conducted on five years of accident data in an urban environment to explore and analyze spatial and temporal variation in the incidence of road traffic accidents and casualties. The current study proposes a spatio-temporal model that can make predictions regarding the number of road casualties likely on any given road segments and can generate a risk map of the entire road network. Bayesian methodology using Integrated Nested Laplace Approximation (INLA) with Stochastic Partial Differential Equations (SPDE) has been applied in the modeling process. The novelty of the proposed model is to introduce "SPDE network triangulation" precisely on linear networks to estimate the spatial autocorrelation of discrete events. The result risk maps can provide geospatial baseline to identify safe routes between source and destination points. The maps can also have implications for accident prevention and multi-disciplinary road safety measures through an enhanced understanding of the accident patterns and factors.