A GIS based approach to identify road traffic fatal accident hotspots in the greater Durban city from 2011 - 2015.

Abstract

Road safety is a serious problem across the globe and its severity is more prominent in developing countries, especially in Africa. South Africa is no exception and road safety is one of the major challenges the greater Durban City in the KwaZulu-Natal Province will have to overcome in order to achieve its vision of being Africa’s most livable city by 2020. The prime goal of road safety is to reduce the number and severity of traffic accidents by identifying, implementing and evaluating measures to improve road safety. Road safety improvements are supposed to be applied to accident hotspots where they have the most significant impact, thus, identification of hotspots is an essential step in safety management. This study employed GIS spatial statistics tools to examine and map the trends and spatiotemporal distribution of fatal accident hotspots in the greater Durban city from 2011 - 2015. A total of 546 fatal accidents occurred that led to 594 fatalities. It was observed that the highest number of fatal accidents occurred in 2011, a year which marks the beginning of the UN Decade of Action for Road Safety. However, there was a declining trend from 2011-2013, with 2013 recording the lowest fatal accidents. Since 2013, fatal accidents saw an increasing trend until 2015. The majority (52.4%) of fatal accidents occurred on weekends, while 39.2% were recorded in March, May, June and December. Furthermore, the time period between 18:00-23:59 recorded the highest number of fatal accidents (33%). Ninety two and a half percent (92.5%) of fatal accidents resulted from human error, 67.9% of these resulted from a vehicle-pedestrian collision due to pedestrians entering the roadway when unsafe to do so. Dual carriageways accounted for 30.4% of the total fatal accidents, followed by freeways (26.7%) and single carriageways (25.8%). Spatial clustering of fatal accidents and spatial densities of fatal accident hotspots were evaluated using Global Moran's I and Anselin Local Moran’s I spatial autocorrelation, Kernel Density Estimation (KDE) and Getis-Ord Gi* statistics respectively. Moran’s I and Anselin Local Moran’s I show that the occurrence of fatal accidents were random, with the exception of 2012 which had a 1% less likelihood that clustering occurred. Ranking of hotspots using KDE was done with the help of pixel values of the observed locations while ranking of hotspots using Getis-Ord Gi* was done with the help of z-scores associated with statistical significance. Results of hotspot analysis delineated 22 hotspot locations at 90% - 99% confidence levels. Based on the Getis-Ord Gi* hotspot results, the N2 freeway is of critical concern as it appears consistently throughout the five-year period under study and accounting for 49.2% (29) of the total recorded fatalities. Fatal accident hotspots in the study area follow a Northeasterly-Southwesterly trend as determined by the directional distribution (standard deviation ellipse) function in ArcMap. The results can be effectively utilised by various agencies for adopting better planning and management strategies to reduce and prevent road traffic fatal accidents in the greater Durban City.