Spatial Analysis of Weather Crash Patterns

Abstract

Spatial statistical techniques can be an effective tool for analyzing patterns and autocorrelation in crash data, especially weather-related crashes. Since weather is a geographic phenomenon, it tends to show distinct geographic patterns affecting certain locations more than others. Accordingly, "weather-related" crashes may also display similar distinct patterns or clustering. The objective of this research was to use spatial statistical techniques to identify significant patterns of weather-related crashes. Weather-related crashes, defined as those crashes which occurred in adverse weather conditions, were analyzed using the Getis-Ord G i * ͑d͒ statistic. The statistic reveals spatial patterns for weather-related crashes which are clustered at different locations depending upon weather conditions ͑snow, rain, and fog͒. The results also show geographic areas ͑counties͒ of statistically significant high and low relative crash rates for each weather condition. Furthermore, the resulting patterns of crashes were validated by comparing counties of high and low crash rates with areas of varying weather data. The establishment of this relationship between weather and crashes is imperative in identifying the variables contributing to these crash types and the implementation of effective countermeasures for road weather safety audit purposes.

Application of Geographically Weighted Regression Technique in Spatial Analysis of Fatal and Injury Crashes

“…Auto-logistic models, Conditional Auto-regression (CAR) models, Simultaneous Auto-regression (SAR) models, spatial error models (SEM), Generalized Estimating Equation (GEE) models, Full-Bayesian Spatial models, Bayesian Poisson-lognormal models are some of the most employed techniques to conduct spatial modeling in traffic safety (e.g. in Flahaut 2004;Guo et al 2010;Huang et al 2010;Khan et al 2008;Kweon and Lim 2012;Ossenbruggen et al 2010;Quddus 2008;Sadia and Polus 2013;Siddiqui et al 2012;Wang et al 2009;Wang and Abdel-Aty 2006). The output of these models are still fixed variable estimates for all locations, however, the spatial variation is taken into account.…”

Rainfall impacts on traffic safety: rain-related fatal crashes in Texas

“…The GetisÀOrd G i à tool was selected for this analysis because it can reveal significant spatial clusters of high values (hot spots) and low values (cold spots) county by county. Khan et al (2008) used the GetisÀOrd G i à tool for the spatial analysis of weather crash patterns and this research uses a similar methodology. The G-statistic was used to measure the high/low clustering of the rain-related crashes.…”

“…Because elements lying in space are affected by a topographical system and usually have a degree of spatial dependency, performing a spatial autocorrelation analysis can lead to severe inaccuracy in statistical examination of results of the data (Getis & Ord 1992;Cressie 1993). Spatial statistical techniques can be used to examine patterns and autocorrelation in weather-related crashes (Khan et al 2008). Using geographic information systems (GIS), Nokhandan et al (2008) studied how environmental factors impacted road accident frequency.…”

A Selection of High Pedestrian Accident Zones Using Traffic Accident Data and GIS: A Case Study of Seoul

“…이러한 사업을 진행할 때, 어떤 지점에 안전 대책을 우 선적으로 수립하여야 교통안전을 크게 개선할 수 있는지 파 악하는 것은 매우 중요하다 (Khan et al, 2008). 많은 연구들이 교통사고 핫스팟 선정에 KDE를 사용하였 으며 (Anderson, 2009;Flahaut et al, 2003;Plug et al, 2011;Prasannakumar et al, 2011;Truong and Somenahalli, 2011;Vemulapalli, 2015), 일부 연구들은 KDE를 시행하여 보행자 사고 핫스팟을 선정하였다 (Blazquez and Celis, 2013;Jang et al, 2013;Loo et al, 2011;Pulugurtha et al, 2007;Rankavat and Tiwari, 2013 일부 연구들은 Getis-ord Gi*를 시행하여 교통사고 핫스팟 을 선정하였다 (Khan et al, 2008;Kingham et al, 2011;Kuo et al, 2011;Manepalli et al, 2011;Vemulapalli, 2015). 특히, Getis-ord Gi*를 사용한 기존 연구들은 Getis-ord Gi*로 …”