Research On Identifing Hotspots In The Urban Road Networks Based On The Network Kernel Density Estimation Method

According to the statistical data, traffic accidents has been the major issue which dangers human health and social development. It is also the main reason that lead to death in our country. While the hazardous road segments are accident-prone locations, they should be the main targets that need first considerations in the safety improvement projects. However, the existing hazardous road segments identification methods demand higher-standard traffic data so that it is hard to been carry out in the practical project. Besides the identification results are showed by the words or forms which are indirectly understand to others.Therefore, Kernel Density Estimation Method is applied to the research on the hazardous road locations in the urban road networks in this paper. The Kernel Density Estimation Method is a nonparametric estimation theory in the space analysis theory. It only needs to know the locations that the traffic accidents happen, while its identification results can be showed in the GIS platform by the three-dimensional form. Firstly, two main parameters in the Kernel Density Estimation Method-bandwidth and kernel functions were introduced in this paper. Then this method was modified to adjust to the hazardous road locations identification restricted by the urban road networks. This paper then evaluated the effectiveness of the identification results by the modified Network Kernel Density Estimation Method. The effectiveness tests indicate that the Kernel Density Estimation Method performed better than the traditional hazardous road locations identification methods, and can be promoted in the practical projects.Then the identification results have been shown in the ArcGIS Desktop which includes build up the road traffic accident database, divide the road networks into units, compute the kernel values of the unit nodes and identify the hazardous road locations based on the cells’ kernel density values.The network kernel density estimation method used in this paper to identify hotspots in the urban road networks can contribute to the urban road safety management by supplying the technical basis.