Speed Variation is accepted to be closely associated with traffic crashes. There is, however, a dearth of research in speed variation related crashes on urban arterials in general and in the context of developing countries in specific. This gap is a due to the lack of the required infrastructure in the urban areas. In addition to the lack of infrastructure, there is also a lack of a measure capable of using a one-dimensional data (speed) to detect crashes in the absence of the mentioned data collection methods.
This study shows a conceptual measure capable of calculating speed variation to predict crash frequency. Speed observations from road segments using conventional or non-conventional methods are plotted on a histogram. This histogram reveals the travel speed regimes. While past studies have shown a single speed regime with a normal distribution for such histograms, current studies have observed and accepted multi-nodal distributions which arise primarily due to the heterogeneity of vehicle type and the differences in nature of travel. The authors hypothesize that the measurement of variation in the multi-nodal distribution can act as a predictor variable to traffic crashes. This study observed that there exists up to three speed regimes, which means that groups of vehicles tend to travel at statistically different speeds. This relationship may be captured as the ratio of the size of each node to the distance between the nodes.
The finding of this study has the potential to predict the crash frequency relying solely on the speed of vehicles. This one-dimensional measure to predict crash frequency is important as crash studies generally rely on multiple factors to predict the frequency of crashes. The conceptual measure developed in this study is capable of quantifying the effect of these various factors which affect speed and in turn lead to traffic crashes.
Roshan Jose, Indian Institute of Technology Kharagpur; Sudeshna Mitra, Indian Institute of Technology Kharagpur
