Citation

Bogenreif C, Souleyrette RR, Hans ZN. J. Transp. Saf. Secur. 2012; 4(3): 179-192.

Copyright

(Copyright © 2012, Southeastern Transportation Center, and Beijing Jiaotong University, Publisher Informa - Taylor and Francis Group)

DOI

10.1080/19439962.2011.646001

PMID

unavailable

Abstract

Although crash rates on U.S. highways have decreased in recent years, nearly 34,000 fatalities were reported for 2009. Many of these deaths occur on high-speed rural roads, where crashes at curves are particularly likely. To analyze and systematically mitigate risk on highway curves, the curves must first be located and measured. This article presents a hybrid manual-computer method that uses a geographic information system (GIS)-based procedure, global positioning system (GPS) road data, circular regression, chord equations, and line simplification to identify and measure curves. Measures are validated with "as-built" design plans, and predicted safety-performance sensitivity to measurement errors is tested using Highway Safety Manual functions. Results indicate predicted performance is more sensitive to curve parameter errors for short tight curves and to errors in length than to errors in radius. The length measurement-induced errors may be reduced by analyzing tangents and curves as a whole. The impact of radius-measurement error is shown to be dependent on curve length. Although the ability of the method to measure curve parameters over a wide range of length and radius is limited, the curve identification strategy provides an efficient means to identify curves for implementation of low-cost safety improvement measures.