ROCA - An ArcGIS toolbox for road alignment identification and horizontal curve radii computation.

Bíl, Michal; Andrášik, Richard; Sedoník, Jiří; Cícha, Vojtěch · 2018 · DOAJ

DOI: 10.1371/journal.pone.0208407

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper introduces ROCA (ROad Curvature Analyst), an ArcGIS toolbox designed to automatically identify road alignments and compute horizontal curve radii from vector line data. The research addresses the lack of automated tools for extracting road geometry, a process that is currently time-consuming, error-prone, and often missing from national road databases. This limitation restricts the scale of traffic safety studies, which frequently rely on small sample sizes of manually identified curves. The authors aimed to develop a fully automated solution that outperforms existing semi-automated methods, thereby enabling large-scale analysis of road geometry and its impact on traffic safety. The methodology employs a naïve Bayes classifier to distinguish between tangents and horizontal curves. Unlike previous methods that relied on simple thresholds or classification trees prone to overfitting, ROCA uses six explanatory variables of road geometry, including bearing angles, cumulative angles, and radii of circumscribed and osculating circles. Probability density functions for these variables are estimated using kernel density estimation. The software was validated using data from the Czech road network, where an expert manually identified geometry for 52 secondary roads. These were split into a training set (32 roads) and a validation set (20 roads). The tool also incorporates the Douglas-Peucker algorithm for data generalization to reduce noise from digitization errors. Results demonstrate that ROCA successfully processed 9,980 km of secondary roads in under ten minutes, identifying 42,752 horizontal curves comprising approximately 43% of the network. The naïve Bayes classifier achieved a success rate of 82.4% on the validation set, outperforming the existing "CurveFinder" method by 26% in correctly identified curves. The authors applied ROCA to analyze crash modification factors (CMFs) using 56,710 traffic crashes recorded between 2009 and 2016. They found that horizontal curves with a radius of 50 meters are approximately 3.7 times more hazardous than those with a radius of 1,000 meters. The significance of this work lies in providing a freely available, automated tool that substantially increases the efficiency of road safety research. By enabling the processing of entire road networks rather than small samples, ROCA allows for more robust statistical analyses of road geometry and crash risk. The study confirms that automated extraction of geometric data is feasible and accurate, facilitating broader applications in traffic engineering and safety planning. The software and associated data are available for non-commercial use, supporting reproducibility and further research in the field.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• road geometry