Development and Evaluation of Geostatistical Methods for Estimating Weather Related Collisions: A Large-Scale Case...

Area of Science:

• Geostatistics
• Transportation Safety
• Road Infrastructure

Background:

• Winter driving conditions increase collision risks, necessitating road safety measures.
• Road authorities prioritize hazardous areas and collision hotspots for improved mobility and accessibility.
• Winter collision (WC) statistics, specifically the WC to total collision ratio, can identify high-risk road segments.

Purpose of the Study:

• To introduce and evaluate regression kriging (RK) as a geostatistical method for large-scale winter collision hot spot analysis.
• To assess the effectiveness of RK in modeling winter collision ratios using auxiliary variables and spatial autocorrelation.
• To determine the robustness and feasibility of RK for statewide implementation in road safety management.

Main Methods:

• Utilized regression kriging (RK), a geostatistical technique incorporating auxiliary variables and spatial autocorrelation.
• Applied RK to analyze winter collision ratios across the northeast quarter of Iowa over five winter seasons (2013/14–2017/18).
• Evaluated RK model performance using statistical measures: mean squared error, root mean square error, and root mean squared standardized error.

Main Results:

• Regression kriging proved highly effective in modeling winter collision ratios.
• The study demonstrated the robustness and feasibility of RK for analyzing road collision data.
• RK provides a powerful framework for understanding contributing factors to winter collisions.

Conclusions:

• Regression kriging is a valuable and effective tool for identifying and analyzing winter collision hot spots.
• The methodology supports data-driven prioritization of road segments for safety interventions.
• RK offers a scalable and reliable approach for statewide road safety assessments.