Predicting black ice-related accidents with probabilistic modeling using GIS-based Monte Carlo simulation

Area of Science:

• Road safety and transportation engineering
• Environmental science and meteorology
• Computational modeling and simulation

Background:

• Black ice is a hazardous road condition, difficult to detect due to its transparency.
• Accurate prediction of black ice accidents requires accounting for spatial, weather, and traffic uncertainties.
• Existing methods often rely solely on road temperature, lacking comprehensive risk assessment.

Purpose of the Study:

• To develop a probabilistic model for predicting black ice accident likelihood.
• To identify and analyze the primary trigger factors influencing black ice accidents.
• To provide a decision-support tool for traffic authorities to mitigate risks.

Main Methods:

• Utilized Monte Carlo simulation with random values to model black ice accidents.
• Integrated spatial, weather, and traffic data for probabilistic predictions.
• Employed sensitivity analysis within Monte Carlo simulations to determine trigger factor significance.
• Visualized accident probability maps using a geographical information system (GIS).

Main Results:

• The average black ice accident probability was calculated at 0.0058 (SD=0.001).
• Sensitivity analysis identified wind speed (0.354), air temperature (0.270), and road angle (0.203) as significant triggers.
• The study successfully mapped black ice accident probabilities across road sections.
• A method for evaluating black ice accident risk beyond simple temperature prediction was established.

Conclusions:

• Monte Carlo simulations offer a robust framework for predicting black ice accident probabilities by decoupling meteorological and traffic factors.
• Identified key environmental and road geometry factors provide actionable insights for targeted safety interventions.
• The developed methodology can enhance road safety management by pinpointing high-risk locations and informing proactive countermeasures.