Comparative analysis of zonal systems for macro-level crash modeling.

Area of Science:

• Transportation Engineering
• Spatial Analysis
• Traffic Safety

Background:

• Macro-level traffic safety analysis lacks clear guidelines for selecting appropriate spatial configurations.
• Existing zonal systems like census tracts (CTs) and state-wide traffic analysis zones (STAZs) may not be optimal for crash modeling.

Purpose of the Study:

• To comparatively analyze the performance of different zonal systems for macroscopic crash modeling.
• To determine the optimal zonal system for traffic safety analysis.

Main Methods:

• Developed Poisson lognormal models for total, severe, and non-motorized crashes using CTs, STAZs, and traffic analysis districts (TADs).
• Compared model performance across zonal systems with and without spatial autocorrelation using a grid-based framework.
• Evaluated model prediction accuracy within grids of varying sizes.

Main Results:

• Models based on TADs consistently outperformed models based on CTs and STAZs for all crash types.
• Models incorporating spatial autocorrelation demonstrated superior performance compared to those without it.

Conclusions:

• Traffic analysis districts (TADs) are recommended for transportation safety planning.
• Census tracts (CTs) are suitable for socio-demographic data collection, and STAZs for transportation demand forecasting.
• The findings aid practitioners in selecting optimal zonal systems for effective traffic safety policies.