Modeling Nighttime Traffic Risk Through Static Infrastructure Typology and Dynamic Environmental Interactions

To address the high frequency of nighttime urban road traffic crashes and the lack of models that adequately quantify the traffic risks, this study develops a dual-layer model that couples static baseline risk with dynamic environmental corrections. Using the analytic hierarchy process and expert judgment, we determine baseline weights for nine road types: intersections (0.198), access points (0.162), curves/narrow roads (0.146), bridges (0.113), internal roads (0.095), tunnels (0.083), road segments (0.071), crosswalks (0.067), and ramps (0.066). Illumination, weather, and time are incorporated as dynamic environmental factors to enable nonlinear weight adjustment. Furthermore, total traffic risk is divided into five levels: low [0, 0.10), mild [0.10, 0.25), moderate [0.25, 0.45), high [0.45, 0.80), and critical [0.80, 1.00]. Quantitative risk analysis reveals that under extreme conditions of no illumination and heavy rain, the risk at intersections reaches 0.388, which is 1.53 times the risk under bright and clear conditions (0.254). Validation using 40,852 nighttime crash records shows that visibility (OR = 1.65) and illumination (OR = 0.56) are the most significant factors affecting crash severity, while weather (OR = 0.89) and time (OR = 0.96) show moderate effects. This study provides a basis for prioritizing lighting improvements and issuing proactive warnings for adverse weather and late-night periods.