Explainable friction-based modeling of winter runway contaminants for GRF-compliant decision support

Runway contamination in winter conditions remains a major safety concern for aircraft ground operations, and the implementation of the Global Reporting Format (GRF) has increased the need for objective tools to complement visual runway surface assessment. This study proposes a GRF-compliant friction-based framework using data collected during a large-scale experimental campaign conducted in January 2024 at Montréal/Mirabel Airport under controlled and repeatable winter contaminant conditions. A physically interpretable penalized piecewise regression model was first developed to describe the longitudinal friction coefficient measured by IMAG (LFC C 1 ) as a function of speed, slip rate, and contaminant type, and an XGBoost multiclass classifier was then trained to identify winter contaminants directly from friction measurements. The classifier achieved an overall accuracy of 0.78, which increased to 0.84 when the absolute difference between friction and drag-related coefficients was included, with particularly strong performance for the most safety-critical contaminants such as ice and wet snow. These results show that operational skid-resistance measurements can provide objective support for contaminant identification and complement visual inspection in GRF-based winter runway condition assessment.