Leakage-Aware Cross-Distress Modeling of Extreme-Weather Effects on Asphalt Pavements: A Controlled Multi-Output Benchmark for Cracking, Rutting, and Roughness
Academic Journal of Applied Sciences · 2026
DOI: 10.54097/4jkq3v27 ↗Kurz gefasst
Fachartikel in der Zeitschrift Academic Journal of Applied Sciences, 2026. Verfasst von Connor Ellison. Die Arbeit nennt 39 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).
Abstract (Originalsprache)
Extreme weather influences pavement deterioration through distinct thermal, moisture, and traffic-mediated mechanisms, yet many data-driven studies model one distress at a time and evaluate repeated section observations with random train-test splits. This paper develops a leakage-aware cross-distress framework for cracking, rutting, and International Roughness Index (IRI). The framework separates current-condition memory from incremental weather information, quantifies distress-specific weather-response fingerprints, and tests whether cross-distress state variables improve short-horizon forecasting. Because a directly downloadable three-distress field panel with aligned extreme-weather indicators was not available in the present reproducibility workflow, the numerical experiment uses a fully disclosed, seed-controlled synthetic longitudinal benchmark rather than presenting synthetic records as observations. The benchmark contains 350 virtual pavement sections, four climate regimes, 3,500 section-year transitions, and mechanistically motivated nonlinear interactions. Under five-fold GroupKFold evaluation by section, ridge regression attained R² values of 0.978, 0.990, and 0.986 for next-year cracking, rutting, and IRI, respectively; HistGradientBoosting achieved 0.975, 0.986, and 0.983. Removing weather reduced HistGradientBoosting R² by 0.0041 for cracking, 0.0041 for rutting, and 0.0014 for IRI, showing that current condition dominates one-step prediction while weather still contributes distinct residual information. Permutation analysis recovered physically coherent signatures: heavy precipitation was the leading weather term for cracking, heat and truck traffic dominated rutting, and precipitation was the strongest weather term for IRI. The proposed evaluation protocol and diagnostic metrics are intended as a reproducible bridge to future LTPP multi-distress field validation.
Zitieren
APA
Ellison, C. (2026). Leakage-Aware Cross-Distress Modeling of Extreme-Weather Effects on Asphalt Pavements: A Controlled Multi-Output Benchmark for Cracking, Rutting, and Roughness. Academic Journal of Applied Sciences. https://doi.org/10.54097/4jkq3v27
Vancouver
Ellison C. Leakage-Aware Cross-Distress Modeling of Extreme-Weather Effects on Asphalt Pavements: A Controlled Multi-Output Benchmark for Cracking, Rutting, and Roughness. Academic Journal of Applied Sciences. 2026. doi:10.54097/4jkq3v27
BibTeX
@article{Ellison2026,
author = {Ellison, Connor},
title = {Leakage-Aware Cross-Distress Modeling of Extreme-Weather Effects on Asphalt Pavements: A Controlled Multi-Output Benchmark for Cracking, Rutting, and Roughness},
journal = {Academic Journal of Applied Sciences},
year = {2026},
doi = {10.54097/4jkq3v27}
}
EUVIMED