Hierarchical Bayesian model for joint prediction of runway pavement metrics considering measurement uncertainty
- 1 December 2025
- journal article
- Published by Canadian Science Publishing in Canadian Journal of Civil Engineering
- Vol. 52 (12) , 2437-2456
Abstract
Accurate prediction of runway pavement conditions is critical for aviation safety and maintenance planning. This study presents a Hierarchical Bayesian Joint Model with latent variables to simultaneously forecast key performance metrics—the International Roughness Index (IRI) and Surface Macrotexture Depth (SMTD)—while explicitly accounting for measurement uncertainties. The proposed model incorporates nonlinear quadratic relationships among axial loads, SMTD, and IRI, effectively capturing both direct and indirect load effects. Model performance was rigorously evaluated through a stratified five-fold cross-validation, achieving mean absolute errors as low as 0.98 for IRI and 0.88 for SMTD, outperforming traditional methods by approximately 15%. Posterior diagnostics confirmed robust convergence and accurate uncertainty quantification. Overall, the hierarchical Bayesian model demonstrated superior predictive accuracy, highlighting its practical utility for data-driven pavement management decisions.Keywords
Funding Information
- Yarmouk University (5/2024)
This publication has 17 references indexed in Scilit:
- Measurement of pavement unevenness and evaluation through the IRI parameter – An overviewMeasurement, 2022
- An ensemble learning model for asphalt pavement performance prediction based on gradient boosting decision treeInternational Journal of Pavement Engineering, 2021
- A hybrid wavelet-optimally-pruned extreme learning machine model for the estimation of international roughness index of rigid pavementsInternational Journal of Pavement Engineering, 2020
- International Roughness Index prediction model for flexible pavementsInternational Journal of Pavement Engineering, 2018
- Framework for Development and Comprehensive Comparison of Empirical Pavement Performance ModelsJournal of Transportation Engineering, 2015
- Evaluating Learning Algorithms2011
- Accounting for measurement error: A critical but often overlooked processArchives of Oral Biology, 2009
- Model Inference and Averaging2008
- Another look at measures of forecast accuracyInternational Journal of Forecasting, 2006
- Estimation of Pavement Performance Deterioration Using Bayesian ApproachJournal of Infrastructure Systems, 2006