Evaluating the Shared Structure of Pavement Deterioration Across Multiple Pavement Indicators

Pavement management systems commonly evaluate network condition using a single indicator, most often the International Roughness Index (IRI), although pavement deterioration is expressed through multiple structural and surface distress mechanisms. This study develops a multi-indicator framework combining treatment-adjusted temporal analysis, cross-correlation analysis, and principal component analysis (PCA), applied to 165,372 section-year observations of IRI, rutting depth, structural cracking density, and functional cracking density on 9739 sections in Baltimore County, Maryland, from 2008 to 2024. After excluding observations adjacent to documented treatments, median IRI declined from 110.58 to 103.77 in/mi while rutting and both cracking indicators increased, and functional cracking worsened in 57.7% of directly observed between-treatment intervals. Most of the variation in IRI lay between sections, whereas most of the variation in cracking occurred within sections over time. Cross-correlation peaked at lag zero for every indicator pair in condition levels and after detrending and pre-whitening, and for all pairs except rutting and functional cracking after differencing; the only consistent lead–lag asymmetry, a small tendency for rutting to precede the other indicators, was much weaker than the contemporaneous association. The first principal component (PC1) explained 49.5% of total variance, to which all four indicators contribute substantially (20.7% to 29.4%); this structure was stable under treatment adjustment, across sub-periods, and when variation between and within sections was analyzed separately. Cracking contributed more strongly to the axis on Intermediate roads and IRI on High-Order roads, a difference consistent in direction across all sensitivity analyses. PC1 and an existing minimum-based remaining-service-life measure (RSL_4) agreed on fewer than half (45.2%) of the observations either flagged as high priority, with PC1 favoring multi-indicator distress profiles. Single-indicator and minimum-based approaches therefore capture different aspects of pavement condition, supporting class-sensitive, multi-indicator assessment for network-level pavement management.