Express highway embankment distress and occurring probability in permafrost regions on the Qinghai-Tibet Plateau

Abstract

Under the influence of climate warming and increasing human activities, permafrost degradation has become severe, impacting the stability of local infrastructure. This is particularly evident in the Qinghai-Tibet Plateau, where the fragile permafrost environment, high ice content and high ground temperature exacerbate the situation. The Gonghe-Yushu Express Highway (GYE), a newly constructed express highway on the plateau, was completed and opened to traffic five years ago. However, even for such a new express highway, engineering distresses have occurred. In this study, we employed ground penetrating radar, unmanned aerial vehicle, and manual investigations to examine the types and distribution characteristics of embankment distresses, analyzed the impacts of different factors, and established a probability prediction model for the occurrence of embankment distresses. Our findings revealed that the main types of embankment distresses on the GYE are wavy pavement, uneven settlement, and cracks, with the highest proportion of distresses in waterlogged sections on plateaus and basins. Among the nine factors affecting embankment distresses considered in this study, permafrost ice content and soil type have relatively larger impacts. Compared to logistic regression and support vector machines, the random forest model demonstrates higher accuracy in predicting embankment distress occurrence in permafrost road sections, with a prediction accuracy rate of 86.4%. Based on the random forest model, the permafrost road sections of the GYE are divided into five levels according to the probability of embankment distress occurrence: slight, low, moderate, high, and extremely high, with corresponding embankment lengths of 264, 61, 41, 34, and 29 km, respectively.