The configuration of the urban built environment is critical for promoting sustainability and achieving carbon neutrality. However, existing studies mostly use linear and spatial econometric models to investigate the relationship between urban built environments and traffic carbon dioxide (CO₂) emissions, in-depth studies exploring the heterogeneous impacts of related features on traffic CO₂ emission by interpretive machine learning models are scarce. Hence, we extract four dimensionless features to depict the size, compactness, irregularity, and isolation of built-up areas, and road network-related features (i.e., average cluster coefficient, road topological density, and road geometric density), respectively. Subsequently, we develop an interpretive machine learning framework based on the extracted features related to the urban built-up areas and road networks. The interpretive results of the proposed framework uncover that urban morphological features, especially population density (POP), GDP per capita (GDPpc), and urban physical compactness (UPC), have a heterogeneous impact on the per capita traffic emission (PCCE) across different cities. GDPpc is more like a linear relationship with PCCE, and UPC has a significant influence on PCCE when its value is between 62% and 78%. Our results also reveal the nonlinear relationships and interactive effects between these features, providing the implications of urban morphological planning and carbon emission reduction.

@en

Traffic barrier crashes have been a major concern in many prior studies in traffic safety literature, especially in the crash-prone sections of mountainous regions. However, the effect of factors affecting the injury-severities resulting from crashes involving different types of traffic barriers may be different. This paper provides an empirical assessment of the performance of ordered and unordered discrete outcome models for examining the impact of exogenous factors determining the driver injury-severity of crashes involving two types of traffic barriers in mountainous regions: w-beam barriers and cable barriers. For the ordered framework, the alternative modeling approaches include: the generalized ordered logit (GOL) and the random thresholds random parameters generalized ordered logit model (RTRPGOL). Whereas, for the unordered framework, the alternative modeling approaches include: the multinomial logit (MNL), the random parameters multinormal logit (RPL), and the random parameters multinormal logit model with heterogeneity in the means and variances (RPLHMV). Using injury-severity data from 2016 to 2019 for mountainous regions in Guiyang City, China, three injury-severity categories are determined as outcome variables: severe injury (SI), minor injury (MI), and no injury (NI), while the potential influencing factors including drivers-, vehicles-, road-, and environment-specific characteristics are statistically analyzed. The model estimation results show: (a) that the MNL model statistically outperforms the GOL model in terms of goodness-of-fit measures; (b) the RTRPGOL model is statistically superior to the MNL and RPL models; and (c) the RPLHMV model is statistically superior to the RTRPGOL model, and therefore the preferred option among the model alternatives. To that end, the RPLHMV model is leveraged to quantitatively describe the impact of explanatory variables on the driver injury-severity and explore how these factors change over the years (between 2016–2017 and 2018–2019). The results further show that the factors affecting driver injury severities and the effects of significant factors on injury severity probabilities change across traffic barrier crash models and across years. In addition, the results of the temporal effects analysis show that some variables present relative temporal stability, which is important for formulating long-term strategies to enhance traffic safety on mountainous roads. Most importantly, the effects of the explanatory factors that exhibit relative temporal stability are found to vary across traffic barrier crashes. For example, trucks, daylight, curved section segments, and high-speed limit (greater than 55 mph) are some of the factors that have opposite effects between traffic barrier crash models. The findings from this paper are expected to help policy makers to take necessary measures in reducing traffic barrier crashes in mountainous regions by forming appropriate strategies, and by allocating properly their available resources at the pre-planning phase.

@en

Knowledge of the hierarchical organization of urban heavy truck flows is important for understanding the structure of urban freight system and underlying interactions dynamics, providing insights to assess and develop freight policies. The complexity and dynamic nature of urban freight system pose significant challenges in comprehensively capturing structured arrangement of heavy truck movements. In this paper, we uncover the hierarchical organization of urban heavy truck flows by using complex network theory. We use large-scale heavy truck GPS data and urban freight location point-of-interest (POI) data to construct urban heavy truck mobility networks, and detect their community structure. The empirical results suggest different sets of locations are closely linked to each other to form multiple clusters. By integrating the categories of locations, we reveal the cluster-specific industry concentration and industry-specific location roles, informing evidence-based policy formulation. To capture the interaction dynamics of locations, we develop a spatial network growth model that considers the spatial agglomeration of industrial clusters and interaction pattern of locations. The model provides a mathematical tool to simulate the formation process of real-world networks for logistics planning and management.

@en

Truck-involved crashes, especially truck-car crashes, are associated with serious and even fatal injuries, thus necessitating an in-depth analysis. Prior research focused solely on examining the injury severity of truck drivers or developed separate performance models for truck and car drivers. However, the severity of injuries to both drivers in the same truck-car crash may be interrelated, and influencing factors of injury severities sustained by the two parties may differ. To address these concerns, a random parameter bivariate probit model with heterogeneity in means (RPBPHM) is applied to examine factors affecting the injury severity of both drivers in the same truck-car crash and how these factors change over the years. Using truck-car crash data from 2017 to 2019 in the UK, the dependent variable is defined as slight injury and serious injury or fatality. Factors such as driver, vehicle, road, and environmental characteristics are statistically analyzed in this study. According to the findings, the RPBPHM model demonstrated a remarkable statistical fit, and a positive correlation was observed between the two drivers' injury severity in truck-car crashes. More importantly, the effects of the explanatory factors showing relatively temporal stability vary across different types of vehicle crashes. For example, car driver improper actions and lane changing by trucks, have a significant interactive effect on the severity of injuries sustained by drivers involved collisions between trucks and cars. Male truck drivers, young truck drivers, older truck drivers, and truck drivers' improper actions, elevate the estimated odds of only truck drivers; while older car and unsignalized crossing increase the possibility of injury severity of only car drivers. Finally, due to shared unobserved crash-specific factors, the 30-mph speed limit, dark no lights, and head-on collision, significantly affect the severity of injuries sustained by drivers involved in collisions between trucks and cars. The modeling approach provides a novel framework for jointly analyzing truck-involved crash injury severities. The findings will help policymakers take the necessary actions to reduce truck-car crashes by implementing appropriate and accurate safety countermeasures.

@en

This study proposes random-parameters multinomial logit models, with heterogeneity in means and variances, to explore the differences in the factors influencing injury severities of drivers involved in different types of two-vehicle crashes. The models are verified using crash data from the United Kingdom (UK) over three years (2016–2018). Three types of crashes are separately identified (car-car, car-truck, and truck-truck crashes). In this study, a wide variety of potential variables, including the driver, vehicle, road, and environmental characteristics, are considered, with two possible injury-severity outcomes: severe and slight injury. The results show that unobserved heterogeneity existed for young drivers in both car-car and truck-truck crash models and the 30 mph speed limit in the three separate models. Remarkably variations are observed in crashes involving different types of vehicles. The driver’s age and gender, speeding, sideswipes, presence of junctions, weekdays, unlit, and weather conditions significantly impact driver-injury severities in various types of vehicle crashes. These findings are expected to help policymakers seek to improve highway safety and implement proper safety countermeasures.@en

Accurate estimation of intercity heavy truck mobility flows is of vital importance to urban planning, transportation management and logistics operations. The inaccessibility of big data related to intercity transport systems and the heterogeneity of trucking activities pose challenges for the reliable estimation. Recently, the advance of Artificial Intelligence (AI) provides a potential solution to this problem. However, most previous studies focused on the estimation of inter-regional passenger mobility. In-depth studies of estimating intercity heavy truck mobility flows by using deep learning techniques are still scarce. To fill in the gaps, we construct a deep neural network based on the Deep Gravity framework, an advanced predictive model for human mobility. We collect a wide range of data related to heavy truck movements, freight locations, road networks and land uses to train the model, and validate its high performance by comparing to traditional gravity model. Furthermore, we use an explainable AI technique to interpret how the city features contribute to the determination of intercity heavy truck movements, and the results can provide valuable policy implications for logistics operations, businesses and urban planning.

@en

Background: Existing studies mainly focus on the relationship between real-time weather and traffic crash injury severity, while few scholars have investigated the operation risk levels caused by traffic incidents. Identifying weather-related factors that affect the incident-induced delay is helpful for estimating the delay levels when an incident occurs. Accordingly, the present study profoundly explores the relationship between weather conditions and traffic delays caused by traffic incidents. Methods: The traffic incident and weather datasets from January 1 to December 31, 2020, in New York State are used. To that end, the hazard-based duration and multinomial logit modeling frameworks are employed to determine the effect of weather conditions on the duration of traffic delay and the delay severity, respectively. More importantly, to account for multiple layers of unobserved heterogeneity, a random parameter with heterogeneity in means approach is introduced into the above two models. Results: (1) The strong breeze (wind speed over 8 m/s) and low visibility (visibility under 5 km) significantly affect the duration of delay. (2) Hot day (between 20 and 30 °C) has a 344.03 % greater probability of minor delay. A strong breeze has a higher probability of severe delay. The low visibility is found to increase the estimated odds of moderate delay and severe delay by 51.15 % and 13.39 %, respectively. In comparison, the normal visibility (between 10 and 20 km) significantly decreases the estimated odds of severe delay by 119.17 %. Conclusions: Compared with other weather factors, wind speed, temperature, and visibility have the greatest impact on the traffic delay levels after a traffic accident, and there are significant differences in the impact under different delay severity. Findings from this study will help policymakers to establish comprehensive differentiating security measures to resolve traffic delays.

@en