Urban Mobility Institute, Tongji University, 200092 Shanghai, China; Intelligent Transportation Research Center, Zhejiang Lab, 311121 Hangzhou, China; The Key Laboratory of Road and Traffic Engineering Ministry of Education at Tongji University, Tongji University, 201804 Shanghai, China.
Intelligent Transportation Research Center, Zhejiang Lab, 311121 Hangzhou, China.
Accid Anal Prev. 2024 Jun;200:107491. doi: 10.1016/j.aap.2024.107491. Epub 2024 Mar 14.
Freight truck-related crashes in urban contexts have caused significant economic losses and casualties, making it increasingly essential to understand the spatial patterns of such crashes. Limitations regarding data availability have greatly undermined the generalizability and applicability of certain prior research findings. This study explores the potential of emerging geospatial data to delve deeply into the determinants of these incidents with a more generalizable research design. By synergizing high-resolution satellite imagery with refined GIS map data and geospatial tabular data, a rich tapestry of the road environment and freight truck operations emerges. To navigate the challenges of zero-inflated issues of the crash datasets, the Tweedie Gradient Boosting model is adopted. Results reveal a pronounced spatial heterogeneity between highway and urban non-highway road networks in crash determinants. Factors such as freight truck activity, intricate road network patterns, and vehicular densities rise to prominence, albeit with varying degrees of influence across highways and urban non-highway terrains. Results emphasize the need for context-specific interventions for policymakers, encompassing optimized urban planning, infrastructural overhauls, and refined traffic management protocols. This endeavor may not only elevate the academic discourse around freight truck-related crashes but also champion a data-driven approach towards safer road ecosystems for all.
在城市环境中，与货运卡车相关的碰撞事故造成了重大的经济损失和人员伤亡，因此越来越有必要了解此类事故的空间模式。由于数据可用性的限制，某些先前研究结果的普遍性和适用性受到了极大的影响。本研究探讨了新兴地理空间数据的潜力，以更具普遍性的研究设计深入研究这些事件的决定因素。通过将高分辨率卫星图像与精细的 GIS 地图数据和地理空间表格数据相结合，呈现出丰富的道路环境和货运卡车运营情况。为了解决碰撞数据集零膨胀问题的挑战，采用了 Tweedie 梯度提升模型。结果表明，在高速公路和城市非高速公路网络中，碰撞决定因素存在明显的空间异质性。货运卡车活动、复杂的道路网络模式和车辆密度等因素变得尤为突出，尽管在高速公路和城市非高速公路地形上的影响程度有所不同。研究结果强调了决策者需要采取具体情况具体分析的干预措施，包括优化城市规划、基础设施更新和改进交通管理协议。这一努力不仅可以提升围绕货运卡车相关碰撞事故的学术讨论，还可以倡导一种以数据为导向的方法，为所有人创建更安全的道路生态系统。
