Analysis of the impact of climate-driven extreme weather events (EWEs) on the UK train delays: A data-driven BN approach
开发了一个数据驱动的贝叶斯网络模型，分析极端天气事件如何影响英国火车延误，识别关键风险因素，并预测不同天气情景下的延误概率和严重程度，为交通规划者制定适应措施提供依据。
• Develop a new Bayesian Network model to analyse how extreme weather events affect train delays in the UK. • Identify the key factors that cause train delays, providing detailed insights into the most influential risks across the UK. • Examine the effects of extreme weather, focusing on severe floods as the biggest impact on train reliability. • Provide a predictive framework to measure the likelihood and impact of different weather scenarios on train delays. Climate change exacerbates the occurrence of frequent Extreme Weather Events (EWEs), directly disrupting railway operations in numerous countries, notably the United Kingdom. Projections for the UK climate indicate an increase in rainfall intensity, warmer and wetter winters, hotter and drier summers, and more frequent and intense EWEs. Such climatic shifts cause increased weather-related railway delays, which in turn result in significant economic loss. This study develops a new risk model using a data-driven Bayesian Network (BN) to analyse the impact of climate-induced EWEs on UK train delays. The model quantifies the influence of various factors on delays, providing deeper insights into their individual and combined effects. The new model and the findings contribute to the disclosure of 1) the interconnections among the different variables influencing train delays, including the origin and destination of the train and traction type, and 2) the prediction of the quantitative extent to which the variables can jointly lead to train delays of different severity levels, incident reason, the month of occurrence, the responsible operator, and the train schedule type. Critical findings highlight the substantial negative impact of severe flooding on the operational reliability of the UK railway system. An important insight was the significant clustering of delays ranging from 80 to 90 min, particularly on Fridays, suggesting the need for targeted operational interventions in specific regions. Additionally, the analysis identified December as the most hazardous month for train delays due to EWEs, with January and July also showing elevated risk levels. This paper offers valuable insights for transport planners, enabling them to prioritise climate-related scenarios causing the most severe train delays and to formulate the associated adaptation measures and strategies rationally.