降雨事件对上海地铁通勤客流时空影响的精细尺度研究

全球气候变化背景下,降雨事件日益增多,严重影响城市交通和居民的日常出行。本文以上海市为例,基于分时降雨数据和地铁OD客流数据,运用Prophet时序模型拟合降雨事件下的客流常态值,从站点和OD二个维度定量评估降雨造成的地铁通勤客流变化时空格局。研究结果表明：① 通勤客流总体随小时雨量增大而下降,不同类型站点客流的降雨波动性呈现差异;降雨会造成进站客流的时间滞后性和堆积性,通勤出行需求越大的站点类型堆积效应越显著;由于出发时间弹性差异,不同时点客流的降雨敏感性也不同,7:00和17:00敏感性较高,8:00—9:00和18:00—19:00则相对刚性;② 降雨会造成行程时间≤15 min的短距离客流显著上升,总体增加7.3%,中长距离客流变化不明显,总体减少1.3%;在不同功能区之间,早高峰居住型→产业型的客流波动和时间堆积性最为显著,晚高峰商服型→居住型的返程客流波动性较低;早高峰降雨敏感性线路的起始站点多分布在大型居住区,晚高峰则位于大型产业园区和商业中心;晚高峰返程客流的波动性低于早高峰。尽管降雨事件对通勤客流总量影响不明显,但会造成局部空间区域和时点的客流激增。本文的研究方法与结果有助于量化降雨对地铁通勤客流的影响程度,并为空间化的交通运行保障提供决策依据。

A Fine-scale Study on Spatio-temporal Patterns of Metro Commuter Flows under Rainfall Events in Shanghai

In the context of global climate change, extreme precipitation events are becoming more frequent and have an increasing impact on urban commutes. In this study, based on hourly rainfall data and metro OD passenger flow data, we use a prophet time-series model to forecast the regular values of commuting flow under rainfall events, and quantitatively assess the spatial-temporal changes of commuting flow caused by rainfall at station and OD levels. Our results show that (1) the commuting flow generally tends to decrease with increasing hourly rainfall. The fluctuation of commuting flow varies from one type of station to another. Rainfall can delay commuting departure time and lead to surge in metro flow in certain times. The higher the commuting demand for a station, the more its flow fluctuates. Flow fluctuation due to rainfall varies in different time periods. 7:00 and 17:00 show high fluctuation with more flexibility in commuting departure time, while 8:00—9:00 and 18:00—19:00 show high rigidity; (2) Rainfall can induce a significant increase in short commuting flow of less than 15 minutes, averaging to around 7.3%. In contrast, the impact on medium and long commuting flow is modest, with an overall decrease of 1.3%. Of the OD flows across various functional zones, fluctuation from residential to industrial stations is most notable during the morning commute, while less so from commercial to residential stations during the evening commute. Most of the departure stations of rainfall-sensitive metro lines during the morning commute are located around large residential areas, and around large industrial parks and commercial centers during the evening commute. Flow fluctuation in the evening commute is lower than that in the morning commute. Although total commuting flow is not significantly affected by rainfall, its surge in certain local regions and times should be highlighted. Our methodology and results will help to quantify the impact of rainfall on metro commutes and provide a basis for spatialized transport coping strategies.