A Study On The Spatiotemporal Pattern Of The Impact Of Shanghai Subway Commuting Trips Under Rainfall Scenarios

With global warming,the frequency and intensity of extreme rainfall in cities are on the increase.Rainfall events overlap with urban morning and evening commuting peak hours,with increasing impacts on public transportation and residential travel.How to quantitatively assess the impact of rainfall on urban commuter transportation and to provide effective spatial response and protection to improve the resilience of the transportation system is a key issue that needs to be addressed.The subway is the main mode of commuting for urban residents in Shanghai.Mining the spatial and temporal patterns of metro commuting flows and changes in travel patterns under rainfall events can provide a basis for traffic management departments to reasonably formulate plans and effectively respond to emergencies.Most existing studies use long series and aggregated daily rainfall data and daily passenger flow for correlation analysis.Commuter flow fluctuations(commuting behavior)are affected by the combination of rainfall onset time,duration,extent and intensity,the geographic location of the exposed station and the surrounding built-up environment,as well as the commuting distance.Aggregate-scale studies cannot deeply reveal and characterize the spatial and temporal heterogeneity of the impact of rainfall events on metro passenger flow,and are difficult to apply to transportation travel protection decisions in key time periods and spatial regions.In this paper,based on hourly-scale rainfall data and metro OD passenger flow data(i.e.,passenger flow at traffic origination and termination points),we investigate the spatial and temporal changes and travel characteristics of commuter passenger flow under rainfall events in two dimensions: station and OD,taking Shanghai as an example.Among them,the morning peak hours are from 7:00 to 10:00,and the evening peak hours are from 17:00 to 20:00.First,the Prophet time series model is used to predict the metro passenger flow values under normality;second,the measured passenger flow on the rainfall day is compared with the normality fitted passenger flow,and the amount of passenger flow fluctuation caused by the rainfall factor is calculated;then,based on the passenger flow data and the POI data around the station,the station function is Next,kriging spatial interpolation is performed on station hourly rainfall data to achieve spatio-temporal matching of hourly rainfall and passenger flow fluctuations;finally,the spatio-temporal patterns of metro commuter flow under rainfall events are quantitatively evaluated in two dimensions,namely,metro station and OD,and the characteristics of station,distance and trip of passenger flow fluctuations are explored.The main research results are as follows.(1)Normal subway commuting is dominated by medium and long distance.Among them,short-distance(?15min)commuting accounts for 8.3%,medium-distance(15?45min)accounts for 65.5%,long-distance(45?60min)accounts for 18.3%,and very long-distance(?60min)accounts for 7.9%.Commuter traffic shows a significant tidal phenomenon,with morning peak inbound traffic mainly distributed in large residential areas on and beyond the outer ring road,and evening peak inbound traffic mainly distributed in large commercial and industrial areas in the city center.(2)Rainfall can cause the volatility of inbound passenger flow at different types of stations.The overall inbound passenger flow decreases as the hourly rainfall increases,with the most pronounced decrease at residential stations and a smaller decrease at commercial service stations.The rainfall delays the departure time of commuters,causing time lag and pile-up of inbound passenger flow,and the pile-up effect is more significant at stations with higher commuting demand.Due to the elasticity difference of departure time,the rainfall sensitivity of passenger flow at different time points is also different,with higher sensitivity at 7:00 and 17:00,and relatively rigid at 8:00?9:00and 18:00?19:00.(3)The impact of rainfall on OD passenger flows of different distances,flow directions and routes shows variability.Rainfall causes a significant increase in shortdistance passenger flows with a travel time ?15min,with an overall increase of 7.3%;the change in medium-and long-distance passenger flows is not significant.The volatility of passenger flow is most significant for residential (?) industrial stations compared to residential (?) commercial stations.The starting stations of rainfall-sensitive routes in the morning peak are mostly located in large residential areas,while in the evening peak they are located in large industrial parks and commercial centers.The results indicate that although rainfall events have insignificant effects on total commuter flows,they can cause a surge of flows in local spatial areas and time points.The research methods and results of this paper help to quantify the extent of the impact of rainfall on metro commuter flows and provide a basis for spatially-based decisions on traffic operation protection.