| In people’s cognition,precipitation in the form of rain or snow usually has an unpredictable impact on urban traffic and increases the incidence of traffic accidents.Nowadays,as an extremely important part of urban public transport system,taxi is often used to analyze or evaluate the traffic operation.Therefore,in this study,we will use taxi GPS data to analyze and study the spatio-temporal characteristics of the taxi operation for different precipitation conditions.This paper firstly preprocessed some taxi data in Lanzhou in 2021,and extracts the corresponding taxi speed and pick-up and drop-off point data.Second,Through the statistics and visualization of the taxi average velocity at different times,got the change of taxi average speed in each period in different regions of Lanzhou city for a week,and the taxi data of heavy rain days,moderate rain days,light rain days and normal days were selected to make comparative analysis on the average running speed of taxis under different weather conditions;Then,On the basis of determining the congestion level of some sections of the city,the multiple linear regression and ridge regression models between weather factors and average road speed were established from different time periods such as morning and evening peak and off-peak,and the influence degree of each factor on average road speed was analyzed by taking working days and non-working days as time variables.The model results show that compared with other indexes,whether it is a working day has a great influence on the change of taxi speed.Compared with other factors,the influence coefficient of maximum and minimum temperatures is very small,so it can be concluded that they have almost no influence on the speed of taxi.The effect of wind is smaller,but there is a slight effect.Third,respectively,the travel time on working days and non-working days and the proportion distribution of travel time under different rainfall conditions were statistically analyzed.On the basis of the previous stage,the extended converted lognormal distribution model(ESLN)based on time variation for different weather conditions was established to fit the travel time of each section in different periods.Buffer index BI and buffer index change rate BIVR were used to analyze the reliability of road travel time.The results showed that the distribution of road travel time of different rainfall levels in Lanzhou is similar and all conform to the lognormal distribution.Heavy rainfall will lead to the enhancement of the reliability of taxi road travel time,and the enhancement effect will be enhanced with the increase of rainfall;The reliability of taxi travel time due to rainfall in urban branches is slightly reduced.Finally,the pick-up and drop-off point data are analyzed statistically and visually in different time periods and different weather conditions.The data of pick-up and drop-off points were extracted,and hot spots were visualized and analyzed by using Gscan algorithm and kernel density analysis method for data of pick-up and drop-off points from different angles,so as to recommend picking up hot spots for taxi drivers and passengers and recommend hot spots for different travel behaviors.The results show that the travel hotspots are mainly near west Railway Station,Cross of West Railway Station,Lanzhou Second Hospital and Lanzhou Station,and residents’ travel areas are mainly hospitals,transportation hubs and transfer stations.In the morning and evening peak hours,the hot spots also include Chenguanying,Xiguan,provincial government and Jiaojiawan during heavy rain and moderate rain.Residents in these areas mainly include going to work and school.In light rain and normal days,the hot spots of the pick-up and drop-off points are similar and evenly dispersed,but the density of hot spots of the pick-up and drop-off in Chenguanying and Jiaojiawan is higher on light rain days,so drivers can go to this area to look for passengers during peak hours in rainy days. |
