土地利用视角下的轨道站点客流预测——以深圳市为例

从土地利用视角分析轨道站点客流的影响因素,基于直接估计模型,选择不同类型用地楼板面积、公交线路数量、站点中心性、站点可达性4个变量建立非线性回归函数,分析不同类型用地客流生成率与站点可达性之间的衰减规律。结果表明：① 相比于线性衰减规律,客流生成率随站点可达性的降低更符合指数衰减规律,用地距离地铁站0~0.5 km时客流生成率下降最快。② 单位面积各类用地客流生成率由大到小依次为：交通用地>办公用地>商业用地≈城中村>工业用地>居住用地>其他用地。③ 客流生成率受站点可达性影响强度由大到小依次为：商业用地>办公用地>交通用地>其他用地>居住用地>城中村>工业用地。

A New Method for Forecasting Station-level Transit Ridership from Land-use Perspective: The Case of Shenzhen City

Analyzing the numerical relationship between transit ridership and land use is significant to station location, land use planning and rail transit property development. Although there have been a growing number of studies on long-term ridership forecasting, limited effort has been made to investigate the impacts of different land use types on ridership and ridership decay rules. Two problems are considered in this paper. First, how do the land development intensity of different types influence station ridership? Second, what are the relationships between the walking distance to the station and trip rates for different land types? This paper attempts to answer the questions based on the direct ridership model. A two-step nonlinear regression model is established and the parameters are estimated by the Gauss-Newton method. Four variables including floor area of different land types, bus lines, centrality and walking distance to the station are included in the model. The first-step model separates the ridership produced by the land use from the total ridership. In the second-step model, two types of distance-decay function, exponential-decay function and linear-decay function, are considered. The parameters of exponential model and linear model are calculated according to . Smart card data of 94 subways in Shenzhen are used to valid the model. The results are: 1) Compared to the linear decay tendency, the decay rule is more in line with the exponential form; 2) Trip rates of different land types ordering from large to small are: transport land > commercial land > official land ≈ urban village > industrial land > residential land > other land; 3) The impacts of walking distance on the trip rate ordering from large to small are: commercial land > official land > transport land > other land > residential land > urban village > industrial land. Due to considering the factors of land use and walking distance to the station, the proposed model provides a quite accurate prediction of the station ridership. Ridership forecasting is crucial for the analysis of project viability and urban rail transit planning. With the help of estimated ridership, subway planners can proactively judge whether a site is suitable for setting transit station. The decay rules of different land types can provide reference for property development around the stations and travel analysis.