1. Introduction
TOD is an urban planning concept aimed at achieving intensive, efficient, and sustainable urban mobility by aligning public transportation with land use around the stations [
1,
2]. It can be divided into station area TOD and citywide TOD based on the research scope [
3]. Station area TOD focuses on developing high-density, mixed-use, and pedestrian-friendly environments around specific transportation nodes, such as railway stations, bus rapid transit (BRT) stops, and bus stations [
4,
5]. It prioritizes factors like land use configuration, the built environment, and station accessibility to enhance ridership and foster sustainable growth [
6]. Citywide TOD seeks to optimize the synergy between public transport, land use, and resource allocation across urban areas [
7,
8]. It emphasizes adapting land use patterns, transit systems, and transit routes (stations) to meet the effective organization and strategic development of travel demand [
9,
10]. Due to the characteristics listed above, TOD has become an important direction of urban renewal efforts [
11,
12].
As an integral component of public transportation systems and a precursor to high-capacity transit systems, the role of bus systems in TOD is undeniable [
13,
14]. Passenger flow volumes destined to and originating from service areas of bus stops, referred to as passenger flow volumes from and to bus stops, serving as a link between the bus system and its service area, play a vital role in urban bus network planning and optimization, particularly in TOD contexts where land use is designed to support efficient transit systems [
15]. However, constrained by the unclear relationship between land use along bus routes and bus travel [
16,
17], urban renewal planning efforts for station area TOD, which involve altering land use structures and optimizing pathways from and to bus stops, as well as those for citywide TOD, which entail adjusting urban functional structures and upgrading bus routes and stops, face challenges in assessing current issues of passenger flow volumes and predicting the benefits of proposed updates. Consequently, this situation often leads to corresponding work either being shelved or confined to the vague growth in passenger flow volumes, which exacerbates issues like job–housing separation [
18], pendulum commuting, and operational losses [
19]. The challenge of regulating the relationship between land use along bus routes and bus travel significantly impedes the high-quality coordinated development of the two.
The aforementioned unclear relationship corresponds to research inadequacies in analyzing bus travel. According to relevant studies, existing analyses of bus travel can be categorized into three levels, micro, meso, and macro, each offering different analytical perspectives. At the micro level, short-term passenger flow prediction models are crucial components of intelligent transportation systems for analyzing dynamic variations of passenger flow. These models are typically divided into parametric and non-parametric models. Parametric models rely on theoretical assumptions and specific conditional parameters to analyze linear models of historical passenger flow data, providing greater interpretability. Examples include linear time series models [
20,
21], historical average models, Kalman filter models [
22], and multi-source data regression models [
23]. Non-parametric models, on the other hand, utilize machine learning methods to analyze nonlinear models of historical passenger flow data, offering better objectivity. Examples include nonlinear time series models [
24], support vector machine models [
25], and neural network models [
26,
27,
28,
29,
30]. At the meso level, passenger flow forecasting models are used to support the adjustment and optimization of bus networks by analyzing bus travel relationships. These models are typically divided into aggregation and decomposition models. Aggregation models iteratively calculate recent travel distributions based on statistical regularities of historical data and the relatively stable relationship of bus travel, featuring a relatively simple structure. Examples include the constant growth factor method and the average growth factor method. Decomposition models, on the other hand, are based on detailed traffic distribution surveys and behavioral analysis to conduct systematic analysis of traffic flow distribution corresponding to comprehensive models [
31], with stronger regularity. Examples include the gravity model method [
32], intervention opportunity model method, and maximum entropy model method [
33]. At the macro level, bus travel demand forecasting models are based on the analysis of land use relationships between bus stops to support integrated development of land use and bus travel. However, perhaps influenced by micro-level development orientation, compared with macro-strategic forecasting based on the classical scientific reductionist approach, which aims to uncover the general principles and underlying laws behind the complex travel phenomena to support strategic planning initiatives, existing research advocates micro-intelligent predictions based on the contemporary scientific paradigm of complexity theory, which focuses on unraveling the dynamics and mechanisms inherent in complex travel phenomena to aid precise control in design solutions. Moreover, passenger flow forecasting models, which are associated with relationships between land use along bus routes and bus travel, typically rely on the four-stage method [
34,
35], which involves dividing TAZs according to the service areas of bus stops [
36], utilizing survey results of the bus travel OD matrices between TAZs as a simulation basis, and constructing complex bus network models based on trip generation and distance impedance of bus travel [
37,
38,
39,
40]. Perhaps it is the unclear logical relationship in distance decay [
41,
42,
43] that leads to misuse of passenger flow forecasting theories and methods. Alternatively, the complexity of bus network models superimposes various differences in land use attributes, and stringent standards of bus travel OD matrix volume simulation magnify the differences in land use attributes [
44,
45]. This results in the segmentation, alienation, and mingling of land use elements in the analysis of bus travel demand. Consequently, the predominant role of land use generation relationships and bus travel distance impedance is weakened, while the subjective influence of non-land use factors such as household and individual attributes is strengthened. Land use idiosyncrasies gradually replace common patterns, resulting in a state of overfitting. In addition to increasing the difficulty of data acquisition and interpretation, as well as issues regarding model portability, this also leads to a disconnect between land use and bus travel demand. Consequently, it becomes challenging to advance bus travel demand management at the land use level.
The systematic analysis of the time-and-distance decay law [
46], which reflects the impedance of bus travel distance in passenger flow forecasting models, contained within residents’ travel time and distance distributions, lays the groundwork for understanding the general relationship between land use along bus routes and bus travel. In order to unravel the general relationship above and to foster coordinated development, the rest of this paper is organized as follows.
Based on the principle of the gravity model,
Section 2 deconstructs the point–line units of bus travel, taking bus stop service areas as bus stop TAZs and passenger flow volumes from and to bus stops as estimation objects. The gravitational logic estimation method corresponding to passenger flow volumes from and to bus stops is constructed by land use types, intensities, and spatial distributions between bus stop TAZs and the respective upstream and downstream collections of bus stop TAZs.
Subsequently,
Section 3 takes the passenger flow volume from and to 38 bus stops in the Xueyuan Square area of Dalian during weekday morning peak hours as the experimental object, and the basic estimation models of two gravity sets corresponding to passenger flow volumes from and to bus stops are constructed using the bus travel generation aggregation of area-based origin unit method and the bus travel distance impedance of the probability density method.
With passenger flow volumes from and to bus stops surveyed and the walking and bus travel distance impedance obtained,
Section 4 verifies the reliability of the estimation method of passenger flow volumes from and to bus stops through regression fitting between the surveyed values of passenger flow volume and the estimated values of the basic models.
5. Discussion
The construction of the gravitational logic of bus travel at the land use level using the bus travel generation aggregation of area-based origin unit method and the bus travel distance impedance of probability density method offers a theoretical foundation and methodological support for fostering fair [
48,
49], orderly [
50], efficient, sustainable [
51], and intensified demand organization in TOD.
The bus travel relative generation parameters (corresponding to generation efficiency) of different land use types in unit area and the bus travel relative distance decay law (corresponding to distance impedance, related to bus travel distance service levels), provide an orderly framework for coordinated development between land use and bus travel. In station area TOD, the basic unit order logic includes, the land use type layout should be organized based on generation efficiency to meet the bus travel intensity, and the same type of land use should follow a fair layout to reduce the impact of accessibility differences on travel efficiency. In citywide TOD, the order between basic units includes, the urban spatial functions should be relatively balanced to reduce pendulum-like commuting and the separation of residential and workplace areas, and the urban spatial structures should align with the operational characteristics (volume, speed, distance traveled, etc.) of different transportation systems to adapt to the structured upgrading of demand organization.
The construction of the estimation method of passenger flow volumes from and to bus stops provides actionable ideas for the straightforward analysis and optimization of TOD-related issues, as shown in
Figure 10.
As depicted in
Figure 10, within the constraints of the study area and time period and with the calibration of parameters such as production, attraction, and distance impedance in the basic estimation models, the application scenarios of the estimation method may include three main categories, as follows.
(a) Estimating passenger flow volumes within bus stop TAZs: When the relationship between land use around bus stops and along bus routes is known, but passenger flow volumes from and to bus stops are unknown, this estimation method can be used to estimate these volumes and diagnose potential deficiencies in passenger flow in terms of land use types, intensities, and spatial distributions.
(b) Updating land use within bus stop TAZs for station area TOD: For bus stops where the relationship between land use around bus stops and along bus routes is known, this estimation method can not only assess walking inhibition from and to the bus stops (due to inconvenient paths, unreasonable layout relationship, etc.) under the current land use layout but also evaluate improvements in passenger flow volumes after adjusting land use layouts and optimizing access paths. This facilitates organic land use renewal within bus stop TAZs from a bus travel perspective.
(c) Updating land use within upstream and downstream bus stops TAZs for citywide TOD: For bus routes where the land use generation and distance impedance are known, the estimation method can not only assess the benefits in passenger flow volumes from scenarios such as adding or removing bus stops, optimizing spacing, and adjusting routes but also consider the redistribution of passenger flow volumes for integrated development in the service area and structural upgrades to meet increased demand.
6. Conclusions
To unravel the general relationship between bus travel and land use around bus stops and along bus routes and to foster their coordinated development, this paper explores the estimation method of passenger flow volumes from and to bus stops (serving as a link between the bus system and its service area) based on land use elements. Through the analysis of the gravity model principle and method of estimating passenger flow volumes, the construction of basic estimation models corresponding to the gravity set based on land use elements and empirical testing of the estimation method, the main conclusions obtained are as follows.
Regression results demonstrate the feasibility of estimating passenger flow volumes based on land use elements. In other words, passenger flow volumes from and to bus stops can be derived from the land use types, intensities, and spatial distributions on both sides connected by these volumes by treating the land use elements as the gravity set, which is related to the supply and demand of passenger flow volumes. Decoding this general relationship provides a fair and orderly methodological framework for the coordinated development and strategic decision-making on bus travel and land use around bus stops and along bus routes, particularly in the context of TOD planning updates.
In contrast to existing complex and non-transferable analytic predictions concerning bus network relationships, this paper introduces the concept of passenger flow volumes from and to bus stops, deconstructing the point–line units corresponding to these volumes from bus network relationships to simplify the estimation of passenger flow volumes at the land use level. By using the area-based origin unit method, which hypothesizes the bus travel generation is closely correlated with land use types and their corresponding area scales, and the bus travel distance impedance of probability density method, which deciphers the decay probability of different bus travel distances under certain conditions, a portable estimation method based on gravity relationships is constructed and empirically verified.
While the function model has been tested and confirmed with relatively high confidence, ensuring the reliability of the estimation approach and method, there are limitations in its application and extension. The estimation model primarily simulates passenger flow volumes from and to bus stops during the morning peak period in the experimental area, so future research should expand the study area and time period for broader verification of the estimation method. Additionally, the simplified basic estimation models used for experimentation may suffer from issues such as poor representativeness of parameter values, inadequate transfer considerations, and insufficient prediction accuracy. To improve the model, it is necessary to minimize the interference of random factors in estimation of passenger flow volumes from and to bus stops and refine the aggregation of generation and distance impedance while improving the aggregation of transfer passenger flow volumes of bus travel.