Land Use Characteristics of Commuter Rail Station Areas and Their Impact on Station Ridership: A Case Study of Japan Railways in the Tokyo Metropolitan Area

Abstract

Exploring the relationship between land use characteristics and ridership in railway station areas provides crucial decision-making support for station area planning. Previous research has mostly focused on subways, with a lack of studies on the land use characteristics and ridership of commuter rail stations, particularly in relation to the differences and impacts across various passenger catchment areas (PCAs). This study employed a multinomial logit model to evaluate the land use characteristics within 1000 m of Japan Railways (JR) stations in four different PCAs of the Tokyo metropolitan area (TMA). Additionally, regression models and a multiscale geographically weighted regression (MGWR) model were used to analyze how land use characteristics in these PCAs affected station ridership. The key findings were as follows: (1) the land use characteristics around commuter rail stations exhibit distinct zonal patterns; within 250 m, public transport stops and public service facilities are the most densely concentrated; the highest residential population density is found between 250 and 750 m; and commercial facilities are mostly clustered in the 500 to 750 m range; (2) the impact of land use factors on ridership varies in intensity across different spatial zones; the density of public transport stops and street network density is most significant within 250 m, whereas commercial facility density is greatest within the 500–750 m PCA; (3) The land use characteristics within 500 m of stations have greater explanatory power for passenger flow, and the goodness of fit of the MGWR model surpasses that of the linear regression model.

1. Introduction

As urbanization continues to accelerate, large cities are experiencing significant growth in spatial scale and population density. This trend has led to increased long-distance travel within metropolitan areas, as suburban expansion of residential areas and businesses has intensified commuting patterns between urban centers and suburban regions. Concurrently, serious issues such as traffic congestion, safety risks, and environmental pollution have prompted urban planners and policymakers to emphasize public transportation and compact urban development. In response, major cities such as New York, Tokyo, and Seoul have developed metropolitan rapid rail systems. Similarly, in rapidly urbanizing China, metropolitan rail network development is also progressing rapidly. Transit-oriented development (TOD) around station areas is widely recognized as a critical factor in influencing passenger volumes and the operational efficiency of rail systems [1,2]. Despite these efforts, significant disparities persist in spatial development patterns and travel behaviors between urban centers and suburban areas in many large cities. Station area development must be context-sensitive, as surrounding environments vary based on regional contexts, rail types, and geographic location within the transit network [3,4,5,6]. Moreover, different types of rail stations exhibit distinct factors affecting passenger volumes [7].

Existing research on the built environment and ridership around subway stations is extensive [2,8,9,10,11]. Additionally, numerous studies have examined the relationship between high-speed rail station accessibility and land use patterns at national or regional scales [12,13,14]. However, studies on the built environment and ridership around commuter rail stations within metropolitan areas remain relatively limited [3,15], and findings regarding the factors influencing ridership are often inconsistent. For example, Wells and Hutchinson [16] concluded that in the Greater Toronto Area, suburban employment growth limited the effectiveness of land use policies. Policies that promoted high-density residential development near suburban commuter rail stations were unlikely to contribute significantly to ridership growth. In contrast, Sakano and Benjamin [17] found that in metropolitan areas in North Carolina, not only residential and workplace locations but also commercial facilities around stations significantly influenced commuters’ travel choices. Similarly, Durning and Townsend [18] demonstrated that land use, facilities, and amenities at Canadian rapid rail stations affected ridership levels. Other studies have argued that different rail types require distinct TOD strategies to effectively boost ridership [11]. Given that metropolitan commuter rail spans both urban centers and peripheral zones, the spatial distribution of station ridership and the surrounding built environment varies greatly, resulting in intricate and context-dependent interactions [7,16,19]. This underscores the need for further case studies to better understand the relationship between the land use characteristics and ridership at commuter rail stations.

Tokyo, as a prototypical high-density metropolitan region, offers valuable insights from its extensive experience with TOD. The JR rail network, which connects the central and peripheral areas of the Tokyo metropolitan area (TMA), presents a wide range of station development types and serves as a typical case study for commuter rail in metropolitan regions. Using the node–place–passenger flow model, Cao [20] found that JR stations, compared with subway and private rail stations, exhibited better balance in node functions and passenger flow. Therefore, studying the relationship between the built environment and ridership at JR stations in the TMA offers valuable insights for the development of rail transit systems in other high-density metropolitan regions in Asia.

This study has two primary research objectives: (1) to assess the land use characteristics of different PCAs around JR stations in the TMA based on TOD principles; and (2) to investigate the impact of the land use characteristics in each PCA on station ridership. To achieve these objectives, a multinomial logit model was employed to analyze the land use characteristics across different PCAs of JR stations, and stepwise regression and mixed geographically weighted regression (MGWR) models were used to investigate the relationship between land use characteristics and station ridership in various PCAs. This study not only offers an empirical analysis of land use and passenger flow around railway stations but also introduces a novel methodological approach to TOD research by precisely subdividing station impact zones into specific sub-areas.

The paper is composed of five sections. Following the introduction in Section 1, Section 2 reviews the spatial scope of TOD land use, socioeconomic factors, and their influence on station ridership. Section 3 describes the study area, data sources, and methods. Section 4 presents the analysis results and explores these findings and their implications. Section 5 summarizes the study and discusses directions for future research.

2. Literature Review

Long-term reliance on motorization and the resulting energy and environmental issues have spurred extensive research on the relationship between land use and transit ridership. TOD has been shown to reduce automobile dependency [21,22], significantly boost ridership at rail transit stations [23,24], and promote the sustainable development of transportation and urban systems [25,26]. However, conclusions regarding how TOD influences travel behavior and which factors are most critical vary across different studies [11,27,28]. This is particularly true for commuter rail, where significant differences in land use and factors affecting ridership exist across different station areas [7]. The following literature review focuses on four aspects: the spatial scope of TOD at rail stations; land use and station attribute factors that influence ridership; the methodologies used to analyze the built environment–ridership relationship; and the context of rail transit ridership and land use in Japan.

2.1. Space Scope of TOD

The spatial extent of railway station influence zones provides a foundation for land use planning strategies, which vary across countries and regions depending on rail types and geo-social characteristics. In North America, tram and light rail stations primarily serve concentrated commercial and residential functions, with TOD areas often defined by a 400 m radius [21], while train stations typically adopt an 800 m radius [29]; in Europe, studies typically employ 700–800 m buffer zones [30,31], and in some cases, up to 960 m is defined as the walkable station area, with enterprise locations extending to a 1km limit, as seen in London and Copenhagen [32,33]. In high-density Asian metropolitan areas, such as Seoul, South Korea, a 900 m radius is used for metro stations [34], while 1500 m is applied for metropolitan rail stations [11]. Similarly, in Japan’s Tokyo Metropolitan Area, a 500 m radius is commonly applied for central city railway stations [35], whereas those for peripheral stations extend to 800–1000 m [36,37], In China, subway stations are generally studied within 800–1000 m [7], while railway stations extend to 1000–1500 m [38,39]. This findings indicate that TOD extents are smaller in central cities with dense rail networks and larger in peripheral areas. Among rail types, railway stations have the largest TOD extent, followed by subways, while tram stations have the smallest.

2.2. Land Use Characteristics and Station Attributes Influencing Transit Ridership

Although research on the relationship between land use and ridership at commuter rail stations remains limited, substantial insights have been gained from studies on subway station areas, particularly regarding the influence of the built environment on travel behavior. Findings on the impact of high-speed rail on land use structure and intensity can also provide valuable references for commuter rail studies. Research often evaluates station areas through TOD factors in the built environment, as well as node and place characteristics, to assess their correlation with ridership levels.

TOD within the built environment is commonly evaluated using the “three Ds”: density, diversity, and design [40], critical determinants of ridership. Among density indicators, employment density has been consistently identified as a significant factor [41], while the impact of residential density has varied across studies [8,42,43,44]. For instance, Liu (2016) emphasized the role of employment density in Maryland’s metro system [7], which was similarly observed by Sung for Seoul [19] and Pan for Shanghai [10].

The impact of land use diversity on station ridership has yielded various conclusions. For instance, some studies indicate a positive promotion effect [15,29,34], while others suggest it has no significant effect [11,43]. Lee argued that the relationship between ridership and the built environment in the Seoul metro system varies based on network accessibility and urban structure [45]. In central business districts (CBD) and fringe areas, ridership is primarily influenced by density, whereas in the inner suburbs, it is mainly affected by diversity. These conflicting findings highlight the need for further research to investigate the impact of land use combinations on station traffic ridership.

Station location and attributes significantly impact ridership, and the measurement of station location and importance is becoming increasingly refined. For instance, the distance to the central business district (CBD) is often used as a measure of regional accessibility and is generally regarded in most studies as having a significant negative impact on station ridership [46]. However, Liu found that this variable did not significantly affect transit ridership at commuter rail stations in the Baltimore–Washington metropolitan area [7]. Additionally, the centrality of a station within the rail network and indicators such as train service are receiving more attention and have been shown to be closely related to ridership levels [9,20,47].

2.3. Methods for Exploring the Relationship Between the Land Use Characteristics and Transit Ridership

In analyzing the factors influencing transit ridership, the ordinary least squares (OLS) regression method is among the most widely used techniques. Some researchers have extensively applied OLS regression to explore the relationship between various attributes of the built environment and ridership levels [10,31,48,49]. However, this method had some limitations, particularly in addressing the spatial variability of transit ridership and the spatial heterogeneity in the relationship between ridership and explanatory variables [32]. To overcome these issues, alternative methods such as The geographical weighted regression (GWR) model, structural equation model (SEM) [50], Poisson model [25], and gradient boost decision trees (GBDTs) [8,38] have been employed. Studies comparing these methods indicate that GWR often provides more robust and geographically detailed results than OLS [31], while MGWR enhances accuracy by capturing both global and local spatial effects of built environment indicators on passenger flow [34].

2.4. Rail Transit Ridership and Land Use in Japan

Tokyo is recognized as a global leader in TOD practices due to its extensive, highly accessible rail transit network and integration of station-based urban development [51]. Numerous studies have explored various aspects of Tokyo’s rail systems, including their relationship with urban form [52,53,54], land use evolution and influencing factors in station areas [37,55,56], commercial development patterns [56], and three-dimensional station area design [57,58]. Additionally, strategies such as TOD planning and urban renewal in the context of Japan’s aging population have received attention [59,60]. However, relatively few studies have investigated the relationship between TOD and passenger flow. Cao (2020) found that JR stations in the Tokyo metropolitan area exhibited better coordination between node, place, and ridership compared with private railways and subways [35]. Zhuang and Zhao (2014) highlighted significant variations in land use patterns and their evolution across different spatial circles around Japanese train stations [36,61]. However, relatively few studies have explored how land use characteristics influence ridership across different spatial circles, despite its critical role in shaping effective TOD planning and design.

In summary, the literature review reveals that transit ridership at rail stations has been influenced by various factors, including land use, the built environment, station location, and network centrality. However, the nature of these influences varies depending on the type of transit system (e.g., subway, light rail, commuter rail), the study region (e.g., low-density areas in the Americas versus high-density regions in Asia), and the specific attributes of the station area. Additionally, differences in analytical methods and model specifications have contributed to significant variations in the research findings. Overall, existing research has placed comparatively less emphasis on commuter rail systems and has often relied on the radius of influence of a single station to analyze the impact of land use characteristics on ridership. TOD practices in JR station areas within the Tokyo metropolitan area offer valuable insights for rail station development in other high-density metropolitan regions, particularly those in Asia.

Previous studies have established a 1 km radius as the accessible area for commuter railway stations, while the average distance between JR stations in the Tokyo metropolitan area is approximately 2 km. Accordingly, this study adopted a 1 km radius as the station catchment area and further subdivided it into four principal catchment areas (PCAs): the core PCA (0–250 m), the primary PCA (250–500 m), the secondary PCA (500–750 m), and the outer PCA (750–1000 m). This refined approach enabled a more detailed examination of the built environment’s impact on ridership across varying spatial scales. This approach allowed a more detailed examination of how the built environment affects ridership across different spatial scales. Furthermore, in terms of methodological improvements, this study applied a multinomial logit model to evaluate land use characteristics in four principal catchment areas (PCAs) based on the principles of TOD (density, diversity, and design). Multiple linear regression and MGWR models were both employed to analyze the relationship between these characteristics and subway ridership.

3. Materials and Methods

3.1. Study Area

The Tokyo metropolitan area (TMA) comprises Tokyo, Saitama Prefecture, Chiba Prefecture, and Kanagawa Prefecture, covering a total area of approximately 13,500 square kilometers. At the end of 2023, the population of this region was around 36.87 million, accounting for 27.8% of Japan’s total population, with a GDP of USD 1.6 billion [36,61]. This metropolitan area is characterized by a highly developed rail transit system, including subways, JR, private railways, and other forms of rail, spanning a total length of 2705 km [62]. Urban spatial development in the TMA is closely tied to the rail network. This study focused on the areas within a 1000 m radius of 481 JR stations distributed across 27 lines within the TMA, as is shown in Figure 1. Considering that JR connects the central urban area with the periphery, and that there is a wide variety of station types and land use along the route, this study divided the station area into four PCAs to explore the differences in the characteristics of the land use and their impact on ridership, as illustrated in Figure 2.

3.2. Data and Variables

Most studies examining the relationship between the built environment and passenger flow have utilized average daily ridership data, emphasizing that total daily passenger volumes better reflect the economic value of a rail network [10,34]. In contrast, other studies have employed weekday versus off-day and peak versus off-peak traffic data to capture the variability of traffic-influencing factors across different types of stations, particularly by identifying commuting characteristics at the line or station level [15,38]. Given the focus of this study on the overall operational efficiency of stations and the broader development of station areas, average daily ridership was chosen as the dependent variable.

Based on the variables identified in the literature review and the three “D” principles of transit-oriented development (TOD) (density, diversity, and design), this study selected residential population density and the density of points of interest (POI) as density indicators. For diversity indicators, commercial facility density, public service facility density, and facility mix were chosen. Design indicators include road network density, the number of road intersections, and the number of bus stops. Additionally, drawing on the research by Xia and Zhang [63] regarding the impact of station attribute characteristics on TOD effectiveness and ridership, this study selected distance to the CBD, degree centrality, betweenness centrality, and closeness centrality of the station within the complex network as station attribute indicators.

3.2.1. Density

Density indicators include not only residential population density but also the density of commercial facilities (such as food services, leisure, tourism, shops, and offices) and the density of public service facilities (including welfare services, cultural facilities, schools, and hospitals). The residential population density data were sourced from the 2015 National Census of Japan (https://nlftp.mlit.go.jp/ksj/gml/datalist/KsjTmplt-mesh500h30.html (accessed on 20 February 2024)). The annual total ridership data for the stations were collected and compiled from East Japan Railway Company (JR East) and the Japan National Land Numerical Information Download Service website (https://nlftp.mlit.go.jp/ksj/, accessed on 8 July 2024). The vector data for facilities, parking lots, bus stops, and street networks were obtained from Open Street Map (https://www.openstreetmap.org/, accessed on 8 July 2024) and the Japan National Land Numerical Information Download Service website (https://nlftp.mlit.go.jp/ksj/, accessed on 9 July 2024).

3.2.2. Design

The design index included the street network density, crossing density, and density of bus stops in the station area, from the perspective of walkability and convenient bus transfer. The data relating to the street network, crossings, and bus stops were derived from Open Street Map.

3.2.3. Diversity

Diversity indicators include the functional diversity of the station area and the proportion of non-built land. The data were sourced from the Japan National Land Numerical Information Download Service website and Open Street Map. The functional diversity was first measured using point-of-interest (POI) mixed entropy [20], considering commercial, educational, medical, cultural facilities, community public services, and parks within the station area. The formula is as follows:

$${\displaystyle {\sum }_{j=1}^L}Qj=1,$$

(1)

$$H_i=-{\displaystyle \frac{{\displaystyle {\sum }_{j=1}^L}{Q}_j \times ln{Q}_j}{{ln}^L}}(if{Q}_j=0,{ln}^{Q_j}=0)$$

(2)

where $H_i$ represents the functional mix indicator of Station $i$, $L$ is the number of facility categories, and $Qj$ is the proportion of facility type $j$.

3.2.4. Station Location Attributes and Socio-Economic Indicators

In this study, the station location attributes included the distance from the station to the central business district (CBD), specifically Tokyo Station, and the centrality of the station within the complex urban rail transit network. Data on the TMA’s rail transit lines and stations is sourced from the transportation bureau’s website. A topological map of the rail transit network was created based on the actual rail transit map within TMA. In this network, stations were represented as nodes, and the rail lines connecting different stations were represented as edges, forming an unweighted complex network. Gephi software (version Gephi-0.10.1) was used to calculate the degree centrality (DC), betweenness centrality (BC), closeness centrality (CC), and eigenvector centrality (EC) for each station.

Network centrality is a graph-theoretic modeling approach used to capture the impedance of networks and the importance of nodes within transportation networks. While it had previously been applied in social network analysis, the application of network centrality in spatial networks, particularly in transportation network analysis, has been increasing. For instance, J. Feng [64] and Xiao [65] analyzed ridership and node vulnerability at rail stations based on complex network analysis. Jang [47] examined the impact of network centrality characteristics of rail transit stations on ridership and average travel distance, finding that accessibility, betweenness, and closeness were significant factors influencing ridership. In this study, we converted the Tokyo rail transit network into an unweighted complex network G = (V, E), where the vertex set V represents the number of stations and the edge set and E represents the connections between stations. If there is a rail line connecting two stations, the corresponding entry in the topology connection matrix is assigned a value of 1; otherwise, it is assigned a value of 0. Using Gephi software, we calculated the DC, BC, CC, and EC in the context of this complex network.

DC represents the extent of a station’s connections to other stations, quantified by the number of links associated with that station:

$${\mathit{DC}}_i={\displaystyle {\sum }_{j=1}^{\mathrm{n}}{\mathrm{a}}_{ij}}(i\ne j), {\mathrm{a}}_{ij}=\left\{\begin{array}{l}1,\mathrm{Site}i\mathrm{is}\mathrm{connected}\mathrm{to}\mathrm{site}j\hfill \\ 0,\mathrm{Site}i\mathrm{is}\mathrm{not}\mathrm{connected}\mathrm{to}\mathrm{site}j\hfill \end{array}\right.,$$

(3)

where n is the total number of interconnected edges in the rail network, j refers to the stations other than i, and a_ij represents the connection between station i and station j.

BC represents the importance of the node as a bridge, expressed as the proportion of the shortest path through node i to the shortest path of the OD matrix of all stations:

$${\mathit{BC}}_i={\displaystyle \frac{{\displaystyle {\sum }_{j\ne k\ne i}{\mathrm{n}}_{jk}^i}}{{\displaystyle {\sum }_{j\ne k}{\mathrm{n}}_{jk}}}},\hspace{1em}i\ne j\ne k,\hspace{1em}j<k,$$

(4)

where i, j, k represent different stations, nⁱ_jk represents the number of shortest paths from station j to station k that pass through station i, and n_jk represents the total number of shortest paths from station j to station k.

CC reflects the geospatial accessibility of the expression node and is expressed as the reciprocal of the sum of the shortest rail network distances d_ij from station i to all other stations:

$${\mathit{CC}}_i={\left[{\displaystyle {\sum }_{j=1}^{\mathrm{n}}{\mathrm{d}}_{ij}}\right]}^{-1},$$

(5)

where d_ij is the shortest path length between stations i and j, and n is the total number of all paths between stations in the network.

EC shows the importance of the station itself and its environment, representing the number of stations connected to the station and the importance of the connected stations:

$$E{C}_i=\mathrm{c}{\displaystyle {\sum }_{j=1}^{\mathrm{n}}{\mathrm{a}}_{ij}{\mathrm{e}}_j},$$

(6)

where c is a proportional constant and the node eigenvector centrality is $\mathrm{e}={\left[e_1,e_2,\dots \dots e_{\mathrm{n}}\right]}^{\mathrm{T}}$, where ej is the eigenvector centrality indicator for station j and T is the number of iterations to the steady state.

Station location socio-economic indicators primarily include the size of the local resident population, the number of local employees in commercial, manufacturing, and private establishments, and the annual sales of commercial goods within the municipality where the station is located. The descriptive statistics of each variable are shown in

Table 1. Description of Variable Indicator System.

Dimension	Indicator	Description		Mean	Std.	Min	Max
Density	Residential population density (104 person/km2)	The number of residents per unit area in the station area.	0–1000 m	9534.75	9223.37	14.09	99,855.96
			0–250 m	1.21	1.55	0.00	15.27
			250–500 m	1.24	1.52	0.00	14.76
			500–750 m	1.24	1.41	0.00	10.77
			750–1000 m	0.62	0.72	0.00	6.62
	Commercial facility density (1/km2)	The ratio of the total number of commercial facilities, leisure, tourism, shops, offices, restaurants, and shopping service POIs within the station area to the area size.	0–1000 m	40.95	89.84	0.00	1039.81
			0–250 m	137.35	321.38	0.00	2423.47
			250–500 m	58.48	142.60	0.00	2123.94
			500–750 m	31.68	81.72	0.00	794.54
			750–1000 m	27.59	75.39	0.00	744.28
	Public service facility density (1/km2)	The ratio of the total number of educational and medical facility POIs within the station area to the area size.	0–1000 m	36.10	36.14	0.00	236.62
			0–250 m	89.16	96.81	0.00	1183.67
			250–500 m	45.72	50.52	0.00	402.38
			500–750 m	6.53	8.13	0.00	81.49
			750–1000 m	46.61	60.77	0.00	488.91
Design	Street network density (km/km2)	The ratio of the total street network length within the station area to the area size.	0–1000 m	33.08	31.22	3.94	364.82
			0–250 m	82.51	73.19	11.54	588.59
			250–500 m	32.01	39.72	2.06	383.08
			500–750 m	40.66	44.06	0.4	445.25
			750–1000 m	23.40	25.90	3.94	267.80
	Crossing density (1/km2)	The number of crossings within the station area.	0–1000 m	34.15	52.60	0.00	354.46
			0–250 m	52.92	58.69	0.00	408.16
			250–500 m	53.78	99.24	0.00	713.07
			500–750 m	9.83	17.25	0.00	144.64
			750–1000 m	42.42	73.13	0.00	504.91
	Bus stop density (1/km2)	The number of bus stops within the station area.	0–1000 m	9.73	7.98	0.00	36.31
			0–250 m	24.07	29.08	0.00	214.28
			250–500 m	11.00	11.22	0.00	81.49
			500–750 m	3.12	3.46	0.00	21.39
			750–1000 m	11.93	10.55	0.00	67.66
Diversity	Functional diversity	The richness and complexity of functions within the station area, measured by the entropy index of POI data. Six types of facilities were considered: commercial, public, cultural, schools, hospitals, and parks.	0–1000 m	1.26	0.27	0.00	1.66
			0–250 m	0.92	0.35	0.00	1.47
			250–500 m	1.02	0.37	0.00	1.55
			500–750 m	1.08	0.41	0.00	1.68
			750–1000 m	1.13	0.37	0.00	1.70
	Proportion of non-built-up land (%)	The proportion of non-built-up land, mainly referring to agricultural, forest land, and water areas	0–1000 m	11.61	21.92	0.00	87.10
			0–250 m	5.98	15.45	0.00	107.15
			250–500 m	9.00	19.46	0.00	88.52
			500–750 m	11.68	22.85	0.00	92.42
			750–1000 m	13.44	24.56	0.00	90.41
Station location attributes and socio-economic indicators	Distance to CBD (km)	Distance from the station to Tokyo Station.		36.71	22.21	0.00	96.23
	Degree centrality	The number of links connected to the station, representing the station’s hub status.		2.56	1.34	1.00	12.00
	Closeness centrality	The closeness of a station to all other stations.		0.06	0.01	0.03	0.09
	Betweenness centrality	The number of shortest paths passing through the station.		23,523.45	9223.37	0.00	366,479.79
	Eigenvector centrality	The importance of the station itself and its environment.		0.08	0.10	0.01	1.00
	Years of operation	The operational years of the station, with stations opened before 1950 recorded as 1950.		61.340	15.119	2.000	72.000
	Number of local resident population (104 person)	The total local resident population in the municipality where the station is located.		19.84	16.40	0.48	74.81
	Number of local employees (104 person)	The number of local employees in commercial, manufacturing, and private establishments in the municipality where the station is located.		15.60	24.46	0.17	128.88
	Sales of commercial goods (106 million yen)	Annual sales of commercial goods in the municipality where the station is located.		2.25	7.56	0.00 9	45

.

3.3. Methods

3.3.1. Multinomial Logit

This study employed a multinomial logit model to analyze the land use characteristics of the station areas. The PCA at four spatial levels around the site was used as the dependent variable of the logit model, that is, the core PCA within 250 m was “1”, the primary PCA between 250–500 m was classified as “2”, the secondary influence zone between 500–750 m was “3”, and the outer PCA between 750–1000 m was “4”, while the average value within a 1000 m radius of the station area was set as the control group “0.” Land use variables (including density, diversity, and design-related indicators) and station attribute variables were used as explanatory variables.

3.3.2. Multiple Linear Regression and MGWR

To explore the relationship between commuter rail station passenger flow and the built environment in station areas, we first conducted multiple linear regression analysis. One of the advantages of using linear regression is its ability to efficiently identify explanatory variables of the built environment through parameter inference and assess their overall impact on the dependent variable. However, linear regression models overlook the spatial auto correlation of station passenger flow and the varying relationships across spatial scales between passenger flow and the explanatory variables. Moreover, a few studies have provided empirical evidence showing that the geographically weighted regression (GWR) model fits better than the ordinary least squares (OLS) model. The GWR model offers analytical and predictive advantages by providing spatial variation in estimated parameters and their statistical significance [31]. Furthermore, the mixed geographically weighted regression (MGWR) model is better suited for capturing variable impacts across various spatial scales, making it more suitable for models composed of both local and global explanatory variables. Therefore, we constructed an MGWR model to compare with the results from the multiple linear regression model. The GWR model incorporated both global fixed and local varying terms for the explanatory variables. The model equation is as follows:

$$y_i={\beta }_o\left(u_i,v_i\right)+{\displaystyle {\sum }_k}{\beta }_{{bw}_k}\left(u_i,v_i\right)x_{ik}+{\epsilon }_i$$

(7)

where $y_i$ represents the dependent variable, which is the ridership of station $i$, $\left(u_i,v_i\right)$ are the x–y coordinates of station $i$, ${\beta }_o\left(u_i,v_i\right)$ is the intercept term for station $i$, $x_{ik}$ is the kth explanatory variable for station $i,$ ${\sum }_k{\beta }_{{bw}_k}$ is the regression coefficient for the kth explanatory variable at station $i$ using bandwidth bw, where a positive or negative value indicates a spatial matching (separation) phenomenon between station passenger flow and the explanatory variable, and εi represents the random error term.

The overall research framework is illustrated in Figure 3.

4. Results and Discussion

4.1. Distribution Characteristics of Station Passenger Flow

As shown in Figure 4, Figure 5 and Figure 6, passenger flow distribution among JR stations is notably uneven. Only a small number of stations experience high passenger volumes, with most stations handling significantly lower ridership. The Yamanote Line and its connections to the Chuo, Tokaido, Tohoku, and Sobu Lines form the primary passenger corridors, linking the urban core with peripheral areas in the metropolitan region.

The ten stations with the highest passenger flows are Shinjuku, Ikebukuro, Tokyo, Yokohama, Shibuya, Shinagawa, Omiya, Shimbashi, Akihabara, and Kita-Senju. These stations, located primarily along the Yamanote Line, serve as major transfer hubs for multiple rail lines. Altogether, the Yamanote Line accounts for approximately 14% of the total passenger flow, reinforcing its role as the backbone of the metropolitan transit network. Passenger volumes decrease along rail lines extending radially outward from Tokyo Station that exceed 30 km, with peripheral areas showing significantly lower ridership.

4.2. Spatial Characteristics of Land Use in Station Area

Table 2. Multinomial Logit Results for the Built Environment in Different PCAs Around Station Areas.

	Core PCA		Primary PCA		Secondary PCA		Outer PCA
	β	SD	β	SD	β	SD	β	SD
Population density	0.148	0.090	0.800 ***	0.092	1.210 ***	0.113	−0.098	0.074
Commercial facility density	−0.0049 ***	0.001	−0.001	0.001	0.017 ***	0.002	−0.011 ***	0.002
Public service facility density	0.010 ***	0.002	0.000	0.003	−0.205 ***	0.014	0.020 ***	0.003
Street network density	0.0329 ***	0.004	−0.037	0.005	0.018 ***	0.005	−0.041 ***	0.005
Crossing density	−0.019 ***	0.0029	0.008 ***	0.002	−0.051 ***	0.008	0.007 ***	0.002
Bus stop density	0.056 ***	0.0099	−0.003	0.009	−0.165 ***	0.024	0.019 **	0.009
Functional diversity	−4.989 ***	0.354	−4.393 ***	0.324	−2.498 ***	0.330	−1.830 ***	0.310
Proportion of non-built-up land	−0.016 ***	0.005	−0.009 **	0.004	−0.007	0.004	−0.005	0.004
Cons	2.476	0.646	−0.792	0.673	−3.868 ***	0.784	3.267 ***	0.556
Log likelihood	−2467.9613

*** p < 0.01, ** p < 0.05.

presents the results of the multinomial logit model for land use characteristics in different PCAs. Population density shows a positive and strong correlation with station ridership in both the primary and secondary PCAs but weaker correlations in the core and outer PCAs. This suggests that the population is more clustered in the primary and secondary PCAs relative to the average population density of 1000 m in the JR station area. Sung [66] also reached the same conclusion in his study of the Seoul metropolitan area railway, finding that residential land density was highest between 500 m and 750 m around the station. However, in other studies of subway station areas, it was found that the population density showed a decreasing trend from the 300 m core area outwards [34]. This may be due to the variation in land use between the subway and the metropolitan area railroad.

The density of commercial facilities in the core, secondary, and outer PCAs exhibits a strong correlation with station ridership, but the correlations are negative in both the core and outer PCAs. This indicates that, compared with the average commercial facility densities within the 1000 m reference range, commercial facility densities are higher in the secondary PCA than in the core or outer PCAs. In Sung’s study, the area with the highest density of large commercial facilities was within 250–500 m of the station perimeter [66]. This is quite different from subway stations; Ma et al.’s study of the Xi’an metro found that commercial facilities were mostly located in the core circle immediately adjacent to the station [67].

Density of public service facilities is strongly positively correlated with station ridership in the core and outer PCAs, but it shows little correlation in the primary PCA and a strong negative correlation in the secondary PCA. This suggests that the public service facilities in the JR station areas of the TMA are mainly concentrated in the core and outer PCAs, while the distribution is relatively scattered in other PCAs. Sung [66] also found similar findings in his study, reporting that the highest density of public service facilities in a metro railway station area was in the area 500–750 m around the station.

The functional diversity index is negatively correlated across all PCAs, and its absolute value decreases from inside to outside. This may be because compared with the degree of overall functional mixing within 1000 m of the station area, the distribution of facilities in each circle has a certain agglomeration, and the closer the core functions of the station are, the more concentrated they are. In other studies, such as Sung’s [66] study on railways in the Seoul metropolitan area, the degree of land use mixing within 500 m of the station area was also greater than within 250 m. The proportion of non-construction land in the core area and the main impact area is significantly negatively correlated, indicating that compared with the overall proportion of non-construction land in the station area, there is less non-construction land in the core area and the main impact area. A similar conclusion was also found in another study [68], which reported that the land development utilization rate in the peripheral areas of commuter rail transit stations is relatively low.

Bus stop density is significantly positively correlated with station ridership in both the core and radiating zones, with significance levels of 0.01 and 0.05, respectively, while showing no correlation in the primary PCA and a strong negative correlation in the secondary PCA. This suggests that bus stops are more tightly connected to rail stations in the core PCA. Road network density is significantly positively correlated in both the core and secondary PCAs, weakly correlated in the primary PCA, and negatively correlated in the outer PCA, indicating a gradual decrease in street network density from the core outward. Crossing density is significantly negatively correlated in the core and secondary PCAs, while showing a significant positive correlation in the primary and outer PCAs. This suggests that, compared with the station area average, crossing density is higher in the primary and outer PCAs. This is similar to Jun’s [34] findings on Seoul subway station areas.

4.3. Analysis of the Impact of the Built Environment on Passenger Flow

4.3.1. Multiple Linear Regression Results

To ensure the validity of the model fitting, we conducted correlation analysis, with multicollinearity tests among the variables before performing the analysis, removing the variables EC and sales of commercial goods, which had a variance inflation factor (VIF) greater than 7.5.

Table 3. Results of the multiple linear regression model.

	Core PCA		Primary PCA		Secondary PCA		Outer PCA
	β Std	t	Β Std	t	Β Std	t	β Std	t
Population density	−0.220 ***	−5.591	−0.244 ***	−6.321	−0.124 ***	−2.682	−0.106 **	−2.447
Commercial facility density	0.060	1.628	0.041	1.132	0.135 ***	2.863	−0.140 ***	−3.225
Public service facility density	0.056	1.563	0.372 ***	8.678	0.062 **	2.023	0.134 ***	2.909
Street network density	0.287 ***	6.060	0.288 ***	4.676	0.096	1.648	0.196 ***	4.317
Crossing density	−0.137 ***	−3.727	−0.153 ***	−2.649	0.056	1.570	−0.040	−0.933
Bus stop density	0.237 ***	7.128	0.170 ***	5.210	0.084 **	2.404	0.088 **	2.240
Functional diversity	−0.005	−0.170	−0.032	−1.131	0.026	0.770	−0.080 **	−2.424
Proportion of non-built-up land	−0.045	−1.473	0.021	0.698	−0.001	−0.015	−0.022	−0.602
Distance to CBD (km)	−0.186 ***	−4.102	−0.174 ***	−4.187	−0.173 ***	−3.306	−0.200 ***	−3.919
DC	0.501 ***	12.133	0.387 ***	9.647	0.588 ***	13.460	0.644 ***	15.052
CC	−0.127 ***	−3.329	−0.101 ***	−2.978	−0.117 ***	−2.881	−0.101 **	−2.431
BC	0.045	1.146	0.087 **	2.320	0.058	1.332	0.001	0.025
Years of operation	0.084 ***	2.888	0.039	1.462	0.090 ***	2.878	0.126 ***	4.088
Number of local resident population	0.023	0.705	0.063 **	2.086	0.068	1.865	0.061	1.641
Number of local employees	0.094 **	2.340	−0.016	−0.416	0.029	0.649	0.037	0.773
Cons		−0.703		−0.060		−1.575		−0.870
R2	0.73		0.78		0.68		0.68

*** p < 0.01, ** p < 0.05.

presents the multiple linear regression results for the different PCAs around JR stations.

The model’s R² for the primary PCA was 0.78, which was higher than the 0.73 in the core PCA and the 0.68 in the secondary and peripheral PCAs. This indicates that the land use characteristics in the primary influence area are relatively better at explaining station ridership. Moreover, the natural logarithms of population density, public service facility density, bus stop density, street network density, distance to CBD, DC, and CC all have substantial statistical significance across the PCAs.

Residential population density is negatively associated with station ridership, with larger coefficients in the core and primary PCAs compared with the secondary and outer PCAs. This differs from the findings of Jun et al. [34] for South Korean subway stations but aligns with similar results from Pan et al. [10] in their study of Shanghai metro stations. This may be because the surrounding environment of subway stations is different from that of suburban orbital stations or commuter railway stations. The impact of public services on ridership is significantly positive across all PCAs except for the core PCA. Commercial service facility density shows no significant impact on station ridership in the core and primary PCAs but exhibits a significant positive correlation in the secondary PCA and a significant negative correlation in the outer PCA. This suggests that commercial facility density primarily influences station ridership in the secondary and outer PCAs, with a pronounced promotional effect in the secondary PCA, while it plays a role in balancing employment and housing in the outer PCA.

Bus stop density positively and significantly impacts station ridership across all PCAs, with smaller standard deviations in the core and primary PCAs, indicating that improved transfer convenience in station areas is crucial for increasing JR ridership, particularly in the core and primary PCAs. Street network density also significantly promotes ridership in the core and primary PCAs, highlighting that the street network density within a 500 m radius of the station positively influences station ridership.

Distance to CBD and CC show negative correlations with station ridership, while DC, years of operation, and number of local employees show positive correlations. Notably, the effect of years of operation is more significant in the secondary and outer PCAs but not in the core and primary PCAs. The influence of number of local employees on ridership is significant only in the core PCA.

4.3.2. MGWR Model Analysis Results

The study used an adaptive bi-square kernel function to estimate the local and global parameters of the MGWR model. The “Golden Section Search” method was employed to determine the optimal bandwidth size by minimizing the corrected Akaike information criterion (AICc).

Table 4. MGWR model results.

	Core PCA		Primary PCA		Secondary PCA		Outer PCA
	Mean	IQR	Mean	IQR	Mean	IQR	Mean	IQR
Intercept	−0.0685	0.0064	−0.0362	0.0018	−0.0215	0.0009	−0.0206	0.0007
Population density	−0.244	0.0128	−0.2449	0.0006	−0.1326	0.0005	−0.1203	0.0006
Commercial facility density	0.028	0.0032	0.0204	0.0003	0.1137	0.0014	−0.1588	0.0007
Public service facility density	0.0332	0.0011	0.3454	0.0014	0.0586	0.0018	0.1424	0.0052
Street network density	0.279	0.0013	0.2711	0.0002	0.0886	0.0007	0.1932	0.0002
Crossing density	−0.1459	0.0028	−0.1751	0.0004	0.054	0.0017	−0.055	0.0009
Bus stop density	0.2626	0.0010	0.2035	0.1239	0.0986	0.0018	0.104	0.0006
Functional diversity	0.0238	0.0017	−0.0111	0.0011	0.0371	0.0016	−0.066	0.0025
Proportion of non-built-up land	−0.0386	0.1122	−0.0063	0.0027	−0.0174	0.0034	−0.04	0.0038
Distance to CBD (km)	−0.2886	0.0413	−0.1925	0.0020	−0.1888	0.0019	−0.2128	0.0012
DC	0.5128	0.1050	0.3933	0.0852	0.6049	0.0979	0.6627	0.1063
CC	−0.1439	0.0016	−0.1171	0.0012	−0.1369	0.0006	−0.1197	0.0015
BC	−0.0001	0.0016	0.0514	0.0009	0.0228	0.0018	−0.0388	0.0012
Years of operation	0.0683	0.0113	0.0302	0.0020	0.0858	0.0022	0.1227	0.0025
Number of local resident population	0.0373	0.0128	0.0903	0.0011	0.0802	0.0028	0.0729	0.0030
Number of local employees	0.0582	0.0032	−0.0255	0.0027	0.0193	0.0011	0.0229	0.0011
R2	0.76		0.81		0.70		0.70
AICc	603.7404		524.7108		689.195		683.5094
R2 (GWR)	0.76		0.77		0.67		0.67
AICc (GWR)	628.5064		567.0857		707.8051		706.2423

reports the MGWR results for the core, primary, secondary and outer PCAs.

The MGWR model exhibited a higher R² and a lower AICc than the linear regression and GWR models. Similar to the linear regression model, the primary impact area and core area models had higher adjusted R² values and lower AICc values than the secondary and outer PCAs. The direction of the variable coefficients in the MGWR model aligned with that of those in the linear regression results. Public service facility density, bus stop density, and street network density showed positive correlations with ridership, indicating that these variables promote station ridership. The density of commercial facilities was found to have a positive impact on ridership in the core, primary, and secondary PCAs, but a negative impact in the outer PCA. The mean values of residential population density and the proportion of non-built-up land were negative, reflecting the specific characteristics of JR commuter rail stations and the land use development patterns in the station areas. Among the station attribute variables, the mean value of DC was positive, while the mean values of distance to the CBD and of CC were negative.

Notably, among the density indicators, the mean value of public service facility density was highest in the primary PCA. However, the mean value of commercial facility density in the secondary PCA was higher than in the other PCAs, indicating that, on average, density of public service facilities in the primary PCA had a more significant effect and density of commercial facilities in the secondary PCA had a greater impact on station ridership. Similar but not completely consistent findings were also reported in previous studies. For example, in the study by Chakour and Eluru [69], the impact of large-scale commercial facilities on station ridership was most significant within 500 m around the station, and the impact of large-scale public service facilities was most significant within 250 m around the station. In Sung’s [66] study, the impact of public service facility density on ridership was primarily observed within a 1000 m radius around the station.

The influence of population density on passenger flow is negative, indicating that the ridership of JR stations in the Tokyo metropolitan area with a small population within 1000 m is larger, and that the lower the population density with 500 m around the station, the larger the passenger flow.

The impact of population density on ridership is negative, indicating that stations with fewer residents within a 1000 m radius in the TMA tend to have higher ridership, and this is particularly the case for a 500 m radius. Similar results were found in studies by Pan et al. [10] on the Shanghai metro and Chen et al. [70] on the Wuhan metro, which indicated that a higher number of jobs within a 500 m buffer zone around a station was associated with greater ridership, while the results regarding residential population numbers were the opposite. However, Li et al. [44]’s research on the Guangzhou metro and Jun et al. [34]’s study regarding ridership at Korean metro stations found positive correlations between population numbers and station ridership. These inconsistencies may stem from differences in urban form. In the Tokyo metropolitan area, railway station areas exhibit higher development density and functional diversity compared with those in regions like North America [71]. Additionally, Japan’s aging population, declining birth rates, and suburban relocation of enterprises and schools have led to the decline of traditionally residential station areas. Such areas now experience lower ridership compared with station areas with mixed or commercial functions, which are better equipped to attract passenger flow [60].

Regarding the diversity indicators, the mean value of functional diversity was positive in both the core and secondary PCAs, indicating that, on average, functional diversity in these PCAs promotes ridership. This finding is supported by the research of many scholars, such as Sung and Oh [26], who found that dense land use and functional mix within a 500 m radius of Seoul metro stations contributed to increased transit ridership. Additionally, Sung [11] reported a statistically significant positive correlation between ridership and functional mix within 250 m of the station. The absolute value of the proportion of non-built-up land was higher in the outer and core PCAs than in other PCAs, indicating that on average, non-built-up land has a stronger inhibitory effect on passenger flow in the outer and core PCAs.

Regarding the design indicators, both street network density and bus stop density declined progressively from the core PCA to the outer PCAs, indicating that the convenience of transportation connections is particularly important for passenger flow in the core PCA. Although there have been few studies that have analyzed the relationship between street network and bus stop density with ridership across different zones, many studies have demonstrated the positive impact of transit connections on station ridership [2,15,29,66,72]. Additionally, Cao and Tao [1] noted that there may be differences in land use and travel behavior thresholds among metropolitan areas of varying sizes and levels of economic development.

Among the station attribute factors, the average coefficients for the distance to the CBD and the CC indicator were negative, while the average coefficients for DC, BC, and years of operation were positive. This indicates that stations closer to the CBD and those with higher DC and BC values tend to have greater ridership. This may be because stations with high DC often serve as multi-line transfer hubs within the rail network, while those with high BC are typically located at key transit interfaces within the network. These findings align with the results of Cao [20]. Additionally, the impact of station age was found to be greater in the outer PCA compared with other PCAs, which may be due to the longer development time required for areas further from the station; thus, the influence of stations’ length of operation is more pronounced in these peripheral regions. The study by Liu [73] also suggested that the impact of station age on ridership is related to the station’s development stage. This highlights the importance of orderly development in station areas based on station location and development stage.

Figure 7 and Figure 8 display the local parameters for commercial facilities and public services facility density. The density of commercial and public service facilities has had a positive impact on ridership to varying degrees. Commercial facilities in the core PCA exert a stronger positive influence on ridership along the Yokohama and Chiba corridors, while in the primary and secondary PCAs, they significantly enhance passenger flow at stations in the core area of the Yamanote Line. It is noteworthy that the density of commercial facilities in the secondary and peripheral zones has a statistically significant impact on ridership, unlike in the core and primary influence areas. Similar conclusions were drawn in Sung’s [66] study of railroad stations in the Seoul metropolitan area, where large commercial densities within 500 m of the station had a greater coefficient of influence on patronage than those within 250 m, and in Chakour’s [69] study of the Montreal Metro and Commuter Railway in Canada, where it was found that the influence of commercial establishments in a 600 m perimeter around the station was greater than that of those within a 200 m perimeter. However, this differs from the findings of some studies on metro station areas, such as in Ma et al.’s [67] study on the Xi’an metro, where commercial facilities were found to have higher travel attractiveness within 400 m, gradually decaying towards the periphery.

The effect of public service facilities on ridership is more significant in the primary and outer PCAs, and the magnitude of its effect is characterized by a spatial distribution that is greater in the Tsukuba and Chiba corridors and the Tama area compared with the other areas in the primary PCA and the outer PCA, respectively. This result may be related to the higher density of commercial and public service facilities near the Yamanote Loop, and the commercial agglomeration along the Tokyo Bay corridor in Yokohama and Chiba.

Figure 9 and Figure 10 illustrate the variable coefficient distributions for functional diversity and bus stop density across the four PCAs. In the core PCA, the level of POI entropy positively influences ridership at stations in the northwest of the TMA, while in the secondary PCA, this variable significantly enhances passenger flow throughout most of the metropolitan area, with minimal impact at the peripheral edges. Furthermore, with the exception of the outer PCA, most local coefficients for functional diversity are not statistically significant, a finding that aligns with the results of the multinomial logit model (Table 4). This may be attributed to stations located in the metropolitan center, which are surrounded by employment, business, and other relatively dense functions. The higher degree of functional complexity in the core PCA is more likely to attract visitors. In contrast, stations situated at the periphery of the metropolitan area are predominantly residential, where increased functional complexity leads to a higher degree of localization for residents, resulting in a lower proportion of long-distance trips on the railroad. Although no scholars have specifically studied the relationship between functional mix and station ridership across different zones, research by Lee [74] on the Seoul metropolitan area found that high-density functional mix in the urban center and secondary centers promotes TOD. In contrast, self-sufficient functional mix in the peripheral areas of the metropolitan region is beneficial for compact development.

Finally, in the core and primary PCAs, a significant impact of bus stop density on ridership was observed in most areas of the metropolitan center, while in the secondary and outer PCAs, strong effects were found only in the eastern and northern regions of the metropolitan area. These findings indicate that JR ridership is influenced by the connectivity of bus facilities surrounding the stations and particularly, by the convenience of bus transfers within the core and primary PCAs. On the other hand, for stations in the peripheral areas of the metropolitan region, the impact of bus stop density in the secondary and outer PCAs on ridership is statistically more significant than in the core and primary PCAs. This suggests that suburban stations have a wider catchment area for ridership and are more affected by the convenience of bus transfers at nearby bus stops. Research by Li [44] also found differences in the impact of bus stop density on ridership between inner and outer suburban areas in the Guangzhou metro system.

4.4. Policy Recommendations

This study reveals that public service facilities and residential populations are mostly concentrated in the core and primary zones of Tokyo’s JR station areas. The impact of land use on passenger flow exhibits spatial heterogeneity across different zones. Overall, public service and commercial facilities, the number of bus stops, and street network density significantly enhance station ridership, whereas residential population density has the opposite effect. The findings of this study have several implications.

First, the mechanism of passenger flow and the land use characteristics in commuter rail stations differ from those of metro stations. Compared with residential population density and functional diversity, the density of commercial and public service facilities has a more significant impact on attracting passenger flow to commuter rail stations. Therefore, it is necessary to enhance the provision of service facilities based on the needs of different types of stations. Second, intermodal connectivity indicators, such as bus stop density, road network density, and intersection density, are critical for increasing passenger flow at commuter rail stations, particularly in suburban areas where convenient transfer connections are a key method to boost passenger volumes. Finally, this study suggests that the development of TOD for commuter rail stations in high-density metropolitan areas should focus on planning and designing the 500 m zone around stations, with particular attention paid to public service and commercial facilities.

These implications align closely with Japan’s current urban and transportation development policies. For instance, the revised Special Measures for Urban Regeneration Act encourages clustering urban functions such as medical care, elderly care, commerce, and public transportation around rail stations, with layouts differentiated by concentric zones [75]. The Ministry of Land, Infrastructure, Transport and Tourism (MLIT) has introduced a transportation hub integration strategy incorporating buses, LRTs, and bicycles to enhance intermodal connectivity [76]. Furthermore, MLIT emphasizes the placement of elderly care facilities within a 500 m radius of rail stations, a guideline similarly adopted by the City of Nagasaki, which designates this radius as a core influence zone for station planning [77].

5. Conclusions

A hierarchy spatial analysis of TOD in metropolitan railroad station areas can provide more specific and targeted guidance for TOD planning and design in these areas. This study considered JR stations in the Tokyo metropolitan area as a case study, employing a multinomial logit model to analyze the land use characteristics of different PCAs within a 1000 m influence range. Additionally, linear regression and MGWR models were used to validate the impact of land use and socio-economic attribute indicators on station ridership across different PCAs. The main research findings can be summarized as follows.

First, land use around JR stations in the Tokyo metropolitan area is significantly more clustered within a 500 m radius than at the periphery, dominated by public services and commercial facilities. Second, the influence of land use elements on station passenger flow varies across different PCAs, reflecting spatial heterogeneity in the impact of the built environment. The key factors influencing station ridership include residential population density, public service facility density, and density of bus stops, with a stronger influence in the core and primary PCAs than in the secondary PCA. Third, the MGWR model offered stronger explanatory power compared with the stepwise regression model when analyzing the impact of the land use characteristics on passenger flow in different PCAs around JR stations. Overall, the land use characteristics within a 500 m radius of stations show the most explanatory potential for analyzing station ridership.

The rational planning and strategic layout of land use in metropolitan rail transit station areas are fundamental to sustaining passenger flow, with a particular emphasis on public service facilities and feeder connections within a 500 m radius of the station. This study of JR station areas in the Tokyo metropolitan area serves as a well-established example of station area development in a high-density metropolitan region.

This study has certain limitations that should be acknowledged. First, this research was centered on the Tokyo metropolitan area, characterized by a highly developed rail network and high urban density. The applicability of the findings to regions with underdeveloped rail transit systems or lower construction densities has yet to be confirmed. Additional case studies are needed to provide broader references for the development of rail-based metropolitan areas. Second, this study relied on average daily passenger flow data, making the findings more applicable to station area design at a relatively macro level, while omitting analyses of peak commuting patterns along transit lines. Future research could leverage dynamic passenger flow data from across various dates and times to conduct more granular spatio–temporal analyses.