1 Introduction

Transport facilities should be made accessible to all and based on the planning and scientific viewpoints that promote social justice. However, these factors are usually overlooked during planning and evaluation of conventional transportation modes that are usually motorized (Litman 2013). Therefore, a transportation planning approach that combines overall affordability with social and environmental factors should be used. Considering the conventional transportation modes, it is believed that roads cannot be built fast enough to keep up with the rising demand (Ewing and Cervero 2010). That is why state departments throughout the world are turning to strategies for improved transport planning to meet the travel demand in a more sustainable way. To foster urban and transport sustainability, transit-oriented development (TOD) has been most successful since the 1980s (Ibraeva et al. 2020). “TOD can be described as land use and transportation planning that makes cycling, walking, and transit use desirable by focusing development around public transport stations/stops” (Thomas and Bertolini 2017). This means that TOD focuses on planning principles that promote equitable transport distribution by locating the stations in a way that increases the tendency to walk and cycle.

Walking stands as a major mode of access to public transport and has been studied with respect to many factors including footpath quality, density, land use mix, and weather (Cervero et al. 2009; Novales et al. 2021; Pafka and Dovey 2017; Tao et al. 2018). Its importance as a mode of access for public transport is evident from the literature (see Table 1). The table shows the walking trips to transit stations as a percentage of total access trips. It can be observed that walking is the main mode of access for transit all around the world. However, the mode share of the walk is quite high in European, Australian, and American cities as compared to the Asian cities, such as Bangkok in Thailand and Dhaka in Bangladesh. The lesser share of walking in Bangkok can be attributed to the dominance of their conventional motorcycle taxi for accessing transit, which has a mode share of 30% (Chalermpong and Wibowo 2007).

Table 1 Mode share of walking for accessing transit services in different cities around the world
Full size table

Walking as a major mode of access to public transport requires keen planning while locating a station/stop. The tendency to walk in an environment depends on socio-demographic, socio-economic, and travel behavior data. Past research efforts revealed that areas with more intersections, high population density, and more interconnected streets result in more pedestrians (Cervero et al. 2009; Xiao et al. 2020). On the other hand, sparse areas that are car-oriented having spacious roadways prevent walking (Cervero et al. 2009). Other research studies have shown that socioeconomic factors like income, age, and gender are non-significant in explaining walking (Daniels and Mulley 2013; Kamruzzaman et al. 2016). Most research studies on station accessibility have a common key finding that streets in the vicinity of stations have a high pedestrian frequency (Kamruzzaman et al. 2016). Especially, areas within the radius of 1 km have been found walking-dominant in terms of accessing public transport (Cervero 2001). These pedestrian-dominated regions around PT stations are referred to as catchment areas or ped-sheds. This pedestrian-rich catchment area has been the focus of many research studies aiming at station accessibility.

Prior studies have approached the problem of station accessibility through conventional methods that rely on the proximity of the station (Foda and Osman 2010). Additionally, the limited research that uses advanced methods fails to relate station accessibility with walking lengths on actual pedestrian networks (Ibraeva et al. 2020). The literature has rather related station accessibility to factors such as urban infrastructure, transit service operational characteristics, simple radial proximity, and pedestrian safety (Cervero et al. 2009; Sun et al. 2018; Xiao et al. 2020). Therefore, it is justified to conduct a study that uses advanced mapping tools and practical methods to determine walking accessibility to transit stations using an actual pedestrian network. This approach can be utilized in locating the future transit stations with an aim to facilitate walking as a mode of access to transit services.

The next section of the paper presents the detailed literature review focusing on walking accessibility to public transit stations under different environments. After the literature review section, the data for the selected case study is described followed by the methodology describing different methods used to evaluate walking accessibility to public transit stations. The results and discussion section is then presented which includes results of walking accessibility to BRT stations and their implications. At the end, major conclusions of the research study are provided which also include limitations of the study.

1.1 Literature review

Different researchers have tried to improve the estimation of walking accessibility in different ways. Ziari et al. (2007) used mathematical relations to find that station spacing of 1500 m results in better vehicle speed and reduces total travel time, hence, improving accessibility by attracting more passengers. Alam et al. (2010) used a gravity model to estimate public transit accessibility to jobs in Florida, Broward County. Instead of using the arbitrary value of a distance-decay parameter in a gravity model, the authors estimated the parameter using socioeconomic survey data in Sacramento County. The estimated distance-decay parameter was applied to Broward County. The study results revealed that the method produced more realistic accessibility values than the simple circular buffer method. The study also demonstrated the transferability of estimated distance-decay parameters from one geographic unit to another (Alam et al. 2010). Foda and Osman (2010) considered the actual road network by making pedestrian catchment areas around the stations. The authors used accessibility indices to evaluate walking accessibility and found that the circular radius method overestimates the accessibility as compared to pedestrian catchment (Ped-shed) by 50%. Eboli et al. (2014) used the Transit Capacity and Quality of Service Manual (TCQSM) method to calibrate the public transit service area coverage (area within walking distance of a public transit stop) in an Italian city. The study results revealed that the street connectivity factor, the population factor, and the grade factor have a significant influence on the public transit service area coverage. Vale (2015) improved the original TOD classification model of Bertolini (1996) by incorporating the pedestrian catchment method concept with 700 m favorable walking distance. This shows the importance of the pedestrian catchment method in dealing with older issues more practically. Bertolini (1996) did a cross-sectional study to analyze the location of transit stations, particularly heavy rails, in the surrounding environment and found that transit stations are places of diverse activities that are affected by global and local dynamics. The research also suggested that transit stations should be given due consideration and treated both as a node and a place while defining a redevelopment strategy in the surrounding land use. Pafka and Dovey (2017) estimated the accessibility to services inside the pedestrian catchments from a transit station. The authors used area-weighted average perimeter and catchment interfaces, a measure based on area and perimeter of blocks within a catchment area. These measures take both street width and block size into consideration. The study found that short blocks that demonstrate a street network with rich interconnectivity performed relatively well. However, the study also found that if the frequency and intensity of short blocks increases, the blocks impede walking due to more pedestrian crossings that act as hurdles. Sarker et al. (2020) used survey data and found that pedestrians prefer a shorter and most direct path to stations in Munich and that pedestrian catchments in high urban density were smaller and more walking-efficient. Sun et al. (2018) studied the competitiveness of public transport and car in Shenzhen city by calculating the travel time ratio of PT and car. This ratio is a time-based analogy to Pedestrian Route Directness test, which is distance-based. The study also established that the value of the travel time ratio below 1.5 shows effective PT service. Alawadi et al. (2021) used the Pedestrian Route Directness (PRD) test for the walking efficiency of streets in different urban environments. The authors utilized the actual catchment area concept introduced by Foda and Osman (2010) to find that the inclusion of alleyways along with the street network can improve the accessibility indices values. Both PRD and actual pedestrian catchment consider the street network characteristics directly in a more practical way. Rahman et al. (2022) studied the first and last mile phase of a multimodal transit trip in the city of Dhaka, Bangladesh. The authors used multinomial logit (MNL) and nested logit (NL) models to find the mode choice of access and egress to different public transport (PT) types. The study concluded that travelers are more likely to use motorized trips at egress than at access, the probability of which is higher for advanced transit types like commuter rail. Rayaprolu et al. (2022) used the concept of iso-access lines that show the same level of utility for different parameters of two transit systems at the same budget. The study proposed an improved combination of two different transit types—A through-routed service with high speed and large station spacing and local service with high user access and spatial coverage. The authors found that the through-routed service is cost-effective at a longer access time (larger buffer), and the local service is cost-effective at a shorter access time (smaller buffer). However, the study did not consider the walking access directly but rather considered multimodal transit access to opportunities optimized in monetary terms. Ramos-Santiago (2022) studied the number of feeder bus passengers directed to rapid transit stations as a function of land use and built environment characteristics in Los Angeles. A statistically significant relation was found between built environment factors around feeder bus stops and the associated passenger boarding at transit stations. Although the research studied the transit patronage provided by feeder bus routes, it found that walking is still the dominant mode of access to transit. Cui et al. (2022) examines the relationship between accessibility to jobs and daily bus stop boarding in Portland, Oregon. The study observed a 1.8-2.0% increase in boarding against a 10% increase in accessibility. However, no consideration was given to the walking mode; rather the accessibility was computed generally. Lastly, the study concludes that optimizing the station location can improve transit patronage and operational efficiency, and can even result in cost savings.

The walkability and land use mix in the vicinity of a transit station has a high impact on mode choice decisions of the commuters (Rahman et al. 2022). The accessibility to the transit stations (bus and rail) can be improved through policy decisions and appropriate land use modifications. The accessibility improvement is dependent on transport policy (Lo et al. 2008). With a modification in the infrastructure the accessibility to use public transport can be enhanced. In Hong Kong, for example, the access to a transit station is provided through a shopping mall (Lau et al. 2005). In one study, the characteristics of transit stations were identified as one of the important factors to increase the ridership (Loo et al. 2010). The research done so far ignores the effect of station location on walking accessibility (Ibraeva et al. 2020). The importance of this research is felt particularly in case of Pakistan because of the growing number of new public transit projects and the lack of research on its public transport system (Baker et al. 2018). In one recent research project, inequalities in walking accessibility to public places has been identified (Khattak et al. 2023). Further, Pakistan has much lower trip satisfaction, like other developing countries, when it comes to motorized feeder modes to transit (Khan 2021). Additionally, with regard to bus station design, Voß et al. (2020) introduced an innovative approach in bus station design, combining mystery shopping and virtual reality. This method uses mystery shopping for service quality assessment and virtual reality for realistic environment simulations, offering a comprehensive tool for improving public transportation infrastructure and services.

The aim of this research is to analyze the effect of public transit station location on walking accessibility to it by using advanced methods of actual pedestrian catchment and PRD test. The objectives set for this study are to:

  • Estimate the walking accessibility to transit stations based on the actual street network.

  • Evaluate the effect of station location in different urban environments on walking accessibility.

  • Give recommendations on the location of stations for future public transit services.

1.2 Case study

The Rawalpindi-Islamabad Bus Rapid Transit (BRT) is a public transit service in the twin cities of Pakistan. The first phase of this project, shown in Fig. 1, is taken as a case study that was first operated in 2015. The service extends over a 22.5 km stretch from Pak-Secretariat Islamabad to Rawalpindi Saddar and has 24 stations. Rawalpindi has 11 stations, and the entire 8.6 km route is elevated. The rest of the 13 stations are in Islamabad over a length of 13.9 km. The local street network in both cities is different from each other. Islamabad is a well-planned city built in the 1960s and its street structure follows a grid-iron pattern (Capital Development Authority Islamabad 2012) and in Rawalpindi, it is closer to a spider-net pattern. The project route has been selected because of the diverse nature of the two cities giving an opportunity to study the effect of station location on walking accessibility in more versatile environments.

Fig. 1
figure 1

Map showing the Route of Rawalpindi-Islamabad BRT phase 1

Full size image

The Rawalpindi-Islamabad BRT project is a transport facility in the two cities of Pakistan. It is rated in the Silver category as per ITDP (Institute for Transportation and Development Policy) BRT rating scale, since it somewhat lacks in non-motorized access (Haider et al. 2021). In 2012, there were over one million trips taking place within Islamabad every day including half a million trips to and from Rawalpindi of which the PT mode share was 35% (Capital Development Authority Islamabad 2012). The feasibility study conducted by one of the oldest local engineering consulting firms M/S NESPAK (Pvt.) predicted a daily ridership of 139,000, which was found later to be ranging between 100,000 and 125,000 with a headway of 3 min (Haider et al. 2021).

Table 2 Important land use activities around BRT stations within 400 m radius
Full size table

2 Methodology

2.1 Concept of ped-shed

In most of the past research, the circular buffer concept has been followed to assess accessibility to transit stations by walking mode (Salvo and Sabatini 2005). This method does not take into account the real distribution of the population and corresponding road network around the station resulting in an overestimated accessibility to a BRT station (Foda and Osman 2010; Salvo and Sabatini 2005). Since this method assumes that people can access the station from anywhere within the buffer ring and ignores the actual topology of the pedestrian pathways surrounding the stops, therefore estimated accessibility results may not reflect actual results.

A limited road network and its connectivity within the buffer to access BRT stations may not allow riders to access the BRT station directly. To overcome this limitation, a network-constrained approach can be adopted assuming a walking threshold up to 400 m approximately from the BRT station. In literature, this area is known by the names of Ped-shed, Pedestrian catchment and Actual buffer area (Ibraeva et al. 2020). To find it, the 400 m distance will be measured along the road network in all directions instead of using a circular buffer. The actual distance-based buffer in all directions from the station can be estimated using a GIS application after modeling pedestrians’ walkways along with station locations. Using an inbuilt tool of Network Analysis in GIS application, 400 m fragments of all the roads surrounding the station were identified. Connecting the ends of these roads, the Ped-sheds around the station were established. Figure 2 shows the difference between the circular buffer and the actual accessible area called Ped-shed.

Fig. 2
figure 2

Showing circular buffer and Ped-shed around the station

Full size image

2.2 Ideal station accessibility index (ISAI)

In order to quantify the walkability inside the ped-shed some parameters are required. The parameters that have been selected are road density and road connectivity inside the ped-shed. Connectivity is discussed in the next section. The road density in this study is represented by ISAI as used in the past research (Alawadi et al. 2021; Foda and Osman 2010). This index shows the accessibility in terms of the road length inside the ped-shed that can be used by pedestrians to reach a public transit station, also known as Metric Reach (MR). Unlike other variables, ISAI considers the total reachable road length as well as considers the area of the circular buffer allowing the comparison of different radii and study zones (Ibraeva et al. 2020). ISAI is calculated as:

$$ ISAI= \frac{{L}_{A}}{{A}_{C}}$$

where, (see Fig. 2) ISAI is the Ideal Station Accessibility Index, \( {\text{L}}_{\text{A}}\) represents the actual accessible road length inside the Ped-shed and \( {\text{A}}_{\text{C}}\) is the area of the circular buffer.

2.3 Pedestrian route directness (PRD) test for street efficiency

The accessibility index value shows the accessible street density of the area and does not provide complete knowledge about the spatial distribution with respect to the efficiency of roads around the transit stations. Rather, it estimates the intensity of road supply around the station. In other words, the road network around a station may be very dense (more intensive road supply) but due to the presence of cul-de-sacs and obstructions, the street connectivity can be very low. A denser road network results in a very high ISAI value but due to limited street connectivity, actual walkability to access a station would be low. This limitation makes the ISAI method inefficient.

To account for this limitation, the PRD test is applied to the station locality within adequate walking distance. The use of this test has been favored by the Dubai Urban Street Design Manual (Alawadi et al. 2021). This design manual recommends a maximum value of PRD of 1.5 for a good and efficient street network. The value of PRD indicates the ratio of the actual walking distance between two points to the straight crow-fly displacement, also known as Euclidean distance, as shown in Fig. 3. This value has a minimum value of “1” and indicates the straight radial path from station to point of interest. The formulation to estimate PRD values is given below:

Fig. 3
figure 3

PRD: Ratio of actual walking distance to radial displacement

Full size image
$$ PRD = \frac{L}{{L}_{R}}$$

where, L represents the length of the actual road from the station to the point of interest (here public transit station), and \( {\text{L}}_{\text{R}}\) is the length of the theoretical radial road from station to point of interest.

For this study, the PRD value has been calculated for each station by selecting points randomly at a distance of 200 and 400 m from stations and then taking their average.

2.4 Data

The data for the research is taken from online sources. The street network is acquired from online OSM (Open Street Map) libraries. The road network map is imported into the QGIS application. The station locations are marked on the map from Google Earth’s coordinates. The road network and pedestrian pathways around the stations were observed and physically visited and adjustments were made accordingly in QGIS. The adjustments include solving the network topology errors such as the stations’ entry and exit point connections to the street network and correcting the pedestrian network in the station vicinity. Direct connections of major highways to stations are allowed only where pedestrian facilities are provided or where noticeable pedestrian activities are observed.

The buffer radius around the station has been finalized from the research literature. Research states that the 1000-meter distance around the transit station is pedestrian-dominated in terms of mode of access (Cervero 2001). Around 34% of the total access trips are walking-based and seldom exceeding 2 km (Chalermpong and Wibowo 2007; Daniels and Mulley 2013). However, most researchers use a radius of 400 to 600 m as a preferable walking area around the transit station considering a 1.3 m/s average walking speed and 5 to 7 min of walk on each access and egress (Foda and Osman 2010). The exact figure depends upon age, environment, and weather protection (Pafka and Dovey 2017). For Islamabad, having long summers, a 400 m walking distance can be considered as appropriate.

3 Results and discussions

Results have been presented descriptively in Table 2 showing the values of the ISAI for all the stations. The table has been divided into two parts for stations in Islamabad and Rawalpindi. The PRD values have also been shown for each station. The results have been calculated using the buffer radius and actual walking distance of 400 m. The results will be categorized and explained further in the coming sections.

Table 3 Results of ISAI and PRD Analysis
Full size table

3.1 Walking accessibility to public transit stations in Islamabad

The graph illustrating the ISAI and PRD values of the stations in Islamabad that have passed the PRD test (PRD less than 1.5) is presented in Fig. 4. The Stations failing the PRD test have been filtered out. The figure also shows the corresponding street network configuration of each station. The stations are arranged/sorted with respect to decreasing values of ISAI. The street network visualized in the figure helps in understanding the values of indices more clearly with respect to the street network configuration.

Figure 4 shows that the station “Chaman” has one of the highest values of ISAI, and “Katchari Chowk” has the lowest value. The street configuration changes from a dense and looping structure at the left to the network having more dead ends at the right. This follows the decreasing trend of the ISAI value from left to right showing that the absence of looping in the network and less density badly influence the accessibility. Moreover, it can be said that the looping in the road network itself is due to closely spaced residential and commercial activities. Nevertheless, looping can result in too many pedestrian crossings, which start acting as hurdles at some point (Pafka and Dovey 2017). However, it is certain that walking accessibility values are higher in dense and looping street configurations.

The main thoroughfares in the street network have been used in the analysis where pedestrian walkways are present. It is worth noting that there are high-rise buildings (3–15 story) near the stations of “Shaheed e Millat” and “Parade Ground”. So, a single road serves the floor area of multiple stories unlike the case of low-rise residential or single-story buildings. If we consider the single road multiple times for each story of these high-rise buildings, the ISAI value will increase noticeably. This is one of the limitations of the method used in this study. But the station location having high-rise multi-story buildings in the vicinity can result in a greater number of pedestrians for a smaller ground area. This conflict has been highlighted by (Rayaprolu et al. 2022) stating that the land use distribution should follow a pyramidal pattern, rather than a uniform pattern, having high-rise dense buildings near stations and low-density development at the periphery.

Fig. 4
figure 4

Index values for stations passing the PRD test in Islamabad

Full size image

3.2 Walking accessibility to public transit stations in Rawalpindi

The results are shown in Fig. 5 for Rawalpindi city in a similar way as for Islamabad. The figure shows only those stations that have passed the PRD test (PRD less than 1.5) and all those stations failing the PRD test have been filtered out. The figure has the stations arranged/sorted with respect to decreasing values of ISAI from left to right.

Figure 5 shows that the station “Waris Khan” has the highest value of ISAI and “Saddar” has the lowest value. The street configuration changes from a very dense and radial type structure at the left (at “Waris Khan” and “Committee Chowk”) to a network having more cul-de-sacs and grid-type structures at the right. This follows the descending trend of ISAI value from left to right showing that a dense radial/spider-net type network configuration like “Waris Khan” is likely to be more accessible than loose-grid type and configuration comprising dead-ends.

Further, the “6th Road” station has a high number of commercial activities on service roads and the presence of pedestrians on roads. It has grade separation for through traffic, U-turns, and pedestrian crossings. This environment depicts the concept of shared space more closely, integrating the pedestrians and vehicles to use the road simultaneously. That is why direct access has been given from stations to the roads. Owing to that, it has more ISAI value even though the network density is lower as we move away from the station. This underlines the fact that station location in infrastructure environments that closely relate to shared space are likely to promote walking to public transport. A similar observation is noted for the “Shamsabad” station.

Fig. 5
figure 5

Index values for stations passing the PRD test in Rawalpindi

Full size image

3.3 Stations failing the PRD test

Figures 6 and 7 show the stations of Islamabad and Rawalpindi cities, respectively, which have failed the PRD test. In these figures, the stations are arranged/sorted in descending order based on the value of ISAI from left to right.

Fig. 6
figure 6

Public transit stations failing the PRD test in Islamabad

Full size image
Fig. 7
figure 7

Public transit stations failing the PRD test in Rawalpindi

Full size image

Table 3 shows the discrepancies in the network around the stations failing the PRD tests (PRD greater than 1.5). The stations either have a discontinuity in the surrounding network, presence of interchange, low-density lands, or large premises in their vicinity.

The “7th Avenue”, “Kashmir Highway” and “Faizabad” stations have interchanges. “Kashmir Highway” has a bigger interchange (inverted/partial cloverleaf) and eventually a higher PRD value and lower ISAI value as compared to the “7th Avenue” and “Faizabad” station. This grade separation in the interchange discontinues the network and thus the PRD value increases.

The other stations have discrepancies like a discontinuity in the pedestrian path and the presence of large or deserted bodies. The “Chandni chowk” station has a dense network which resembles a grid-iron type, but the immediate vicinity of the station has no direct pedestrian paths emerging radially from the stations. The same problem exists in the case of the stations “Rehmanabad”, “Stock Exchange” and “PIMS Hospital”.

Large bodies of land means the presence of big institutions, depots, and electricity grid stations near the transit stations that create discontinuity in the surrounding network; details are presented in Table 3. All these factors create discontinuity in the road network and decrease the network efficiency (increased PRD) hence impeding the tendency to walk.

Table 4 Problems in the street network around the stations failing PRD, Islamabad & Rawalpindi
Full size table

Figure 8 shows five different types of street network configurations that were observed in the twin cities, in sequence of decreasing walking accessibility to transit stations. The primary vertical axis (corresponding bar chart) shows the average values of ISAI, and the secondary vertical axis (corresponding line chart) shows the frequency of the PRD test passing for each type of street network. It has been observed that the street networks having a spider-net shape have the highest ISAI values and the highest likelihood of passing the PRD test, followed by the grid-iron street network and the shared space environments. The street networks that have major interchanges and disconnected streets have the lowest values of ISAI. Moreover, all the street networks, classified into these last two categories, have failed the PRD test.

Fig. 8
figure 8

Average ISAI and Percent Passing PRD Test on BRT Stations by Street Type

Full size image

4 Conclusions

This study aims at estimating the walking accessibility of the transit stations using actual pedestrian distances and advanced mapping tools. It also compares the performance of stations located in different street network types and characteristics using measures of accessibility. The main innovation of this study lies in the use of the actual walking distance for pedestrians to determine how the street network configuration and features affect the overall walkability toward transit stations—a technique not found in previous research. This approach can be employed in planning the locations of future transit stations with the goal to boost walking as a means of access to transit services. The study area selected in this study is Islamabad-Rawalpindi BRT. This project spans the twin cities of Islamabad and Rawalpindi, Pakistan, allowing the analysis on walking access to public transit stations in diverse environments. Some major findings of this study are:

  • The stations located in the radial spider-net-shaped network have a higher value of ISAI and, hence, are more accessible than the grid-iron network. Rawalpindi-based stations have a better performance than the grid-iron network in Islamabad in general.

  • The road network in Islamabad has a higher value of ISAI when the street network has more loops and is interconnected.

  • The station location that has high-rise buildings in the vicinity can serve more pedestrians with a lower ISAI value. However, further studies are required to ascertain this.

  • Stations located near shared-space environments also have the potential for promoting walking for transit.

  • There should be direct radial paths emerging from the station into the surrounding areas in order to increase the network of their pedestrian access routes and efficiency.

  • Stations that are located near large uninhabited areas like power grid stations, depots, major interchanges, and deserted areas discourage walking as well as lower the walking efficiency of a network.

Based on the findings, the research article proposes an ideal case scenario of the transit station location. It proposes that the stations should be in densely populated commercial areas having high population density. Large uninhabited areas and traffic interchanges should come at the peripheries of the accessibility buffer. A shared-space concept should be encouraged near stations. There should be direct radial paths near the station connecting it to the inner areas of the ped-shed. In residential areas, paratransit, and shuttle services (feeder route buses) should be provided to increase the coverage of the BRT stations. The method used in this study can be improved by considering the effect of multi-story buildings in the analysis. Since the current study is based on road network length alone, it ignores the effect of buildings with many inhabitants that can occur in the actual accessibility buffer or ped-shed. By considering the total floor area inside the ped-shed, a more precise network of a station’s pedestrian access routes can be obtained. The future prospect in this area should also target the correlation between accessibility indices with the real passenger counts on the stations. This will conclude the effect of the station location on the transit ridership.