1 Introduction

Railway system planning is a multi-branched restrained problem in which every stakeholder has its own goals and requirements. The key aspect is to find a way which allows acceptable solutions, taking stakeholders' divergent interests into consideration. It is crucial to identify and fully understand precisely what kind of system structure can capture diverse stakeholders’ perspectives for an efficient and effective participation process in rail transport planning.

Planners and governmental agencies are becoming increasingly interested in decision support techniques capturing the impacts that cannot be easily quantified for rail transport planning. Decision-makers are likely to perceive a role for multi-criteria decision analysis (MCDA) in the transport planning process. Hence, MCDA is a suitable tool to assess multiple criteria within rail transport planning.

Local governments and community groups place special emphasis today on finding new evaluation methods/approaches for development of transport investments. It is helpful to increase the rail transport competitiveness relative to other types of infrastructure regarding decision-making and prioritizing alternative rail projects.

In Turkey, rail transport planning is divided into authorities at the local, regional and national levels. Istanbul metropolitan city has a complex governance system. There are a large number of departments involved from all levels of government, resulting in overlapping responsibilities. In decision-making processes, all these stakeholders must be involved in a more collaborative way by taking all aspects into consideration.

This research is aimed at determining the best location for rail transit corridors or rail stations based on multiple criteria which should be considered in site-selection decision-making. This paper provides a framework for the application of multi-criteria decision-making (MCDM) methods in rail transport system planning.

The rest of this paper is organized as follows. The next section discusses previous studies in the relevant literature. In Sect. 3 the primary rail transport planning objectives are discussed. Section 4 discusses the study methods and procedure for the cartographic modelling and multi-criteria analysis (CMCA). Section 5 contains a case study applied to four districts (Beylikdüzü, Esenyurt, Avcılar and Başakşehir) in the Istanbul metropolitan area and the results are presented. Finally, the discussion and conclusion are presented in Sect. 6.

2 Literature Review

The literature provides a critical appraisal of previous studies related to geographic information systems (GIS) multiple-criteria decision-making (MCDM) to evaluate transportation systems, as described below.

Economic analysis techniques have been the predominant methodology for the assessment of transport systems. Previous researchers have reviewed these methods from different perspectives such as the role of appraisal methods. Johansson et al. [1] searched for improvement of cost–benefit and transportation system analysis and categorized the appraisal tools as cost–benefit analysis (CBA), multi-criteria analysis, transport system analysis, social impact analysis, environmental impact assessment and location impact assessment. Tudela et al. [2] compared the cost–benefit and multi-criteria analysis and found that the results of the two were different when evaluating non-monetary items and decision-making needed to perform other aspects to achieve better solutions by considering all actors.

With respect to the current state-of-the-art methodologies and techniques, multi-criteria analysis is a well-acknowledged technique for the inclusion of several decision criteria and multiple stakeholder opinions.

In the early stages of the decision process, it is essential to define not only the specific evaluation criteria to be considered in the analysis, but also how variables are obtained and how these decision variables will be assessed. With the development of GIS, GIS-based multi-criteria evaluation techniques are increasingly used according to obtainable data (environmental, social, topographic, geological, etc.). Malczewski [3] reported that GIS packages have the ability to accomplish cartographic modelling and to map algebraic operations like weighed summation. The study highlighted the systems designed by cartographic modelling. There are different ways to solve multi-criteria decision/evaluation problems. Malczewski and Jankowski [4] categorized these methods which can be integrated with GIS. The first category is value function methods, multi-attribute value models, analytical hierarchy/network process (AHP/ANP) and reference point (RP) methods. The second category is outranking relation methods including ELECTRE and PROMETHEE.

It is worth noting that MCDM methods are an important research area in the field of transport, especially for location analyses and the selection of alternatives/scenarios for the development of public transport systems. There are valuable review studies based on the world scientific literature. Yannis et al. [5] presented a progressive review of the literature on MCDM in the transportation industry. The applied MCDM methodologies can differ significantly; for instance, during the decision-making process, the aim, objectives, criteria and indicators must be defined. Broniewicz and Ogrodnik [6] conducted a comprehensive literature review pertaining to multi-criteria methods applied in transport-related projects.

There is a relative lack of up-to-date and comprehensive research papers addressing urban railway corridor planning based on MCDM techniques for the Istanbul metropolitan area. Kırlangıçoğlu [7] studied railway route design processes with GIS and MCDM techniques. The study considered the most important factors affecting the railway corridor planning process and illustrated the spatial suitability levels for current, ongoing and proposed railway lines for Istanbul. This study can be described as the only current and up-to-date source which shares similarities with our research aim, specifically for the Istanbul metropolitan area. In another study conducted specifically in Istanbul, Yücel [8] prioritized different rail system projects using the MCDM methodologies for the evaluation of transportation investments in Istanbul. Istanbul Metropolitan Municipality (IMM) projects are considered as an alternative, which are planned to be constructed. For transportation investment evaluation problems, the analytical hierarchical process (AHP) was employed and alternative weights were discovered using the best–worst method (BWM).

Therefore, we need to examine the public engagement which should be part of the transportation planning process. As noted by Gamper and Turcanu [9], previous studies have not examined the influence of the MCDM methods on public decision-making, and rather have generally focused on the experts and government authorities. In order to fill this knowledge gap regarding public transport users' expectations, willingness and opinions, applications were taken for new railway lines in the Istanbul metropolitan area within the context of our study. The study area was selected according to the results taken from the public applications platform, which gave information on the district and sub-district level. The spatial locations of railway system requests were sorted and prioritized from judgements by the public.

It is worth noting that the set of criteria should be supplementary and essential to a given decision problem. The same decision-making problems can be solved with the use of different sets of criteria. Generally, research authors incorporate the criteria lists frequently used by previous literature studies or from specialists in their field. Ahmed [10] proposed four main criteria (population size, structure of the city, availability of rail transit system and geographical features) in his thesis titled ‘Determination of Potential Transit Station Locations’. Parry et al. [11] selected the variables of slope, altitude, land use and service status for the suitability assessment. Özceylan et al. [12] reviewed site suitability analysis based on GIS-based MCDM modelling from scientific papers which were published between 2005 and 2019 in academic journals. This study takes a step forward by classifying the main criteria according to the frequency of use. Surface water/wetland, distance from urban areas, distance from roads, slope, land use, geology and ground water are the most frequent sub-criteria. Criteria analysis results similarly show that surface water, geology, land use and slope criteria are among the most commonly used criteria groups.

Apart from these applied methods, Ahmed et al. [13] proposed a genetic algorithm-based solution to optimize the locations of stations from the set of candidate stations and the connected line network using GIS. The study analyzed the feasibility of station sites before the optimization process by considering multiple planning requirements that come up from transport sector stakeholders.

In the light of these significant literature reviews, we categorized the scientific papers to learn about past experiences and compare them by study objectives, geographic extent, criteria assessment, shortcomings and study outcomes. With this review, we found that there are only a few up-to-date and comprehensive research papers addressing rail transportation planning based on MCDM techniques, especially for the Istanbul metropolitan area. Additionally, we examined what is already known pertaining to the opportunities of spatial MCDM, which suggests that it is a practical, user-friendly, cost-effective and useable tool, especially for public expectations when carrying out the urban rail transport planning process.

Rail transit suitability was evaluated using the CMCA method for five primary criteria (Geology, Land use, Population Density, Slope and Stream). Our study aims to draw attention to this method’s applicability specifically for urban rail transit planning. Weighted overlay analysis was conducted by calculating the urban rail transit suitability levels as a matrix of cells. Model results are evaluated for the candidate metro line by station buffer analysis. Additionally, public inclination, initiatives, requests regarding urban rail transit, socio-economic conditions and urban functions were taken into consideration. This study underlines public involvement as a neglected research dimension in comparison with expert and government authority involvement. The study supposes that the usage of multi-criteria techniques and multi-criteria analysis should have a statutory obligation in urban rail transit project appraisal stages.

Furthermore, previous studies still have limitations and insufficient spatial coverage, and incompleteness of assessment of criteria and potential applicability for urban rail transit planning. The major contributions of this paper are as follows: (1) This study raises the issue of how to integrate and meet multiple study objectives during rail transit planning processes and discusses how our model is applied, step by step, for urban rail transit development. (2) This study has a novel urban rail transit suitability framework for finding the best solution for the rail transit system planning problem. (3) This research employs a combination of co-production with the urban transport user’s expectations and an emphasis on areas with low socio-economic status points, thus introducing a new paradigm in MCDM.

We hope that our work will contribute to a better understanding of rail transportation planning, and we identified clear and concise research questions. The following research questions enable us to select the appropriate research methodology and design: (a) Which subjects, problems or decision criteria are studied? (b) What are the specific MCDA methods used for urban rail transport project evaluation? (c) What is the appropriate research methodology that best answers the research aims and objectives? (d) Are multiple actors or stakeholders included in the planning process?

To sum up, this study comprehensively explores the systematic assessment of the innovative potential of the MCDM framework in order to obtain ‘a structured, easily applicable, multi-dimensional, collaborative, reliable, productive and extensible’ model structure, and the study effectively demonstrates the adopted methodology for the optimum and suitable urban rail transit route/station location selection.

3 Rail Transport Planning Study Objectives

Successful rail transport planning depends on the selection of primary objectives and components for finding the optimum solution based on multiple factors and constraints. In such a planning process, it is important to evaluate distinguishing characteristics of rail transport planning. Thus, prominent objectives are briefly described as follows:

3.1 Increasing Public Awareness and Public Involvement in the Planning Process

As outlined in an article by Tischler [14], transport projects should be assessed by new appraisal methods, and transport users should be involved in project planning and design development phases to expand public awareness of social and environmental impacts of mega-transport infrastructure projects. Gamper [9] noted that MCA tools have the ability to address interdisciplinary issues and strengthen the decision-making process. MCA-assisted applications will continue to promote attention in public decision-making.

As a different perspective for transport decision-making, Giuffrida et al. [15] defined three ingredients for performance- and consensus-based decisions and proposed a Public Participatory GIS-MCDA-based framework for the evaluation of transport projects. This framework has the potential to promote good social acceptability and robustness of decisions. It supports the evaluation of the alternatives for both technical performance and degree of consensus. Subsequently, Quick [16] focused on ‘public participation in transportation planning’. This approach exhibits significant advantages to the public by giving information, getting involved in the process of identifying priorities, and determining decision-making parameters in a collaborative manner. Transportation policy stakeholders comprise the general public who have different transportation needs based on their geographic locations with specific interests. Participation by the public can lead to a more equitable distribution of scarce public resources.

It is essential to provide a comprehensive planning approach for all community participants and focus groups of rail transport projects. In order to gain a greater understanding of public needs for a new railway line, public transport users’ expectations, willingness and opinions for new railway lines were taken from the Istanbul Metropolitan Municipality e-services application platform from January to March 2021. The ‘Beyaz Masa e-service’ information was applied for 230 public transport users associated with rail transport infrastructure, and then these values were grouped and structured. The application form includes application number, date, personal contact information, entity, subject, district, application details and further explanations. Many public transport users request accessible and transferable new railway lines. Walking distance and time are crucial factors that individuals consider when deciding whether or not to use public transportation. The distance to the railway station is the most concerning factor for passengers. They expected to be close to the railway station catchment area. Public transport users are interested in up-to-date, reliable and accurate information about ongoing and planned projects of railway lines. Railway line expectations were substantially taken from Beylikdüzü, Esenyurt, Sultangazi, Başakşehir and Beykoz districts. The first objective of our study was to provide a rail line that satisfies the public’s transport needs and meets the travel demand with a sustainable living environment. The decision modelling framework in this study can be distinguished from other related studies by its consideration of public users’ expectations in the railway transport planning process.

Modelling travel demand is a challenging task that is required for urban transport planning and evaluation of transportation systems. Istanbul’s Integrated Urban Transportation Master Plan (IUAP), approved in 2011 by the IMM Transport Coordination Center, is fully aligned with the provisions of the Environmental Order Plan of 2009 (Istanbul Transport Annual Report) [17]. The IUAP plan studies continued with several updates, but it has not yet been officially approved. It is vital to accurately predict future passenger demand from model results derived from the urban transport modelling system. Public transport demand forecasts are not shared as an authentic technical published and open-source document, and consequently these research findings are excluded from our study.

3.2 Serving Disadvantaged Districts

The second objective of our study was to provide a rail line that serves vulnerable groups and economically disadvantaged areas. The directorate of Earthquake and Ground Research of Istanbul Metropolitan Municipality [18] carried out a social vulnerability study in Istanbul, which is based on 40,000 household surveys covering the entire jurisdiction of the metropolitan municipality. The study covered all 955 sub-districts with residential occupation, and a social vulnerability score was calculated for each household, sub-district and district. The research is composed of three phases. The first phase includes the literature review, determinations of relevant indicators to evaluate social vulnerability level, production of household survey to provide data to indicators, and validation of a household survey through pilot survey.

The selected indicators were socio-demographic characteristics, urban belonging, socio-economic status, access to health services, social solidarity, risk perception and actions taken to reduce risk and values. In the second step, a validated household survey was applied at a city scale with face-to-face interviews. In the third step, survey results were analyzed and statistical tests were applied on indicators in order to test the representation quality of the indicator themes. As a result, indicators of socio-demographic characteristics, socio-economic status, risk perception/actions and values were validated for analysis. According to these results, socio-economic status values were used for evaluating social criteria, as shown in Fig. 1. Socio-economic status points are taken from ‘Understanding Social Vulnerability Against Disasters Survey Results’ [18] which were calculated according to education, income and property ownership for households. These values are covered for all sub-districts in Istanbul. Socio-economic status points are sorted from low to high values. Rail transport infrastructure serves the public’s needs much more in urban areas where the socio-economic point is low. In our study, study area selection was conducted by taking into consideration these low point areas.

Fig. 1
figure 1

Socio-economic status values (illustrated by author)

Full size image

There are four classifications of urban function and urban density according to the Ministry of Industry and Technology’s ‘Measuring Urbanization Level in Turkish Districts by Population Density and Urban Functions’ [19] national study report in Turkey.

Category-1 refers to ‘high urban function, high urban density’; category-2 refers to ‘high urban function, low urban density’; category-3 refers to ‘low urban function, low urban density’; and category-4 refers to ‘low urban function, high urban density’. Urban function provides an opportunity to examine not only the urban agglomerations in the cities, but also the intensity of socio-economic activities. Studies on socio-economic characteristics of urban settlements also constitute a basis for evaluation of urban functions. This national study report is crucial to understanding the region’s urbanization patterns. Arnavutköy, Bağcılar, Esenler, Esenyurt, Sancaktepe, Sultanbeyli and Sultangazi districts are placed in the fourth category, which can help in searching for the areas that need improvement in terms of public transport.

3.3 Exclusion of Forest Areas

The third objective of our study was to provide a study area that excluded forest areas from selection sets in order to reduce negative environmental impacts for the Istanbul metropolitan region.

In this context, findings by Orhon et al. [20] contain valuable information about the mega project’s environmental impacts for Istanbul. The reduction of forest area was found using the changes in land classification acquired by Landsat data for Istanbul. The changes in land cover conversions in Istanbul were examined from 2009 to 2016, where the urban and built-up classes evolved with a growth rate of 35.6%. Agricultural land and green habitat were affected by the growth of built-up areas. Forest areas were destroyed at an approximate rate of 15,000 ha in 7 years (2009–2016), corresponding to a 7% decrease in the total forest area.

In our study approach, forest areas in the northern region of Istanbul should be protected to reduce the environmental impact of railway project-related urbanization of the surrounding area. This study highlights that continuous population growth and build-up of areas in northern Istanbul should be restricted. In this study, the total area of forest land on the European side was 1296.77 km2 and on the Asian side was 970.24 km2. They were excluded from the study area and are referred to as ‘unsuitable areas’ for new urban rail transport projects. The term ‘unsuitability’ excludes ‘intercity’, ‘interregional’ and ‘international’ transport projects.

3.4 Primary Land-Use Zones

The fourth objective of our study was to provide primary land-use zones which the new railway line and stations would initially serve. This parameter is used to place the proposed route and stations in an area with a high concentration of high-density housing areas, schools, hospitals and commercial land-use zones. Preferable zones were assigned high scores; the other zones like farming areas and military zones received low scores when reclassifying map layers.

According to Pacheco-Raguz [21], changes in transport infrastructure will cause variations in urban land use based on the relative attractiveness of certain areas. Areas that are more accessible may have a higher interaction potential, as they may provide a comparative advantage.

Land use criteria are a significant aspect of rail transit planning when evaluating the social implications of a project—for instance, the level of access to schools and hospitals, and impacts on disadvantaged groups, people with disabilities, pedestrians, children and elderly people. Ahmed [10] supported this criteria by mentioning that travel patterns, land use development, and economic, social and environmental characteristics interact with developing a rail transit network.

3.5 Geological Sustainability

The fifth objective of our research was to evaluate geological sustainability in the identification of suitable railway corridors. Geological layers contain information about the lithology. The geological characteristics of a region influence the availability, type and quality of material, according to Karlsson et al. [22].

Lithology types are classified by score values as follows: Qal = 1, Yd = 1, Tdg = 2, Tik = 4, Tçç = 4, Tçg = 4, Tds = 5, Tc = 6, Tdç = 6, Ct = 7, Tsğ = 7, Tçb = 8. The Istanbul Geological Map (1:100,000) jpeg file with the correct coordinate system information was obtained from the Directorate of Earthquake and Geotechnical Investigation [23]. For creating a thematic geological raster map, image classification techniques were tested as ‘supervised’ and ‘unsupervised’ classifications using ArcGIS 10.8 software. Because of the low resolution of the input data file and heterogeneous areas, classification accuracy was not sufficient for lithological unit segmentation. For this reason, lithological segmentation/geological formation database creation processes were carried out by drawing polygon lines from the boundary lines manually. After creating and editing the geo_polygon shape file, the type and quality score of lithological formation data were entered, and then the polygon data file was converted into raster format (see Fig. 2).

Fig. 2
figure 2

Geological map of the study area (illustrated by author)

Full size image

4 Methods and Procedure

This section summarizes the research method for finding the best solution for the rail transit system planning problem by utilizing a novel urban rail transit suitability framework. This framework performs the aggregation of relevant selection criteria in a spatially distinctive way. The step-by-step flow chart of the methodology adopted is presented in Fig. 3.

Fig. 3
figure 3

Flow chart of methodology

Full size image

4.1 Determining the Criteria Weights Based on Responses from Key Stakeholders

Decision-makers were involved in the weighting of criteria in order to add value from their knowledge gained by different specialities. Special academic staff and specialists in the associated agencies were selected (Republic of Turkey Ministry of Transport and Infrastructure, Head Department of Design and Construction of Urban Rail Transport Systems, Istanbul Metropolitan Municipality Railway Project Directorate, Department of Transport Agencies, academic professionals)

AHP was applied to estimate the criteria weights. In this respect, Broniewicz [6] stated that the AHP method was most frequently used with regard to MCDM methods for transport-related decision-making issues. In the next step, the relationship between the five main criteria was tested with the AHP. A survey was conducted to collect comparative judgement and pairwise comparison between the five main criteria. A sample size of 10 was selected for the experts to complete the designed questionnaire. After receiving the results of the respondents, the criteria were arranged and averaged using an AHP Excel template. All judgement matrices were checked for consistency using the consistency ratio. Two survey forms were eliminated because the ratios were inconvenient. According to the AHP results, weights were allocated in the following order: Population density 38% > Stream 19% > Geology 15% > Land use 14% = Slope 14%.

4.2 Survey Analysis/Expert Questionnaire

The survey form was designed and completed by experts (specialists in the field of rail transport engineering) to gather their opinions.

The experts were asked to assess the importance of each criterion on a nine-point Saaty scale to give the relative rating of two criteria. They were asked to indicate which criterion was more significant for the optimal location of the railway line/station area. Moreover, the survey form was divided into four parts. The first section comprised five main criteria in a pairwise comparison matrix (A-Land topology, slope (%), B-Land geology, lithological formation, C-Stream density, D-Population density, E-Land use).

Section 2 comprised seven main criteria and sub-criteria in a pairwise comparison matrix:

A-Railway length and station number (A1 Railway length and station area coverage, A2 Station number and distance between stations, A3 Proximity to other railway lines)

B-Socio-economic status (B1 Zone/District population, B2-Income, B3-Education, B4-Fulfilment of future public transport demand especially for home to work, home to school trips)

C-Land use (C1 Proximity to industrial areas, C2 Proximity to educational area, educational catchment area, C3 Proximity to health service areas)

D-Environment (D1 Environmental impact to agriculture and forest areas, D2 Environmental impact to natural protected areas, D3 Proximity to flood plain areas, D4-Proximity to streams D5-Proximity to basin, watershed threshold)

E-Transport infrastructure (E1 Proximity to airports, E2 Proximity to intercity bus terminal, E3 Proximity to high-speed train terminal)

F-Economy (F1 Geological suitability, F2 Topographical suitability, F3 Land expropriation suitability, F4 rail construction engineering technique)

G-Integration and accessibility (G1 Proximity to transfer centers, G2 Park and ride applicability to station surrounding areas, G3 Travel time reduction, G4 Walking distance to railway station, G5 Operating fare (urban-high, suburban-low), G6 Integration to other railway projects, G7 Integration to sea transport terminals)

In Section 3, experts were asked to give their opinions about main and sub-criteria and whether there were criteria not taken into consideration. They were also asked whether they had any suggestions about evaluation criteria according to the railway route/station location selection problem.

Experts were asked whether, when the government was preparing the investment program for infrastructure projects, there were any regulations for using multi-criteria techniques or any pilot projects to implement such techniques in the project evaluation process for future projects. The survey with governmental organization managers indicated that MCDA techniques had not been a part of any planning mechanism in their organizations in Turkey. Governmental institutions noted that MCDM methods were not used for any current or future transportation projects.

In Section 4, transport agencies were asked to provide information about whether there were completed, ongoing or future projects incorporating public participation into the Mega Infrastructure Project’s approval process.

4.3 Construct Map Layers for Decision Criteria Using ArcGIS Analytical Tools, Spatial Analysis Tools–Overlay Analysis

The AHP technique in combination with GIS overlaying was used to assign weights for each parameter. In this study, the criteria maps were constructed by editing geographical attributes for five main criteria to overlay the weighted map layers. All spatial data layers were registered to the UTM (Universal Transverse Mercator) projection system.

Additionally, a ‘stream model’ was generated from the digital elevation map (DEM). Using ArcGIS Arc Hydro tools, a geographically referenced database containing geographic and hydrographic data including basin stream networks was created for the study area (see Fig. 4), and the Stream Network Delineation ModelBuilder Flow is shown in Appendix A.1 in order to demonstrate the modelling process.

Fig. 4
figure 4

Stream network (illustrated by author)

Full size image

Geographical features of the evaluation area for planning a rail transit system are vital to taking an investment decision. For instance, hilly terrain, narrow valleys and bodies of water increase the total project cost. The ‘slope model’ was generated from the digital elevation map (DEM). Slope values were categorized as 0–5% (1), 6–10% (2) and > 10% (3). The slope of the terrain was considered for the railway route selection process, as it directly influences the construction and operating costs. Digital topographical maps [24] were used to create TIN and DEM and derive layers such as slope and stream delineation.

4.4 Using Overlay Techniques in Multiple-Criteria Decision-Making

We use one of the most common additive methods, weighted linear combination (WLC), in this study. This method can be implemented with the help of a GIS system with overlay capabilities. Malczewski [25] used overlay techniques to aggregate attribute map layers (input maps) to produce a composite map layer (output map).

In this study, the weights are not directly assigned or obtained by directly asking the decision-maker. We conducted a criteria assessment survey to predict the weights by the AHP, an MCDM process which is based upon the pairwise comparison method (see Sects. 3.1 and 3.2).

Suitability scores were assigned for categories that we determined for five input variables: Population, Land use, Slope, Geology and Stream. For instance, the Slope variable has three categories: low (0–5%); medium (6–10%); high (> 10%) (see Table 1). The slope map layer was reclassified by these categories in a raster map format in a GIS environment. Raster reclassification was done for five thematic map layers, and then we created the weighted overlay map.

Table 1 Weighted indexing table
Full size table

Suitability scores of five evaluation factors were multiplied by the criterion weights and then summed to get a total suitability score in the GIS environment, which can be seen in Eq. (1).

$$S=\sum_{i=1}^{n}wi*fi$$
(1)

where S = suitability score, Wi = weight of the ith evaluation factor, fi = suitability score of the ith evaluation factor, n = total number of evaluation factors.

Weighted layers were integrated into the GIS field calculator as shown in Eq. (2).

$$\mathrm{WLC}= ([\mathrm{POP }] *0.38+[\mathrm{STREAM }] *0.19+ [\mathrm{GEO }] *0.15+[\mathrm{LAND }] *0.14+[\mathrm{SLOPE }] *0.14$$
(2)

The weighted overlay creation workflow is composed of four processes: A—Prepare the input dataset (convert any input vector datasets into raster datasets); B—Classify the raster dataset (raster reclassification); C—Create the weighted overlay map; and D—Station buffer analysis. Different geographically referenced layers can be overlapped with each other according to the study’s objective.

Input raster layers are assigned by influence weight value (%) according to predetermined AHP weights. The total influence weight value should be 100%. A weighted indexing table was created to find the most suitable location for eight railway stations using the five input parameters, as shown in Table 1. The survey of decision-makers was used to obtain the influence weights for overlaying map layers.

5 Study Area Description and Research Findings

5.1 Study Area

Data for September 2021 showed that the Istanbul public transport system accommodated 11,243,226 passengers daily (average weekday). Public transport distribution in average weekday daily percentage statistics were taken from the Istanbul Transport and Mobility Report [26]. The percentage of trips by transport with rubber-tired vehicles like buses, minibuses, metro-bus, service-bus and taxi systems was 79%; the percentage of trips by rail transport (metro and light metro systems) was 19%, and the percentage of sea transport was 2%.

Istanbul, with a surface area of 5340 km2, is the most populous city in Turkey and one of the mega cities in Europe. The study area covered 227.23 km2 (see Fig. 5).

Fig. 5
figure 5

Study area (illustrated by author)

Full size image

The study covered four districts (Beylikdüzü, Esenyurt, Avcılar and Başakşehir) and 73 sub-districts. Table 2 shows the population, average income, average household size, average years of education, socio-demographic indicator point, socio-economic indicator point and average socio-economic status point values for Beylikdüzü, Esenyurt, Avcılar and Başakşehir districts.

Table 2 Study area population, socio-economic and socio-demographic characteristics
Full size table

This study region was selected to resolve existing bottlenecks in terms of increasing efficiency, competitiveness and economic growth. Additionally, it covers environmental and social effects. For instance, forest lands and floodplains were excluded for an environmental point of view. For social effects, rankless districts and sub-districts had precedence according to socio-economic status values.

The selected study area combines several potentially conflicting objectives such as quality-focused, long-term-oriented and more transport-balanced planning approaches. In our study, the study area selection and overview were done during the first stage by taking these issues into consideration.

5.2 Results

The final map of overlay modelling (cartographic modelling) as a method for rail transit suitability is depicted in Fig. 6. Based on the results of the weighted overlay analysis map, suitability was classified as low, medium, or high for the 227.23 km2 study area. The flow chart of the methodology was given in Sect. 4 and the cartographic modelling ModelBuilder flow is shown in Appendix A.2 in order to demonstrate the overall modelling process.

Fig. 6
figure 6

Weighted overlay analysis map (illustrated by author)

Full size image

The survey results are a resource for obtaining criteria weights for overlaying map layers. AHP was applied to estimate the criteria weights. The experts were asked to assess the importance of each criterion. According to expert opinion results, the weights were obtained as follows: Population 38% > Stream 19% > Geology 15% > Land use 14% = Slope 14%. The GIS-based multi-criteria analysis was carried out using WLC. This method has enough flexibility to generate a complete range of decision support maps by changing the weights. Combined maps produce a final decision map depicting the suitability levels.

According to the final rail transit suitability results, the grid cells obtained with the study area ranking have values from 1 to 7, reflecting low-ranking to high-ranking classes. According to this ranking, the following count values were obtained: Class 1: 186 grid cells, Class 2: 3850 grid cells, Class 3: 4913 grid cells, Class 4: 6054 grid cells, Class 5: 3009 grid cells, Class 6: 1170 grid cells, Class 7: 75 grid cells. The most promising areas are between the fifth and seventh classes. A Class 4 ranking represents the overall study area suitability level (see Fig. 7).

Fig. 7
figure 7

Suitability classes (S1-S7)

Full size image

The final weighted overlay map was used for buffer analysis to evaluate the site suitability of the eight rail stations (1000 m buffer zone). The process involves generating a buffer around existing station point features and clipping buffer areas with weighted overlay polygon features by geoprocessing tools (see Fig. 8). Coverage areas are calculated according to input data (grid code classification). If the grid code value is high, the suitability ranking is high; otherwise the suitability ranking is low. According to the results, in Station 1, the maximum area is covered for gridcode 4. For Station 2, the maximum area is covered for gridcode 3. For Station 3, the maximum area is covered for gridcode 4. For Station 4, the maximum area is covered for gridcode 4. For Station 5, the maximum area is covered for gridcode 4. For Station 6, the maximum area is covered for gridcode 3 and gridcode 4 (area values are near each other). For Station 7, the maximum area is covered for gridcode 3. Lastly, for Station 8, the maximum area is covered for gridcode 4. All stations generally take grid code 3 and 4 values, which indicates a medium suitability level.

Fig. 8
figure 8

1000 m station buffer analysis (illustrated by author)

Full size image

As depicted in Fig. 9, the most suitable site is Station 8, because this station receives a maximum grid value of 7, which covers 30,000 m2 for its buffer area in the Beylikdüzü district.

Fig. 9
figure 9

Area (m2) and grid code values for buffer zones (1000 m)

Full size image

Rail transport infrastructure was evaluated as a section of a candidate metro line M 18 (Başakşehir-Esenyurt-Beylikdüzü) with eight railway stations for the preliminary stage evaluation. According to our study area, station locations are generally positioned where the transport systems are relatively less developed than the urban activities. Existing, ongoing railway lines and metro-feeder bus lines are depicted in Fig. 10.

Fig. 10
figure 10

Railway lines (existing and ongoing) and feeder bus lines (illustrated by author)

Full size image

Station 8 (Beylikdüzü) is planned to be an important transfer hub in the future. Station 1 (beginning station) and Station 2 will be integrated with the Bağcılar-Olimpiyatköyü existing metro line, Station 2 will be integrated with the Halkalı-New airport metro line project, and Station 5 will be integrated with the Mahmutbey-Bahçeşehir-Esenyurt metro line project. Station 8 (end station) will be integrated with the Sefaköy-Beylikdüzü metro line project.

6 Discussion and Conclusion

In the first stage of this research, the emphasis was on understanding and discussing how we could reach an idealized planning process, find the optimum solution, identify the research methodology, and build a study framework with collaboration and coordination between stakeholders.

The implemented study framework can be utilized to assist decision-makers and practitioners in the design of efficient participation processes, as well as dealing with various stakeholders and complex issues related to transportation decisions. The CMCA was implemented for five main criteria (Geology, Land use, Population density, Slope and Stream) to evaluate rail transit suitability. A rail transit suitability map was obtained from both a technical and social perspective with the aid of GIS–MCDA techniques.

This framework is applicable for planning a new rail transit system and expanding an existing rail transit system. In our study, we proposed a candidate railway line with eight railway stations and implemented this model. The model structure enables the automation of GIS-based MCDA to find the rail transit suitability levels within the ArcGIS ModelBuilder 10.8 application by performing the following sequence of steps:

  • Step 1 Preparing thematic maps: Criteria maps are constructed by editing geographical attributes for five main criteria.

  • Step 2 Standardization of maps: Raster maps are reclassified by categories which are described in Table 1 in the previous section (category and score columns).

  • Step 3 Overlaying maps: Weighted overlay analysis is performed with the determination of weights using a pair-wise comparison (AHP) based on the opinions of experts. Input raster layers re assigned by influence weight value (%) according to predetermined AHP weights.

  • Step 4 Final suitability map: A weighted overlay analysis map is created to evaluate route and station site suitability as a preliminary analysis for transforming and developing the transport network.

  • Step 5 Model implementation of candidate metro line M 18 (Başakşehir-Esenyurt-Beylikdüzü) with eight railway stations for the preliminary stage evaluation.

In summary, the objective of the study was to investigate the quality and the practical worth of the multi-criteria-based modelling technique by implementing this technique for rail transport planning. The study aimed to combine several potentially conflicting objectives including a quality-focused, long-term-oriented and more transport-balanced planning approach.

The innovative aspect of our approach is the use of public passenger opinions/preferences with overlay analysis (a cartographic model) and the AHP method to capture the perspectives of multiple stakeholders for an efficient and effective participation process in rail transport planning. This study will close the gap by utilizing a novel suitability framework in which, for the first time, public transport users’ expectations and requests along with community needs specific to socio-economic conditions and urban functions are taken into considered simultaneously. A pairwise comparison was conducted to obtain criteria weights and overlay analysis (cartographic model), which were systematically incorporated, and a final suitability map was then obtained by applying GIS-MCDA techniques. This study underlines public involvement as an unattended gap in contrast to expert and governmental authority involvement. Public participation should be a necessary condition for project appraisal and approval stages. Public passenger opinions and preferences can change dynamically as time progresses. From this point of view, a database should be built, monitored and reported periodically by governmental agencies and municipalities.

A transparent, easy-to-understand and equally accessible open-source web-based GIS is essential as a means to convey data in particular for non-expert local communities, agencies and associations without the need for proprietary software, and to collect feedback from all public transport users. People can easily participate in surveys and can be informed of meetings, workshops, etc. Technical guides and legal government documents are required to increase applicability in specific application areas (road, rail, air, freight transport), criteria and sub-criteria, methods, processes and stakeholders. This paper draws attention to whether MCA techniques have legitimacy on a governmental level, and if not, why they have not attained the same level of legitimacy as other decision-support tools such as CBA or cost-effectiveness analysis (CEA). In light of the benefits of MCDM methods, we suggest incorporating these methods into governmental action plans to enhance decision-making and policy planning efforts in the near future.

The topic of multi-criteria decision support systems and tools is a growing research area and offers practicable and extensive methods. Further research can be done by identifying and evaluating additional and goal-driven assessment criteria according to new goals and requirements. New methods in spatial systems for MCDA can be attempted by structuring and developing model implementation stages. This study framework is adaptable for all kinds of urban rail transit systems to fulfil the need for well-accepted alternatives and scenarios. Not only new rail transport route selection, but also pre-determined railway lines can be analyzed depending on different circumstances. Every project alternative is unique and has a spatial surrounding. It is obvious that the evaluation process cannot be the same. Additional factors can be implemented to increase the benefit to society.

As a result, research findings meet the vision of the city transport plan and provide an innovative planning framework by focusing on the public’s needs. Additionally, this modelling methodology offers a practical approach by visualizing the level of spatial suitability for an integrated, advanced, sustainable urban transport development by providing a roadmap for decision-makers.