Digital traces: Mapping Bogotá’s unmapped transit network using smartphones and networked databases

Abstract

New tools have enabled “civic hackers” and transportation researchers to map previously uncharted transit networks previously confined to the purview of locals and insiders. These new datasets reveal the extent of these systems and their role in providing access to the city. In this paper, we describe the methodology regarding the mapping process of Bogotá’s semi-formal SITP Provisional bus system, accomplished using smartphones, cloud-based data management systems, and GIS software. By visualizing the semi-formal SITP Provisional system alongside the centrally planned bus system, we provide the first complete picture of Bogotá’s transit system. We also develop two types of data analysis based on General Transit Feed Specification (GTFS) data generated from this mapping process. Using spatial algorithms, we identify parallelism between the semi-formal transit routes and the formal transit network. We then visualize the degree of access to job opportunities that each system provides. We find that integrating semi-formal and formal transit services, that is, the entire network, increases accessibility levels for workers, especially at urban peripheries. Results suggest the importance of considering semi-formal transit services in transportation planning, the services often neglected in the planning process, and the advantages of integrating them into the network to increase accessibility to opportunity areas. We recommend that other cities harness GPS-enabled apps to map transit systems, generate GTFS data, and empower local actors to make use of the data. Based on this bottom-up approach, semi-formal transit networks can provide additional inputs for urban transportation planning processes regarding the transportation user´s accessibility needs.

Keywords

mapping urban transport informal transit information communication technologies Latin America

Introduction

New tools, specifically smartphones and networked databases that allow large volumes of data to be shared and edited, have enabled citizens to go out into their cities and document whatever phenomena interest them. Platforms such as Open Street Maps have transformed the relationship between citizens and their cities. Instead of relying solely on data and maps provided by governments or other professional sources—maps often with incomplete information which exclude entire populations—these tools have given rise to civic mapping and new forms of engagement (Jones et al., 2015). Those in the field of transportation studies have begun to harness the power of these tools to map traditionally unmapped transit systems formerly solely the purview of locals and insiders who know how to navigate them. Between 2012 and 2013, a group of researchers and students from the University of Nairobi, Columbia University, and the Massachusetts Institute of Technology (MIT), showed how simple tools, elaborate data processing, and diligent mapping via smartphones and handheld GPS devices could produce a legible map of Nairobi’s privately owned or cooperatively owned minibuses (a network known in the plural as matatus), thereby informing conversations around transport planning and infrastructure investment in Nairobi (Williams et al., 2015).

Taking our lead from these researchers, their Digital Matatus methodology and the United Nations’ Sustainable Development Goal 11, which is intended to increase access to safe and reliable transport by 2030, we set out to map Bogotá, Colombia’s entire transit network (UN, 2020). Bogotá’s transit network consists of 1) a centrally planned network of buses privately operated under the name Sistema de Transporte Público de Bogotá (SITP); 2) the SITP’s Bus Rapid Transit (BRT) service, Transmilenio, which has one of the highest ridership levels globally (BRTDATA.ORG, 2021); and, 3) a decentralized network of 4920 buses and vans, known as the SITP Provisional. The SITP formal transit network provides 11.1% of trips, the BRT provides 18% of the trips, and the Provisional network provides about 6.7% of the total trips per day in Bogotá (SDM, 2020b). The SITP provisional system users pay their fares in cash and stops are not fixed (Guzman et al., 2018). Bogotá’s municipal government implemented policy measures to gradually absorb this service into the formally planned conventional bus network (SDM, 2020a, 2021). The SITP Provisional service usually starts in the city’s periphery and provides access to low-income groups located in informal settlements.

Mapping Bogotá’s SITP Provisional service presented us with several challenges unique to Bogotá but similar in some ways to those encountered by other researchers in documenting Nairobi’s matatus, Accra’s trotros, Mexico City’s microbuses, as well as the minibus taxis of Kampala and Cape Town. Bogotá’s SITP Provisional minibuses do not have a centralized transport hub from which trips originate or at which they terminate, unlike the networks found in African cities. In this case, the origin points of the SITP Provisional buses are spread out along Bogotá’s urban periphery at multiple depots (patios) that were difficult to identify at first. For instance, we had to take trips from the city center to these depots to determine the location of the terminal points. Moreover, mapping these routes required our mappers, group of students working as data collectors, to take trips to the end of the route, which left mappers in desolate areas of the city; they also were required to input points on their smartphones in plain sight of other passengers. This implies a challenge in terms of safety for data collectors given that they arrived at unknown areas of the city in places that they were unfamiliar with the local context. Considering these challenges, specifically the decentralization of terminal points, we developed a few original techniques to identify Provisional routes. In this context, we sought to answer the following research questions: (1) How can we adapt previous data collection methods to Bogotá? (2) How does the GIS data of our mapped locations enable us to better understand the relationship between the unmapped systems and their more legible counterparts?

In this paper, we describe how we adapted the Nairobi Digital Matatus methodology to Bogotá by collecting data on all of Bogotá’s transit routes. This accomplishment required the performance of two distinct tasks. First, we gained access to the route, stop, and schedule data for the formal SITP (the centrally planned bus network). Second, we went out into the field and manually mapped the SITP Provisional. After mapping the Provisional system, we converted our data to the widely used GTFS format, a specification that allows the input of information on the agency providing the data, the location of transit stops, and the schedule of service (Eros et al., 2014). We did this to obtain comparable data across systems, thereby enabling us to perform original analyses in which we compared the two systems and also stitched together a singular view of all of the Bogotá transit options in (Google, 2020; GTFS, 2020). We developed an approach based on the use of low-cost data collection methods, revealed the previously unmapped network of conventional buses in a systematic manner used multiple platforms and software, and generated open data sources for data analysis, an approach that could be replicated in other contexts.

The paper is structured in four parts. We review the existing and growing literature on mapping transit networks. We then describe our methodology in terms of our data collection methods, how we standardized our field data to conform to the GTFS protocol, then analyzed the data feed using GIS software. Finally, we include a discussion section regarding our findings, and we conclude the paper with insights regarding this type of data collection, opportunities for replication, and recommendations.

Literature review

Information and Communication Technologies ICT, open data and mapping semiformal transit

As smartphones have become more ubiquitous and powerful over the last decade, researchers have begun to use them to map unmapped transit systems across the world. Each of these cases offered us important insights into how to obtain, manage, and process data. The use of experimental data collection techniques such as flocksourcing, which maps transit service processes by merging crowdsourcing and outsourcing services. This approach was successfully implemented in Dhaka, Bangladesh, enabling researchers and local practitioners to generate databases with the potential of providing transit network maps with route level information (Ching et al., 2013). Nairobi, Kenya’s matatus have spawned papers detailing their pioneering mapping techniques in which they used information and communication technologies (ICT), including innovative approaches such as the generation of GTFS (Klopp et al., 2014; Klopp, Williams et al., 2015; Williams et al., 2015).

In 2016, civic hackers supported by the Mexico City government mapped paratransit (informal transit) services known as “Peseros,” which are informal transit services provided by local operators in small buses (OECD, 2017). Researchers determined that the use of open data with real-time transit information can support transit reforms intended to combine formal and informal systems (Zegras et al., 2015a). Rather than relying on a small group of data collectors, the Mexico City Mapaton incentivized more than 4000 transit users via a gamified smartphone application to help map 1500 routes of total length 50,000 km (OECD, 2017). (Sandoval et al., 2017) found that gamification, that is, the prospect of prizes, increased participation and transformed the government’s relationship with its citizens through the use of open data services, in this case, sharing information regarding local trips to inform policy and planning processes.

Mapping projects’ data, when spatially analyzed, reveals key transportation nodes, the extent of service, and how people move around their city. Throughout the African continent, cities (e.g., Kampala, Uganda; Accra, Ghana; Cape Town, South Africa; and Nairobi, Kenya) have used mapping to visualize their dominant transit networks as well as depict service distributions and demand for services (Coetzee et al., 2018; Ndibatya et al., 2016; Saddier et al., 2016). The data from urban transport mapping projects is important to planning the location of social services (e.g., public health facilities in Nairobi) and because analysts can use it to quantify who can geographically access these facilities and whether transportation network investments may improve access to key facilities. Researchers use the results of urban transport mapping analyses to highlight the importance of spatial equity issues in efforts to improve accessibility for low and middle income groups by integrating informal transit services with land use development (Campbell et al., 2019). Klopp and Cavoli (2019) who compared Nairobi and Maputo, highlight the importance of collaborative mapping exercises to inform transportation planning from an inclusive approach (Klopp and Cavoli, 2019). This bottom-up approach provides the bases for the integration of multiple transportation modes by considering the informal transport users’ needs.

Given the high level of informal transit services in Latin America, several cities from the region have also developed mapping projects to understand the spatial distribution of routes, how transport services have adapted to demand, especially along the urban periphery, and the existence of these services in response to user’s need in the region. A mapping project conducted in the city of Managua, Nicaragua, with the support from the government, developed a transit app for data collection purposes, and the data collected was included in already operational apps that inform transit users regarding routes in the urban area (DATUM, 2022). As mentioned before, Mexico City conducted the “Mapaton” project, one of the pioneer mapping projects in the region, which collected data based on a crowdsourcing data collection process in which citizens shared the georeferenced information of their daily trips in the city using smartphones (ITDP, 2016). The city of Bogotá (Colombia) developed another “Mapaton” project (Goldwyn and Vergel-Tovar, 2018), which provided the data used in the present paper for further analysis. The city of Cochabamba (Bolivia) identified transit routes through a mapping project providing inputs later used by Google Maps and transit apps, with similar exercises of open data generation in cities such as Farroupilha in Brazil, Mérida and Veracruz in Mexico, and Cuenca in Ecuador (DATUM, 2022). In fact, the mapping project in the city of Santiago de los Caballeros in República Dominicana set in motion the generation of the DATUM Web site that provides a summary of the mapping projects in Latin America and the Caribbean LAC region (DATUM, 2022). The development of the mapping project in Santiago de los Caballeros received support from the World Resources Institute WRI, the Inter-American Development Bank IADB, Columbia University, and the Massachusetts Institute of Technology (MIT), with the participation of local partners, generating methodological guidelines, open data sources in the GTFS format (IADB, 2019). More recently, the city of Cartagena in Colombia also carried out a mapping project replicating the experience of other Latin American cities, generating open data that enables the visualization of formal and informal transit routes (Vergel-Tovar et al., 2021).

In the LAC region, policy makers are interested in understanding how informal transit services respond to local users’ needs and the challenges of these services impose on the performance of formal transit systems such as Bus Rapid Transit (BRT) services. IADB (2020) examined the role of semiformal transit services, the challenges for transport reforms in light of the permanence of informal transit services, and the relevance of mapping projects for the generation of open data that provide the basis of a further understanding of informal transit supply in cities of the LAC region.

Given that rapid growth of mapping projects and the creation of open data, the next step consists of data analysis, especially in terms of accessibility based on the coverage of the informal and semiformal transit networks. This important step provides the ground to inform policymaking and decision-making processes. Considering that accessibility analysis requires the training of local staff in cities with a high level of informality in their transit services, it is important to continue conducting this type of data analysis approaches based on the successful generation of open data in regions such as Africa and Latin America and the Caribbean. In order to develop this type of spatial data analysis, it is important to consider the key role data in the GTFS format provides.

Urban transport, GTFS format, and data analysis

GTFS is becoming the standard language for transit data worldwide as public transport agencies, researchers, and planning agencies seek information on route location, frequencies, and passenger demand (Kujala et al., 2018). The GTFS format provides route locations and trajectories that facilitate the development of multiple types of network analysis that support the spatial visualization of transit supply in relation to the movement of passengers in the system (Tao et al., 2014).

The GTFS format provides greater flexibility in understanding the relationship between transportation networks and cities. This format allows researchers to look at different scales to examine broad measures, such as job access across a region, and more refined, local impacts such as access to hospitals along a corridor. For instance, GTFS is a data format that allows researchers to conduct different types of analysis, including the development of gravity models (Hansen, 1959). The use of data in GTFS format has focused primarily in looking at accessibility to job opportunities in the literature. The study looking at job accessibility by public transport at the metropolitan-area scale using GTFS and demographic data shows that this approach allows the spatial analysis to conduct comparative approaches and determine the level of accessibility by different social groups to opportunity areas through public transportation (Bok and Kwon, 2016). Ma and Jan-Knaap (2014) used GTFS to evaluate the job access benefits created by the new Purple Line in Maryland. In Sao Paulo, Slovic et al. (2019) used GTFS data to show how access to jobs was positively correlated with a neighborhood’s socioeconomic status. Recently, mapping projects have provided useful GTFS data to conduct this type of data analysis; this is the case of the study conducted in Nairobi looking at accessibility levels using a gravity model to measure differences across socioeconomic groups in terms of access to social services (Campbell et al., 2019). By showing spatial inequalities clearly, this kind of research can be used to promote equity in the transportation planning processes and direct investment decisions to address longstanding inequities.

Despite all of this interest in GTFS data, there are still transportation networks that do not have publicly accessible GTFS data. The pioneering work of the Digital Matatus project showed how to bridge this divide and develop relevant GTFS data for unmapped transportation systems (Klopp et al., 2015). The World Bank has promoted the development of GTFS for transit systems in Asian cities, such as Manila, Haiphong, and Zhengzhou (Krambeck and Qu, 2015). There are certainly challenges associated with use data generated from mapping projects regarding informal transit services for their conversion into the GTFS format given that this type of data was designed for formal transit services with certain requirements such as the location of transit stops and frequencies. Usually, the flexibility of informal transport services without fixed stops, changing routes to avoid congestion and adapting to the transportation demand implies challenges when analyzing this data in the GTFS format, for instance, it is required to estimate the location of stops according to the data points generated in the mapping process. In this paper, we document our approach to mapping, data collection, and data management as part of the ongoing effort to develop a standardized approach to mapping and generating GTFS data for unmapped transit services.

Methodology

Study area

Bogotá is the capital of Colombia with a population of 7.2 million inhabitants in an area of 380 km2 (DANE, 2018). Bogotá is well known in the field of transportation due to the design and implementation of its BRT system. Since the year 2001, the BRT system began commercial operations transforming completely the operation of urban transport services in the city. Before the BRT, Bogotá had conventional buses operating within a business scheme known as the “penny war” due to bus drivers had to compete for passengers, picking them up at every corner or at any location along the roads (Ardila, 2004). Although the expansion of the BRT system aimed to change the operation of urban transport services in Bogotá, the conventional buses still operated in several areas of the city. With the beginning of commercial operations of phase three of the BRT system between 2012 and 2013, with the operation of the Transmilenio corridors along Av Calle 26 and Av Cra. 10, the local government began the implementation of the Integrated Public Transport System known as SITP because its acronym in Spanish (Sistema Integrado de Transporte Público) (SDM, 2018). In the supplementary material, we provide a detailed description of the study area and the SITP network with Supplemental Figure S1 (Bogotá, 2021; TransmilenioSA, 2022)

Data

Route data

We mapped all transit corridors within the built-up area of Bogotá, as determined by the Atlas of Urban Expansion, which was developed and maintained by New York University’s Marron Institute (Angel eta l., 2012). In this case, the built-up area map is very similar to the municipal boundaries with some slight exceptions. In order to ensure that we could draw the routes of the Provisional system, we hired 14 students from the Universidad del Rosario to go out into the city and map the system with smartphones (Goldwyn and Vergel-Tovar, 2018). This method—arming data collectors with smartphones—has been tested in multiple cities; however, the specific smartphone app used in each city has differed. After selecting three apps to test in New York and Bogotá, TransitWand, Open GPS Tracker, and Flocktracker, we selected Flocktracker (MIT, 2011); with this app, we received support from the Flocktracker team (Zegras et al., 2015b; Zegras et al., 2014). We provide a detailed description of the generation of route data in the supplementary material section with Supplemental Figures S2 and S3.

We generated a data base from the Provisional system that includes the total routes available at the time of data collection, the frequency data of the routes mapped. Based on this data, we were able to estimate additional indicators, which we compared with the formal SITP network (Table 1). Although the Provisional network has fewer routes than the SITP formal network, its performance shows that the Provisional bus routes move a higher number of passengers per route than the SITP formal system.

Table 1.

Comparison between SITP formal routes and SITP Provisional semiformal routes.

Characteristic	SITP formal transit routes	SITP Provisional
Number of routes	300	137
Average weekday headway (min)	23	21
Number of vehicle trips in 2019	1,483,030	8,98,208
Percentage of modal distribution in 2019	11.1%	6.7%
Coverage of the urban area^†	90.0%	82.6%
Percentage of the population covered by transit routes^†	99.2%	93.1%

Source: Data collection, (SDM, 2020b, 2022; TransmilenioSA, 2022) ^†Buffer of 500 m for all transit routes

Frequency data

The team also collected frequency data, a key input for the generation of GTFS data. This process included the selection of intersections in Bogotá where we had already identified a high level of Provisional routes. Data collectors stood at these intersections manually counting the minibuses that passed by during peak hours, morning, and afternoon. This data collection effort had some limitations such as missing off-peak and weekend data. While mapping the extent of the system was a key goal of our research, we also wanted to get a clearer understanding of key operating characteristics, such as fares, speeds, and frequencies along each route. While Flocktracker was used to collect some of this data, we also counted vehicles leaving patios to get a sense of the frequency during morning peak hours, spoke with drivers and people working at patios, and recorded posted fare data.

GTFS data

In order to maximize the utility of our data for government officials, researchers, and users of the Provisional bus system, we processed the raw field data and converted it into GTFS (Supplemental Figure S4). Using Python, we first imported the field data into a PostGIS database. For each GPS trace, we selected one out of every six points to represent a potential stop. This results in a high concentration of stops that can simulate the fact that Provisional services stop on demand. To avoid skewing travel times, we assigned a dwell time of zero to all stops. We then calculated the arrival time at each stop using the travel distance from the route origin and the overall cycle time. Using both PostGIS queries and Python algorithms, we repackaged the data into nine distinct tables that together form a valid GTFS feed. TransitFeed, a validation tool, revealed that several stops in our initial feed were too close to be legible in maps. We consolidated stops in such cases and were able to generate a feed that met the strict standards of the Google Maps routing engine. The GTFS data generated by this project is included with this paper with the purpose to make it publicly available.

In summary, our data collection and management processes suggest the importance of sequence in a collaborative mapping project. It is important to take an inventory of depots, identify origin points of routes, record GPS traces along routes, and conduct multiple frequency surveys. Even with all of the promise of technology, errors are inevitable. Specifically, we recommend manual trace corrections when GPS traces and snapping still do not draw perfect routes. This added step enhances accuracy but requires additional effort and data collectors who are knowledgeable about the transit system.

Data analysis

We conducted two types of data analysis that reveal how the Provisional complements, competes with, and adds redundancy to the SITP. First, we identified cases where Provisional and SITP routes run parallel. To automate the analysis, we developed an algorithm in PostGIS that implements the following steps:

1. Create a POLYLINE object for each route using the shapes.txt file of the GTFS feeds

2. Create an 800m buffer around each route to represent its catchment area

3. For each pair of routes, A and B:

a. Intersect the buffers to generate a MULTIPOLYON object with area

b. Calculate the collinearity of the pair according to the formula

C o l l i n e a r i t y_{A, B} = \frac{A r e a_{A \cap B}}{A r e a_{A}}

where

A is a Provision route

B is a SITP route

$A r e a_{A \cap B}$ is the area of overlap between the catchment areas.

$A r e a_{A}$ is the catchment area of route A

Second, we compared how Provisional and SITP systems currently provide access to formal jobs. We defined a Job Accessibility Index as the percentage of all formal jobs (SDP, 2017) accessible within 75 min by transit at 7:30 a.m. on Thursday, 11 October 2018. To calculate Job Accessibility for each cell of the hexagonal grid, we adapted code from Pacheco (2018) and Pereira (2019) that implements the following formula for cumulative opportunities

J o b A c c e s s i b i l i t y_{i} = \frac{\sum_{j = 1}^{n} J o b s_{j} * A c c e s s i b l e_{i j}}{T o t a l J o b s}

Where:

i = index of the origin hexagonal cell

j = index of the destination hexagonal cell

n = number of hexagon cells in study area

Accessible_ij = 1 if travel time by transit between cells i and j is less than 90 min, 0 otherwise

Total jobs = number of formal jobs in the federal district

We compared how Provisional and SITP systems currently provide access to formal jobs, and then assessed how fare integration would affect accessibility. Currently, users of the Provisional system have to pay double fare in order to transfer, while SITP users can transfer within the formal system at a discounted rate within a limited time window. If the SITP successfully integrates the Provisional system, its users would benefit from the same transfer discounts.

Results

We developed the map of SITP Provisional routes shown in Supplement Figures S1 and S5 based on more than 500,000 GPS data points. We collected this information over a 7-week period between June and August of 2017. Supplement Figure S5 includes the SITP formal transit network but excludes the BRT network since our analysis focuses on the Provisional and the SITP networks. We can see that the goals and operating logic of the Provisional system and SITP differ. The Provisional system routes covers longer distances than the centrally planned SITP system despite operating fewer routes. These longer routes allow Provisional minibuses to link multiple neighborhoods, shopping centers, and job clusters. SITP, on the other hand, offers different kinds of service, from high-capacity articulated buses playing primary thoroughfares and linking multiple zones to local service that concentrates on a specific zone.

Collinearity analysis

Since Provisional routes often start outside of the congested city center, but then funnel down main avenues, we found many cases of collinearity; the four most severe cases are mapped in Figure 1. If these SITP routes struggle to meet demand because of fleet constraints, the collinear Provisional routes may offer welcome relief. If, however, these SITP routes run half-empty or there is capacity to improve service, Transmilenio S.A. will likely integrate the overlapping Provisional service to reduce competition. Transmilenio S.A. must take care to properly accommodate Provisional passengers during these integration processes (SDM, 2018). Based on our data analysis, we found four routes with high levels of overlaps: i) Provisional ZP-261 and route SITP 736 overlap (87%): the Provisional reaching the urban border at the north and hilly areas at the south, ii) Provisional ZP-C42 and the route SITP C115 (85%): both routes are serving the southwest by providing access to the northeast, which is the area with a high concentration of job opportunities in the city; iii) Provisional ZP-C140 and the route SITP 344 (92%): both routes connect the northeast with the northwest, with a very similar route connecting the urban borders; and, iv) Provisional ZP-P4 and the route SITP 910 overlap by 86%: these routes connect the south west with the southeast with clear differences at the end points, serving two different hilly areas of informal settlements at the southeast of the city.

Figure 1.

Collinearity analysis. Source: Data collection and analysis. SITP routes are shown in blue.

Accessibility analysis

We found that Provisional, which has historically been excluded from transit maps and navigation apps, covers the entire federal district of Bogotá. How well the system provides access to the city’s opportunities, though, also depends on its transfer policy, travel times and operating schedules. We account for all of these factors by using the GTFS feeds to calculate a Job Accessibility Index for all cells of a 500-m hexagonal grid encompassing the urban district. Table 2 summarizes the quantitative results of the Job Accessibility Index for the following systems: i) Provisional system with no transfers: it benefits more formal and informal workers in terms of access to formal jobs; ii) Provisional system with transfers: when adding the transfers, the percentage of jobs within increased for all workers by 22.4%; iii) SITP with transfers: the advantages of the integrated system with the option of transfers between routes providing access to more than 73% of formal jobs in 75 min; and iv) Integrated SITP and Provisional System: all workers access to a 77.97% of formal jobs within 75 min of travel time when integrating both systems.

Table 2.

Results accessibility analysis by system and integration scenarios.

Type	Provisional without transfers 75 min	Provisional with transfers 75 min	SITP with transfers 75 min	Combined with transfers 75 min
All workers	40.56	62.60	73.30	77.97
Formal workers	40.76	62.90	73.59	78.23
Informal workers	35.96	55.77	66.75	72.09

Source: data analysis

The choropleth maps in Figure 2 visualize the Job Accessibility Index. Comparing Provisional and SITP reveals that the Provisional system provides much lower access to formal jobs than SITP for users unwilling to pay double fares to transfer to or from a Provisional minibus. Provisional with transfers reveals that fare integration would dramatically improve job accessibility for these users. Comparing SITP with Provisional and SITP shows that, while integrating Provisional routes will slightly improve access in peripheral areas at the west and south districts, the integration will not significantly improve job accessibility for most users of SITP. The result suggests that the Provisional system does not provide complementary access to formal jobs.

Figure 2.

Accessibility analysis. Source: data collection and analysis.

Discussion

Data collection techniques

Our research replicates data collection techniques that were applied in other environments such as Nairobi, Kenia (Klopp et al., 2014; Klopp et al., 2015; Williams et al., 2015), but adding new features such as developing new approaches when dealing with lack of information regarding the location of the origin and destination points of semiformal routes in cities. This project promotes the development of mapping processes that generated databases similarly to those techniques implemented in Dhaka, with the potential of informing citizens, local government agencies and researchers regarding transit routes (Ching et al., 2013; Zegras et al., 2014). We uncovered the location of “patios” or origin points of the SITP Provisional system, including key information regarding frequencies and transit routes that usually follow the demand by covering urban peripheries. Mapping projects generating open data sources regarding informal and semiformal transit services are helping to bridge the gap regarding the understanding transportation services that have been often neglected in the transportation planning and public policy processes (IADB, 2020; Klopp, 2021).

The mapping project of the SITP Provisional in Bogotá provided the visualization of the main transit networks of this semiformal transit service, including the distribution of routes following similar exercises in Kampala, Uganda; Accra, Ghana; Cape Town, South Africa; and Nairobi, Kenya (Coetzee et al., 2018; Ndibatya et al., 2016; Saddier et al., 2016). Our work also informed planning and public policy exercises given that the GTFS data that we generated has been used by Transmilenio and the Mobility Planning Department of Bogotá for the formalization and regularization of routes in order to integrate them into the Formal Transit Integrated Network of the city. Informing transportation planning processes including transportation user’s needs (Klopp and Cavoli, 2019) constitutes an important step in this process, and thus, our project certainly highlighted the importance of providing coverage at urban peripheries where “patios” of the SITP Provisional are located.

Data analysis and findings

The results of this project are important in terms of understanding the challenges associated with the formalization and regularization of the SITP Provisional system, especially in terms of the parallelism with the formal transit network (BRT and SITP). Our findings suggest that SITP Provisional routes have high collinearity levels in relation to the SITP Formal transit routes. This data analysis have already informed public policy and planning by providing open data that can facilitate the integration of the semiformal transit routes into the integrated transit network of the city. The generation of new data regarding transit services constitutes an important step toward generating responses regarding inequality problems in cities in terms of accessibility issues (Klopp et al., 2019). Our accessibility analysis adapting the codes from (Pacheco, 2018) and (Pereira, 2019) for their applicability in the case of Bogotá provided interesting insights regarding the access to job opportunities. The analysis shows how the integration of both systems, the SITP Provisional and the SITP Formal, can certainly increase access to more than 77% of the total formal jobs in the city.

Another key point regarding our work is related to the paradigm shift associated with the transition from informal transit or paratransit services operating alone toward their integration into transit networks which certainly understand the key role of these services from an equity perspective (Klopp, 2021). Our research findings provides a better understanding regarding the operation and coverage of the SITP Provisional service, and most importantly the relevance of their integration in order to improve accessibility for urban residents. This is important given that local residents at urban peripheries in Bogotá usually have to commute longer trips, including several transfers, while low-income users have lower levels of satisfaction with transit services in the city (BCV, 2021).

Limitations

We collected data that reflects the transit service provision of the SITP Provisional at one point in time. Thus, our data visualization provides information regarding the transit services before the formalization and integration of the semiformal transit system took a faster pace, especially during the last year. Another limitation of our work is that we focused on formal job opportunities based on available information. However, informal jobs data in Bogotá is unavailable, making difficult to conduct a similar accessibility analysis for informal job opportunities.

Conclusion

We learned a tremendous amount about how to carry out this work while mapping the unmapped transit system of Bogotá. The level of commitment required to map a system as large as the Provisional of Bogota proved too demanding for the civic mapping community, so we undertook a more coordinated approach that would be necessary in many cities. Considering the dynamic nature of Provisional’s service, this mapping exercise is a snapshot of the system. Thus, our work will likely remain as the last record of the semiformal transit network in the city, considering that the formalization of the Provisional is near completion. In 2022, the city already replaced the Provisional system with the formal network, and our work provided important inputs for this process with the location of Patios and the overlap of routes. We were fortunate to obtain data of the formal transit system directly from the Secretary of Mobility, though we also reached out to Google and others before the Secretary agreed to share the data with us.

Regarding the methodology, our findings suggest that the data collection methods proved to be useful to uncover the network of semiformal transit systems, especially how to identify the location of the origin points known as patios in Bogotá. In comparison to other semiformal transit systems, where there is a main hub in centrally located areas, the case of Bogotá highlights the dispersed geography of patios that are usually located in areas where there is more vacant land at urban peripheries. Also, the data collection techniques of frequencies by identifying 32 intersections with the highest number of transit routes constituted a successful approach in order to generate this key input for the GTFS data.

Mapping unmapped transit systems is an enormous undertaking that requires a lot of labor, coordination, and some sophistication. Despite these barriers, it can be done quickly, within a matter of months, and will become easier as data collection tools improve. For cities that are reliant on these systems, it is critical to understand where they operate so that investments can be made to improve access for passengers. BRT projects in Bogotá, Cape Town, and other cities have been used as a justification to overhaul unmapped systems (Paget-Seekins, 2015). Through this research, however, we have shown that cities can gain a clearer understanding of their transit networks by simply mapping. Using this type of fine-grained data, they can begin to redraw routes, plan new routes, and even choose to invest in BRT.

The two types of data analysis conducted in our paper show how the centrally planned bus system and the legacy franchised system coexist and relate to one another. First, we confirmed the high level of collinearity between Provisional and SITP routes in Bogotá. This is a huge challenge for the formalization process Transmilenio is conducting in order to integrate the two systems. This analysis provides useful findings that can inform the transportation planning process in order to identify the gaps filled currently by Provisional, areas that are usually in the periphery of the city originally developed as informal developments. This gap filled by the Provisional in Bogotá supports the theory that informal transport emerges to fill gaps in formal systems and satisfy transportation underserved demand (Cervero and Golub, 2007). Second, the accessibility analysis highlights how the axis along the Av Calle 26 toward the Airport concentrates a high number of job opportunities due to the concentration of several commercial and office activities along this major arterial road in Bogotá. The accessibility analysis also highlights how integration of the two systems can provide better access to all citizens, improve congestion, and help achieve sustainability goals.

Based on the literature review, our data collection techniques, the generation of GTFS data and subsequently the development of the data analysis, we consider that there are three research lines in mapping projects of semiformal and informal transit services. First, it is important to be flexible with data collection techniques and consider that they are context dependent. While it is possible to replicate the use of some tools such as Flocktracker, the geography of transit routes, their operational schemes and type of vehicles involved will influence significantly in the design of the most suitable approach that will capture the entire network by using smartphones. Second, the generation of GTFS databases constitutes an important step in terms of promoting open data sources and the possibility to conduct multiple types of data analysis regarding transit services that have been neglected in the transportation planning process (Klopp, 2021). We consider extremely important to train planners, scholars and practitioners in the techniques that facilitate the generation of GTFS data and the subsequent analysis in order to empower local actors (Williams, 2021; Zegras et al., 2015b; Zegras et al., 2014). Third, the generation of further open data bases regarding transit services will provide the basis of integrating these services into the transportation planning process of cities with high levels of informal transit services (IADB, 2020).

The research findings of this paper provide some key insights for transportation planning in Latin America and other regions with the coexistence of informal and formal transit services in urban areas. It is important to consider the network of semiformal and informal transit services in the transportation planning process of cities with these services. Avoiding competition and fostering cooperation through integration will not only have significant benefits in terms of the operational schemes but also will solve the coverage of transit deserts with low accessibility, mostly at urban peripheries. Our results also highlight the importance of integrating the semiformal and informal transit networks into the formal transit systems with advantages such as the reduction of parallelism between routes, reaching urban peripheries in a more effective way and increasing accessibility levels for residents art urban borders.

Finally, this data is critical to any plans to redesign Bogotá’s transit network, something that is an ongoing debate (Oviedo Hernandez and Dávila, 2016). Not only does this data relate to plans to extend BRT service or build an elevated train, but also to plans to redesign routes to improve reliability and efficiency. As we saw with some Provisional routes, they run between two peripheral points at a very slow speed. Before designing efficient and reliable routes, planners need access to the existing routes and operating characteristics. Right now, the private operators hold a knowledge advantage over the public sector, which has the potential to undermine the formalization process that has been in place since 2000. Therefore, this paper aims to promote the open data culture by including the GTFS data and Google Maps resources as supplementary materials to explore further the potential of this data in research, transportation planning and policymaking.

Supplemental Material

Supplemental Material - Digital traces: Mapping Bogotá’s unmapped transit network using smartphones and networked databases

Supplemental Material for Digital traces: Mapping Bogotá’s unmapped transit network using smartphones and networked databases by C Erik Vergel-Tovar, Eric Goldwyn and Jonathan Leape in Environment and Planning B: Urban Analytics and City Science

Footnotes

Acknowledgements

This project received funding from the Marron Institute of Urban Management at New York University. The data collection process received support of the Urban Management and Development (UMD) Program in the School of Political Science, Government and International Relations at the Universidad del Rosario. The authors greatly acknowledge the support for data collection from the following students of the UMD program at Universidad del Rosario: Maria Fernanda Cortes Duran, Sebastian Orjuela Rincon, Carlos Andres Patino Lopez, Laura Sofia Morales Alvarado, Anderson Julian Acevedo Herreno, Paulo Federico Martinez Ramirez, Martin Ricardo Silva Tovar, Camilo Hernando Arenas Mejia, Laura Daniela Ujueta Rodriguez, Maryfely Rincon Malaver, Juan David Galeano Moreno, Nicolas Gutierrez Garcia, Maria Camila Roa Urrego and Daniel Santiago Mayorga Riano. The authors also acknowledge the support of, Zolangwi Garzon, Diego Bermudez and Alix Herrera from Universidad del Rosario and Vipassana Vijayarangan, Alex Blei, Nicolas Galarza, Kelly Miller, Brandon Fuller, Clayton Gillette and Dr. Solly Angel from New York University's Marron Institute of Urban Management. The corresponding author also acknowledges the support given by the administrative staff at Universidad del Rosario, to which Professor Vergel-Tovar was formerly affiliated. We also wish to thank Sarah Williams and Dr. Jackie Klopp for sharing their experiences with mapping in Nairobi and answering our general questions in the research design phase of the project. We would like to thank the Flocktracker team for fielding our questions and helping us figure out how to achieve our goals with their app. We greatly appreciate the support from the Secretaria Distrital de Planeacion (SDP) for providing us the Census of Public and Private Facilities (Censo de Establecimientos), which allowed us to conduct the jobs accessibility analysis. We also greatly acknowledge the support of TRANSMILENIO S.A. and the Secretaria Distrital de Movilidad (SDM) for providing us insightful feedback on our work and guidance related to the SITP Provisional operational schemes.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by New York University.

ORCID iD

C. Erik Vergel-Tovar

Supplemental Material

Supplemental material for this article is available online.

Author Biography

C. Erik Vergel-Tovar Assistant Professor in the Department of Architecture at Universidad de los Andes. PhD in City and Regional Planning at the University of North Carolina at Chapel Hill (US), Master’s on Urban Management and Development with Distinction (IHS, Rotterdam) and Architect from the National University of Colombia in Bogotá.

Eric Goldwyn Program director at the Marron Institute of Urban Management and an Assistant Professor in the Transportation and Land-Use program at the NYU Marron Institute. Ph.D. in Urban Planning from Columbia University.

Jonathan Leape Expert advising clients developing projects and policies in the built environment. Master's in Civil Engineering and City Planning from the Massachusetts Institute of Technology MIT.

References

Angel

Blei

Civco

, et al. (2012) Atlas of Urban Expansion. Cambridge, MA: Lincoln Institute of Land Policy.

Ardila

(2004) Transit Planning in Curitiba and Bogota. Roles in Interaction, Risk and Change. Cambridge: Massachusetts Institute of Technology MIT.

BCV (2021) Informe Calidad de Vida en Bogotá 2020. Bogotá. Retrieved from. https://bogotacomovamos.org/vigesimo-tercer-informe-de-calidad-de-vida-en-bogota-2020/

Bogotá , Datos Abiertos Bogotá (2021). accessed January 2021 https://datosabiertos.bogota.gov.co/

Bok

Kwon

(2016) comparable measures of accessibility to public transport using the general transit feed specification. Sustainability 8(3): 224. Retrieved from: https://www.mdpi.com/2071-1050/8/3/224

BRTDATA ORG. (2021) Global BRT Data. Retrieved from: https://brtdata.org/location/latin_america

Campbell

Rising

Klopp

, et al. (2019) Accessibility across transport modes and residential developments in Nairobi. Journal of Transport Geography 74: 77–90. DOI: 10.1016/j.jtrangeo.2018.08.002.

Cervero

Golub

(2007) Informal transport: a global perspective. Transport Policy 14(6): 445–457.

Ching

Zegras

Kennedy

, et al. (2013) A User-Flocksourced Bus Experiment in Dhaka: New Data Collection Technique with Smartphones. Transportation Research Record: Journal of the Transportation Research Board.

10.

Coetzee

Krogscheepers

Spotten

(2018) Mapping Minibus-Taxi Operations at a Metropolitan Scale–Methodologies for Unprecedented Data Collection Using a Smartphone Application and Data Management Techniques. Pretoria, South Africa: 37th Annual Southern African Transport Conference (SATC 2018), Proceedings ISBN Number: 978-1-920017-89-7.

11.

DANE (2018) Censo Poblacion, Departamento Administrativo Nacional de Estadistica, DANE. Bogotá, Colombia: Departamento Administrativo Nacional de Estadística.

12.

DATUM (2022) Datos abiertos de transporte urbano y movilidad. Retrieved from https://datum.la/.

13.

Eros

Mehndiratta

Zegras

, et al. (2014) Applying the general transit feed specification to the global South: experiences in Mexico City, Mexico—and beyond. Transportation Research Record: Journal of the Transportation Research Board 2442(1): 44–52.

14.

Goldwyn

Vergel-Tovar

(2018) Mapping the Unmapped Transit Network of Bogotá, Colombia. Washington DC, United States: Transportation Research Board 97th Annual Meeting. Available at: https://trid.trb.org/view/1495944

15.

Google (2020) GTFS Static Overview. Retrieved from: https://developers.google.com/transit/gtfs

16.

GTFS (2020) GTFS: Making Public Transit Data Universally Accessible. Retrieved from http://gtfs.org/.

17.

Guzman

Oviedo

Cardona

(2018) Accessibility changes: analysis of the integrated public transport system of Bogotá. Sustainability 10(11): 3958.

18.

Hansen

(1959) How accessibility shapes land use. Journal of the American Institute of Planners 25(2): 73–76.

19.

IADB (2019) DATUM: A Resource Center for Participatory Urban Transit Mapping — The case of Santiago de los Caballeros. Retrieved from https://blogs.iadb.org/ciudades-sostenibles/en/datum-open-transit-mapping-dominican-republic-santiago-caballeros/

20.

IADB (2020) Informal and Semiformal Services in Latin America: An Overview of Public Transportation Reforms. Washington DC: Retrieved from.

21.

ITDP (2016) Mapatón CDMX busca generar una base de datos abiertos del transporte público concesionado. Retrieved from https://mexico.itdp.org/noticias/mapaton-cdmx-busca-generar-una-base-de-datos-abiertos-del-transporte-publico-concesionado/

22.

Jones

Layard

Speed

, et al. (2015) MapLocal: use of smartphones for crowdsourced planning. Planning Practice & Research 30(3): 322–336.

23.

Klopp

Mutua

Orwa

, et al. (2014) Towards a Standard for Paratransit Data: Lessons from Developing GTFS Data for Nairobi's Matatu System. Retrieved from.

24.

Klopp

Williams

Waiganjo

, et al. (2015) Leveraging cellphones for wayfinding and journey planning in semi-formal bus systems: lessons from digital Matatus in Nairobi. Planning support systems and smart cities. Springer, 227–241.

25.

Klopp

(2021) From “para-transit” to transit? Power, politics and popular transport. Advances in Transport Policy and Planning 8: 191–209.

26.

Klopp

Cavoli

(2019) Mapping minibuses in Maputo and Nairobi: engaging paratransit in transportation planning in African cities. Transport Reviews 39(5): 657–676. DOI: 10.1080/01441647.2019.1598513.

27.

Klopp

Delattre

Chevre

(2019) Open Data for Inclusive Urban Public Transport Globally. Agence Française de Développement (AFD). Available at: https://www.digitaltransport4africa.org/covid-19-hit-public-transport-users-hard-heres-how-open-data-a-digital-commons-approach-helps/.

28.

Krambeck

(2015) Toward an open transit service data standard in developing Asian Countries. Transportation Research Record 2538(1): 30–36.

29.

Kujala

Weckström

Darst

, et al. (2018) A collection of public transport network data sets for 25 cities. Scientific Data 5(1): 180089. DOI: 10.1038/sdata.2018.89.

30.

MIT (2011) Flocktracker https://mfc.mit.edu/flocktracker

31.

Ndibatya

Booysen

Coetzee

(2016) Mapping the Informal Public Transport Network in Kampala with Smartphones: Making Sense of an Organically Evolved Chaotic System in an Emerging City in Sub-Saharan Africa.

32.

OECD (2017) Case Study Mapatón – Mexico City, Mexico. In: Embracing Innovation in Government: Global Trends. Retrieved from https://www.oecd.org/gov/innovative-government/embracing-innovation-in-government-mexico.pdf

33.

OECD (2017) Zoom in or Zoom Out: Scaling Government.

34.

Oviedo Hernandez

Dávila

(2016) Transport, urban development and the peripheral poor in Colombia - placing splintering urbanism in the context of transport networks. Journal of Transport Geography 51: 180–192. DOI: 10.1016/j.jtrangeo.2016.01.003.

35.

Pacheco

(2018) Moradia, Localização e o Programa Habitacional “Minha Casa Minha Vida” No Município Do Rio de Janeiro. (Master in Public Administration and Government). Fundação Getulio Vargas, Sao Paulo. Retrieved from: http://hdl.handle.net/10438/27253

36.

Paget-Seekins

(2015) Bus rapid transit as a neoliberal contradiction. Journal of Transport Geography 48: 115–120. DOI: 10.1016/j.jtrangeo.2015.08.015

37.

Pereira

RHM

(2019) Future accessibility impacts of transport policy scenarios: equity and sensitivity to travel time thresholds for bus rapid transit expansion in Rio de Janeiro. Journal of Transport Geography 74: 321–332. DOI: 10.1016/j.jtrangeo.2018.12.005.

38.

Saddier

Patterson

Johnson

, et al. (2016) Mapping the jitney network with smartphones in accra, ghana: the accramobile experiment. Transportation Research Record: Journal of the Transportation Research Board (2581): 113–122.

39.

Sandoval-Almazan

Gil-Garcia

Valle-Cruz

(2017) Going Beyond Bureaucracy Through Gamification: Innovation Labs and Citizen Engagement in the Case of “Mapaton” in Mexico City. Beyond Bureaucracy. Public Administration and Information Technology. Springer, 25.

40.

SDM (2018) Secretaría de Movilidad y TRANSMILENIO S.A. arrancan el desmonte definitivo del servicio Provisional. Retrieved from https://www.movilidadbogota.gov.co/web/node/2531

41.

SDM (2020a) Arrancó el proceso de desintegración de 58 vehículos del SITP Provisional que adquirió el Fondo para el Mejoramiento de la Calidad del Servicio. Retrieved from https://www.movilidadbogota.gov.co/web/noticia/arranco_el_proceso_de_desintegracion_de_58_vehiculos_del_sitp_provisional_que_adquirio_el

42.

SDM (2020b) Encuesta de Movilidad 2019. Retrieved from: https://www.movilidadbogota.gov.co/web/encuesta_de_movilidad_2019

43.

SDM (2021) Las 15 acciones del Distrito en el proceso de desmonte del SITP Provisional. Retrieved from: https://bogota.gov.co/mi-ciudad/movilidad/sitp-provisional-acciones-del-distrito-en-el-proceso-de-desmonte

44.

SDM (2022) Sistema integrado de información sobre movilidad urbana y regional SIMUR. In: Secretaría Distrital de Movilidad SDM de Bogotá.

45.

SDP (2017) Censo de Establecimientos. Bogotá, Colombia: Secretaría Distrital de Planeación SDP.

46.

Slovic

Tomasiello

Giannotti

, et al. (2019) The long road to achieving equity: job accessibility restrictions and overlapping inequalities in the city of São Paulo. Journal of Transport Geography 78: 181–193.

47.

Tao

Rohde

Corcoran

(2014) Examining the spatial–temporal dynamics of bus passenger travel behaviour using smart card data and the flow-comap. Journal of Transport Geography 41: 21–36. DOI: 10.1016/j.jtrangeo.2014.08.006.

48.

TransmilenioSA (2022) Portal de Datos Abiertos de TransMilenio. Retrieved from https://datosabiertos-transmilenio.hub.arcgis.com/

49.

UN (2020) Sustainable Development Goals. Goal 11: Make Cities Inclusive, Safe, Resilient and Sustainable. Retrieved from https://www.un.org/sustainabledevelopment/cities/

50.

Vergel-Tovar

Villegas Carrasquilla

Peñas

, et al. (2021) Collaborative Mapping of Urban Transport in Cartagena, Colombia. Retrieved from.

51.

Williams

(2021) Data Action Using Data for Public Good. Cambridge, MA: MIT Press.

52.

Williams

White

Waiganjo

, et al. (2015) The digital matatu project: using cell phones to create an open source data for Nairobi’s semi-formal bus system. Journal of Transport Geography 49: 39–51.

53.

Zegras

Eros

Mehndiratta

(2015a) Digitalized Public Transportation Data as Disruptive Innovation: Prospects in Mexico City.

54.

Zegras

Butts

Cadena

, et al. (2015b) Spatiotemporal dynamics in public transport personal security perceptions: digital evidence from Mexico City’s periphery. ICT for Transport. Opportunities and Threats 115.

55.

Zegras

Eros

Butts

, et al. (2014) Tracing a path to knowledge? Indicative user impacts of introducing a public transport map in Dhaka, Bangladesh. Cambridge Journal of Regions, Economy and Society 8(1): 113–129.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.11 MB