Dataset Development of Global Railway Length in Operation (1825?C2021)

Li, Y. L.  Yu, Y. H.^*  Cai, J.

School of Architecture and Fine Arts, Dalian University of Technology, Dalian 116023, China

Abstract: As the backbone of modern transportation infrastructure, railways not only embody the level of transport development in different countries, but also capture the profound transformations in economic, social, and geopolitical structures. To address the gaps in historical data on the global length of railways in operation, this study integrates data from Brian Mitchell??s International historical statistics, the World Bank, the International Union of Railways, and national statistical offices to construct a dataset of the global length of railways in operation for the period from 1825 to 2021. The missing data were systematically categorized into 3 types: mid-series gaps, end-period gaps, and gaps caused by national boundary adjustments. To complete the dataset, we apply linear interpolation, regression-based forecasting, and static/dynamic weighting methods tailored to each type. Building on the reconstructed dataset, we conduct a spatiotemporal analysis of global railway development at regional as well as national scales. The findings reveal that: (1) global railway development can be divided into 5 distinct stages with clear phases and fluctuations; (2) significant regional disparities exist, with Europe and North America leading in the early stages, and Asian countries emerging as key growth engines in the 21st century; and (3) railway development has been shaped significantly by geopolitical and geo-economic dynamics, with shifting patterns of interaction across historical periods that determine the trajectory of global railway evolution. This study provides systematic data support for understanding the historical evolution of global infrastructure, establishing a solid foundation for exploring the interplay between transportation and socioeconomic changes.

Keywords: global; railway operating length; stage division; evolution trajectory; regional differentiation

DOI: https://doi.org/10.3974/geodp.2025.03.03

Dataset Availability Statement:

The dataset supporting this paper was published and is accessible through the Digital Journal of Global Change Data Repository at: https://doi.org/10.3974/geodb.2025.08.07.V1.

1 Introduction

As a core component of modern transportation systems, railways play a vital role in national, economic, and social development. Since the advent of modern railway technology in the 19th century, railway networks have expanded rapidly, forming highly interconnected transport systems on a global scale. Railways have profoundly influenced urban systems, regional development, national spatial organization, and the global transport network by improving accessibility, reducing transport costs, and strengthening the connectivity of key cities and regions. The length of operational railways serves as a key indicator of railway development, directly reflecting the scale of infrastructure construction and service capacity in a given country or region. Analyzing long-term trends not only reveals the overall patterns and evolutionary paths of global transport infrastructure but also deepens our understanding, through cross-country comparisons, of how factors such as economic stage, geographic conditions, and institutional settings shape railway construction. Moreover, such analysis elucidates shifts in global infrastructure investment priorities and highlights the persistent issue of regional imbalances. Thus, systematically examining the spatiotemporal evolution of the length of operational railways and its driving forces is of significant theoretical and practical value for advancing our understanding of the interaction between transportation infrastructure and socio-economic development.

However, integrating historical data on the length of operational railways on a global scale presents several challenges. Wars, economic disruptions, and technological constraints have frequently resulted in missing data for certain historical periods, undermining the continuity of the time series. Additionally, profound geopolitical transformations, such as the dissolution of multinational states, including the Soviet Union and the Austro-Hungarian Empire, have made it difficult to reconcile historical records with current national borders, further complicating data integration. These issues have constrained systematic, long-term, and cross-regional comparative research on railway development. Consequently, building a consistent and comparable historical data framework is crucial for advancing the global railway research.

To address these challenges, this study systematically compiles data on the global length of railways in operation from 1825 to 2021 (covering conventional railways but excluding high-speed railways). The primary data source is Brian Mitchell??s International historical statistics, 1750?C2010^[1], while the post-2010 data are drawn from the World Bank, International Union of Railways (UIC), and official publications of national statistical offices. The resulting dataset covers 133 countries and regions across 5 continents.

2 Metadata of the Dataset

The metadata of the Global dataset on the length of railways in operation (1825?C2021)^[2], including the title, authors, geographic coverage, temporal resolution, data format, dataset composition, etc., are summarized in Table 1.

3 Methods

3.1 Data Sources

The dataset is primarily compiled from International historical statistics, 1750?C2010^[1], the World Bank^[4], and official publications of national statistical offices. It covers the annual data on the length of railways in operation for the period from 1825 to 2021. Given that some countries have not yet developed railway infrastructure and that historical archives are incomplete or not publicly accessible in certain cases, data acquisition proved to be challenging. The final dataset includes 133 countries and regions worldwide, accounting for approximately 56.7% of all countries and territories. These are distributed across the 5 continents: 29 in the Americas, 36 in Europe, 30 in Asia, 36 in Africa, and 2 in Oceania.

Table 1  Metadata summary of Global dataset on the length of railways in operation (1825?C2021)

3.2 Missing Data Treatment

The dataset spans 1825?C2021, covering 133 countries and regions across 5 continents. Owing to incomplete historical archives, approximately 17.07% of the data were missing. Missing data can be classified into 3 categories: (1) mid-series gaps, where data for a given country are missing for one or several consecutive years within an otherwise continuous time series; (2) end-period gaps, where data are missing toward the end of the time series, requiring extrapolation of trends; and (3) boundary-related gaps, where geopolitical changes, such as national boundary adjustments, necessitate the redistribution of historical records to align with present-day borders.

To enhance completeness and usability, we adopted different strategies for each type: linear interpolation for mid-series gaps, linear regression forecasting for end-period gaps, and static/dynamic weighting methods for boundary-related gaps. Additional adjustments and optimizations were performed to ensure consistency.

3.2.1 Mid-series Missing Data

For cases where the length of railways in operation is missing for one or several consecutive years within a country??s time series, we apply a linear interpolation method. This approach uses observed values from adjacent years, assuming that the railway length evolves linearly within the missing interval. Compared with more complex interpolation or fitting techniques, linear interpolation requires no additional parameter assumptions, ensuring the continuity of the time series while maintaining simplicity and interpretability. It is particularly suitable in contexts where railway development is relatively stable and the differences between pre- and post-gap values are not substantial. This helps avoid biases that may arise from overfitting or inappropriate parameterization. The Equation is as follows: 

                                                                                          (1)

where  denotes the length of the railways in operation (km) for the target year t; ​​ and  represent the 2 adjacent years with known values, andandare the corresponding railway lengths (km). For each missing year within the interpolation interval, the value can be estimated recursively using this equation. This method not only demonstrates strong numerical stability and logical consistency, but also maintains controllability and transparency in reconstructing data when historical records are incomplete or gaps are extensive. Consequently, it is widely employed for the restoration and completion of historical statistical time-series data.

3.2.2 End-period Missing Data

For missing data occurring at the end of the time series, requiring extrapolation of railway length in operation for future years, we applied a linear regression method to fit recent observations and construct a predictive model. In most countries, recent trends in railway length indicate relative stability. Railway systems in developed countries have largely matured, whereas the expansion of railways in developing countries is typically constrained by policies and funding, resulting in slow year-to-year changes. Therefore, linear regression is well-suited for short-term forecasting, offering both applicability and interpretability.

Specifically, we fit a linear regression model using observations from the 5 years immediately preceding the missing period. The slope extracted from the fitted trend was then used for extrapolation. The parameters of the linear regression model??the intercept a and slope b??were estimated using the following Equations:

                                                                                             (2)

                                                                                            (3)

whereandrepresent the year and corresponding length of the railways in operation (km), respectively, and n denotes the number of years used for model fitting (here, n=5). Based on this model, the value of the k_th future year can be predicted as follows:

                                                    ,                                                (4)

where k denotes the forecasting horizon in years (e.g., k=1 indicates a one-year-ahead prediction).

3.2.3 Data Splitting and Allocation for Countries with Boundary Reconstructions

Historical changes in national boundaries pose significant challenges to the attribution of railway length data. The processes of political disintegration, mergers, and repeated boundary adjustments often result in historical records being compiled under the jurisdiction of larger political entities, which cannot be directly mapped to the present-day system of nation-states. Such spatial inconsistencies undermine both the temporal continuity of railway statistics and cross-country comparability. To address this issue, we distinguished 2 types of boundary-related cases based on the structure of historical data:

The first scenario is early aggregate data with later disaggregated national statistics. In situations where only aggregate data for a larger political entity were available in earlier periods and separate national records appeared after disintegration or boundary stabilization, we employed a static weight back-casting method. Specifically, the aggregate data were retroactively disaggregated according to each country??s share of the regional total after the borders were stabilized, thereby estimating the historical railway lengths for individual countries.

The second scenario is aggregate data during specific periods with independent statistics before and after. This case typically arises when states undergo cycles of merger and separation. In such situations, we adopted a dynamic weight interpolation method. Here, the aggregate data were disaggregated through a year-by-year linear interpolation of country-level shares observed at the preceding and subsequent time points, allowing for a more accurate reconstruction of structural changes in railway development.

(1) Static weight retrospective method

For regions lacking separate national railway data during historical periods and possessing only modern national statistics after a country??s dissolution or border demarcation, this study employed a static weight retrospective method for data disaggregation. The specific procedure is as follows: first, obtain the railway operating mileage data for each country after its borders have stabilized, and calculate each country??s proportion of the total regional mileage. This proportion was used as a weight to allocate the overall historical statistical data retrospectively, thereby estimating each country??s contribution to the historical total.

Taking the Soviet Union as an example, this study selected the average annual railway operating mileage of each member republic from the decade following its dissolution (1992?C2001). This was used to calculate the respective weights of the total regional mileage, which were then applied to disaggregate the overall data from the Soviet era into individual countries. By constructing an average weight from a stable period, this method effectively smoothened short-term data fluctuations. Furthermore, considering the path-dependent nature of railway networks ensured reasonable consistency between the disaggregation of historical data and the actual spatial layout, thereby enabling a retrospective estimation of each country??s historical railway data.

Assuming the total railway operating mileage for a given year t before a change was  (km), and the average railway operating mileage for the present-day country j in the decade following the completion of border adjustments is ​, the corresponding allocation weight for country j is:

                                                                                                           (5)

where J is the set of all the newly formed national units. Based on this, the estimated railway operating mileage for country j in historical year t can be calculated as:

                                                                                                            (6)

(2) Dynamic weight interpolation method

For some regions, historical railway data is only available as a regional total without country-specific breakdowns; however, independent national railway data exists both, before and after this period. To address this issue, we proposed a dynamic weight interpolation method. This method uses the proportions of railway mileage at two distinct time points (before and after the period of combined data) to create a smoothly changing year-by-year weight series, enabling the disaggregation of historical data by country.

Consider the cases of Uganda, Kenya, and Tanzania, from 1948 to 1974, the railway systems of the three countries were unified under the British East African Railways and Harbors Corporation, with data recorded only in aggregate form. However, country-specific railway data are available for 1947 (before unified management) and 1975 (after countries began independent operations). Based on the mileage proportions at these two time points, it was assumed that each country??s share of the railway system gradually changed linearly during the unified management period. This approach generates dynamic weights for each year, which are then used to disaggregate the data into country-specific figures. This method not only maintains consistency with regional totals but also reflects the evolving structure of the railway system, aligning the disaggregated results more with actual development trends.

Specifically, let the total regional railway operating mileage for a given year t be ​, which needs to be disaggregated into n national units. Assuming that independent country-specific data are available for both the year before disaggregation, , and the year after, ​, the weights for these two time points can be calculated as:

                                                                    (7)

where  is the railway operating mileage (km) of country i in year t, i=1,2,, n are the indices for all countries involved in the allocation, and  is the proportion of country i in the regional total (%).

Based on this, we constructed a linear weight function for each year t,  within the time period [, ]:

                                                                                    (8)

Then, the merged total for that year, , was allocated according to the weight to obtain the estimated value for country i:

                                                                                                              (9)

The dynamic weight interpolation method is suitable for groups of countries where historical statistics lack country-specific data for an intermediate period; however, comparable independent data were available both before and after this period. This method was particularly well suited for regions with relatively stable railway development and a clear evolutionary pattern in each country??s proportion of the total. Compared to the static weight retrospective method, the dynamic interpolation method is better at reflecting how each country??s share changes over time, thus improving the time sensitivity and accuracy of historical data disaggregation.

4 Data Results and Validation

4.1 Dataset Composition

The Global dataset on the length of railways in operation (1825?C2021) was archived in an .xlsx format. It contained 27,186 data entries, comprising 1.43 MB of data. The dataset included country names, continents, years, length of railway in operation, railway operating growth rates, and data sources.

4.2 Data Results

4.2.1 Regional Level

Figure 1 shows the evolutionary trajectory of the railway operating mileage across the five continents from 1825 to 2021. Based on developmental characteristics and major historical events, the evolution of the global railway network can be divided into 5 distinct phases, each exhibiting significant regional differentiation.

Phase 1 (1825?C1870): The period of emergence and diffusion. Driven by the Industrial Revolution and the demands of colonial expansion, railways emerged first in Europe and North America as a key transport mode for industrialization. Railway construction on both continents demonstrated synchronous and rapid growth.

Phase 2 (1870?C1913): The golden age of railway construction. The railway network in the Americas experienced explosive expansion; Europe maintained stable growth; and Asia, Africa, and Oceania successively launched their own railway construction booms, propelling global railway mileage into an unprecedented period of high-speed growth.

Phase 3 (1913?C1945): The slump. Severely impacted by the two world wars and the Great Depression of 1929, global railway development fell into a slump. The length of the railway in operation of the Americas shrank significantly because of the widespread bankruptcy of US railway companies. Asia, Africa, and Oceania also experienced varying degrees of decline as war-related destruction and economic depression collectively constrained network expansion.

Phase 4 (1945?C2000): Regional differentiation. This phase exhibited clear regional differences. While Asia maintained its expansion momentum driven by government-led infrastructure projects, railway operating mileage in other continents generally showed a downward trend.

Phase 5 (2000?C2021): New growth cycle. Global railway development entered a new growth cycle with Asia becoming its primary engine. The region achieved a remarkable 24.8% increase in its railway operating mileage and its share of the global total reached 26.5%. This growth was primarily fueled by China??s railway network upgrades through large-scale new-line construction and the renovation of existing lines.

4.2.2 National Level

Analyzed at the national level, Figures 2?C6 illustrate the typical characteristics of railway development in major countries across each continent.

In Europe (Figure 2), the United Kingdom, as a pioneer of the railway revolution, rapidly built a national network after completing the world??s first railway in 1825. Its technical standards, management systems, and construction experience profoundly impacted global railway development, with its length of railways in operation reaching over 30,000 km before World War I. In contrast, Russia exhibited a continental-style railway development model. The construction of trunk lines, such as the Trans-Siberian Railway, in the late 19th century established one of the world??s longest railway networks, which further expanded to over 80,000 km during the Soviet era, becoming a core infrastructure supporting national development and economic planning.

Western and Central European regions showed differentiated development paths: Germany, after unification, accelerated network integration through railway nationalization, boosting heavy industry, and the formation of a unified national market; France, conversely, adhered to state-led centralized planning, constructing a radial network centered on Paris. Notably, despite its small size, Belgium maintained a leading position in Europe for a long time because of its early railway construction and high-density network.

The two world wars caused widespread destruction of European railways, and post- war, they faced fierce competition from road transport, leading to a continuous reduction in the length of railways in operation for most countries. This trend reversed with the advent of the high-speed rail era. The railway development trajectories of these countries reflect the inherent logic of technological evolution and profoundly embody the interactive relationship between national governance models and geopolitical patterns.

In the Americas (Figure 3), railway development exhibits distinct regional characteristics. The United States, a bench­mark for global railway development, saw its railway network peak in the early 20th century, with the length of railways in operation exceeding 400,000 km, setting a world record. This remarkable achievement was due to a period of rapid expansion in the latter half of the 19th century, when railways were seen as critical infrastructure for connecting the east and west coasts and promoting westward expansion. However, the Great Depression in 1929 was a turning point. With the rise in road and air transport, the dominance of railways in the transportation system gradually weakened, leading to a continuous reduction in the length of operational railways^[5].

Canada, as North America??s second-largest railway nation, shares similarities with the US but also has unique aspects in its development trajectory. Throughout the 20th century, Canada??s railway network experienced stable growth, a trend that shifted only after the Canadian National Railway implemented line optimization adjustments in 1983.

Railway development in Latin America exhibits a different pattern. The railway networks of major countries, such as Brazil, Mexico, and Argentina, have been constrained by factors, such as economic development levels, industrial policy adjustments, and geographical conditions, generally exhibiting progressive development characteristics, with the length of railways in operation fluctuating within a relatively stable range over the long term. This developmental disparity not only reflects the unique paths of industrialization in each country but also profoundly illustrates the far-reaching impact of changes in transportation modes on infrastructure development.

In Asia (Figure 4), railway development exhibits the distinct dual characteristics of the colonial legacy and modern transformation. As a product of the British Empire??s colonial system, Indian railways began in the mid-19th century, building a vast network of over 60,000 km by the eve of independence in 1947. This colonial legacy ensured its status as having the longest operational railway network in Asia. During the Meiji Restoration (1868?C1912), Japan incorporated railway construction into its national modernization strategy through technology introduction and institutional innovation^[6], forming a railway system with both military and economic functions before World War II, and ranking among the top in Asia in terms of operational efficiency and technological level.

China??s railway development has undergone a unique evolutionary trajectory: development was slow from the late Qing Dynasty to the Republic of China period due to control by foreign powers and wartime disruptions. After 1949, China gradually achieved independent construction and entered an accelerated development phase after the reform and opening up. Since 2008, the implementation of the high-speed rail strategy has caused leapfrog development in China??s railways. In just over a decade, the completion of the ??four vertical and four horizontal?? high-speed rail backbone network made China the global leader in high-speed rail length^[7]. This development not only reshaped the Asian railway landscape but also marked a historic shift in global railway development, moving the focus from Europe and North America to East Asia. The distinct railway development paths of these three countries reflect the complex influences of colonial history, state capacity, and developmental strategies on infrastructure construction.

On the African continent (Figure 5), railway development exhibits a deep colonial imprint and an unbalanced character. As a typical representation of the colonial transport system, the construction of the African railway network began in the mid-to-late 19th century, with its layout entirely serving the resource import needs of the European powers. South African railways were oriented towards gold and diamond transport, Congolese railways focused on copper and rubber exports, and East African railways became part of Britain??s ??Cape to Cairo?? colonial scheme^[8].

This distorted development led to Africa??s railways exhibiting ??three highs and three lows??: high density of resource export lines but low inland connectivity, high accessibility to coastal ports but low regional interconnectivity, and high single-line transport capacity but low network coverage. South Africa was an exception, with its railway network exceeding 20,000 km in the mid-20th century owing to abundant mineral resources and a relatively mature industrial base. Despite an early start, North African countries, such as Egypt and Algeria, experienced long-term stagnation in railway development owing to their colonial economic structure. Central African countries, such as the Democratic Republic of Congo, witnessed severe decline in their railway systems due to post-independence political instability, with some lines built during the colonial period falling into complete disuse.

In Oceania, railway development shows significant national differences (Figure 6). Australia, the dominant force in the region??s railway development, began railway network construction in the mid-19th century and experienced a construction boom around the time of the Federation (1901). The peak length of railways in operation exceeded 45,000 km in the 1920s. In the mid-to-late 20th century, influenced by competition from road transport and fiscal tightening, some branch lines were abandoned, while major trunk lines remained operational. Notably, since the 21st century, with the boom in mining and the development of urban rail transit, some regional railways have experienced a resurgence. In contrast, New Zealand??s railway development is characterized by being ??small and stable??, with its network size peaking at 5,700 km in the 1950s before slowly contracting to approximately 4,000 km currently.

4.3 Data Validation

To assess the accuracy of the imputed data, this dataset employed the hold-out method for validation, and 20% of the known data were randomly sampled from the original dataset and treated as pseudo-missing values. These values were then imputed using the methods described above and compared with their original values. The Mean Absolute Percentage Error (MAPE) was used as the evaluation metric using the following equation:

                                               ,                                         (10)

where n is the number of pseudo-missing points;is the true value of length of railways in operation (km); andis the predicted value of the pseudo-missing points (km).

The results showed that the mean absolute percentage error for the relevant data of each country was within 10%, indicating a high level of accuracy of the imputation methods (Figure 7).

Figure 7  Data prediction error for selected countries

5 Discussion and Conclusion

This study systematically constructed a long-term dataset of the length of operational railways globally from 1825 to 2021. Various imputation and validation methods were employed to address missing historical data, thereby providing systematic and reliable data support for research on global infrastructure development. Analysis based on these data indicates that railways are not only a vital mode of transport but also a crucial lens through which the modernization process of countries can be observed, with their development trajectory profoundly mirroring the evolution of the world??s economic and political landscape.

The historical evolution of the global length of railways in operation exhibits 3 significant characteristics. First, railway development shows distinct stages, with rapid growth in the 19th century driven by the Industrial Revolution, a slowdown in the mid-20th century due to wars and the impact of emerging transport modes, and a resurgence in the 21st century with the rise of emerging economies. Second, regional development disparities are significant, with core countries, such as Europe and North America, dominating early, while peripheral countries, such as those in Asia, caught up rapidly. Finally, global railway development is deeply influenced by geopolitical and geoeconomic factors, with interaction patterns in different historical periods shaping the trajectory of railway evolution. For example, the international division of labor and colonial expansion spurred railway construction in Latin America, Africa, and South Asia.

Despite achieving relatively comprehensive progress in systematically organizing and imputing missing data for railway operating mileage, this study has several limitations. On the one hand, constrained by differences in historical statistical systems, data for some countries in the early years suffer from missing values or inconsistent definitions. On the other hand, in disaggregating data for countries undergoing border restructuring, static and dynamic weighting methods rely primarily on linear assumptions, making it challenging to fully capture nonlinear evolution in actual historical processes. Additionally, some imputation methods demonstrate insufficient adaptability when faced with limited samples in extreme years. Future research could further introduce machine learning and multisource data fusion methods and deepen the linked analysis between railway mileage growth and socio-economic variables. These efforts would promote the construction of a cross-scale, interdisciplinary research framework for infrastructure evolution.

Author Contributions

Li, Y. L. contributed to the data collection and manuscript writing; Yu, Y. H. contributed to the data processing and visualization; Cai, J. proposed the research idea. All authors contributed to the development of the research idea, content review and revision, as well as guidance on statistical analysis.

Conflicts of Interest

The authors declare no conflicts of interest.

References

[1]        Mitchell, B. R. International Historical Statistics, 1750?C2010 [M]. London: Palgrave Macmillan, 2013. DOI: 10.1057/9781137305688.

[2]        Li, Y. L., Yu, Y. H., Cai, J. Global dataset on the length of railways in operation (1825?C2021) [J/DB/OL]. Digital Journal of Global Change Data Repository, 2025. https://doi.org/10.3974/geodb.2025.08.07.V1.

[3]        GCdataPR Editorial Office. GCdataPR data sharing policy [OL]. https://doi.org/10.3974/dp.policy.2014.05 (Updated 2017).

[4]        World Bank. Rail lines (total route-km) [DB/OL]. https://data.worldbank.org/indicator/IS.RRS.TOTL.KM.

[5]        Cohen, J. Private capital, public credit and the decline of American railways, 1840?C1940 [J/OL]. Journal of Transport History, 2010, 31(1): 42?C68. DOI: 10.7227/TJTH.31.1.4.

[6]        Satya, L. D. British Imperial Railway in Nineteenth Century South Asia [M]//Nayak, G. The Railways in Colonial South Asia. London: Routledge, 2021: 85?C113.

[7]        Li, C. J. The great achievements and future prospects of China??s railway development [J]. SASAC Report, 2019(6): 26?C29.

[8]        Scott, M. A. Transcontinentalism: technology, geopolitics, and the Baghdad and Cape-Cairo railway projects, c.1880?C1930 [D]. Newcastle: Newcastle University, 2018.