railway_delays — Italian and Chinese railway networks with delays (2020-2024)

Description

Railway networks of Italy and of the Chinese high-speed railway, compiled by Faccin, Scagliarini and De Domenico (CoMuNe Lab) to study how train delays respond to rainfall. Nodes are railway stations, and an undirected edge connects two stations between which trains run without intermediate stations of the network. The station locations and the shapes of the lines were extracted from OpenStreetMap, and combined with train schedules and reported delays: for Italy, from the TrainStats service (trainstats.altervista.org), with the years 2021-2023 used for model fitting and 2024 for testing; for China, from the dataset of train operation records and weather data of Zhang et al. (2022). Each station was assigned the population of its Voronoi cell (from WorldPop estimates), and a precipitation value from the ERA5-Land reanalysis of the Copernicus Climate Change Service. The coordinates are longitude and latitude in WGS 84 (EPSG:4326/CRS84).

italy: the Italian railway network. The upstream edge list contains 14,572 directed edges, which list both directions for 6,016 pairs of stations and a single one for 2,540 pairs; as in the authors' code, they are merged into 8,556 undirected edges. For 13 of the pairs listed in both directions the two geometries differ, and the length is that of the first one listed. Since edges join consecutive stops, some of them (e.g. of long-distance trains) connect stations hundreds of kilometers apart. The network has two connected components (the largest with 2,367 stations).
china: the Chinese high-speed railway network, from the GeoPackage of the repository (the station metadata from its separate GeoJSON file). The edges were obtained by merging the OpenStreetMap ways of the lines between stations; the edge properties below are OpenStreetMap tags of these ways, one value per edge as given in the data (other tags, mostly sparse names in various languages, are left out). Three pairs of homonymous stations were split by the authors into <name>_0 and <name>_1 (Jingzhou, Fengchengdong, Taian); for these, the english_name, city and province are those of only one of the two stations. The degree attribute of the node data (not equal to the degree in this network) is left out. The network has two connected components (the largest with 702 stations).

The time series of the delays per station and the counts of train transitions per hour, also in the repositories, are not included.

The networks have the following properties:

name: the station identifier (lower-case Italian name, for italy; pinyin name, for china).
pos: the position of the station (longitude, latitude).
pop: the population assigned to the station.
capacity: the maximum number of trains passing through the station in one hour.
rain: the average rain at the station (ERA5-Land precipitation, as given in the data; the units and the averaging period are not documented). For china, it is nonzero for only 6 stations.
italy only: osm_name, the name of the station in OpenStreetMap; railway, its OpenStreetMap railway tag (station, stop, halt, manual, service_station; empty for 4 stations); region, the Italian region; delay_q10, delay_q50, delay_q90 and delay_mean, the 10%, 50% and 90% percentiles and the mean of the delays experienced by the station (in minutes).
china only: osmid, the OpenStreetMap node identifier; english_name, the English name; name_zh, the Chinese name (empty if missing); osm_name and osm_name_en, the name and name:en tags in OpenStreetMap (empty if missing); city; province.
Edge length: the length of the line between the two stations, along its geometry, in kilometers.
china only, edges: line_name and line_name_en, the name and name:en tags; railway, highspeed, usage, service, electrified, the OpenStreetMap tags of the same names; maxspeed and maxspeed_designed, the maximum speed and the designed maximum speed in km/h (-1 if missing). Empty strings mean missing tags.

Tags

Transportation Spatial Unweighted Metadata

Citation

M. Faccin, T. Scagliarini, M. De Domenico, "Systemic collapse emergence from climate-induced perturbations in transportation networks." Proc. Natl. Acad. Sci. U.S.A. 123(42), e2532180123 (2026) 10.1073/pnas.2532180123 Sci-Hub
D. Zhang, Y. Peng, Y. Xu, C. Du, Y. Zhang, N. Wang, Y. Chong, H. Wang, D. Wu, J. Liu, H. Zhang, L. Lu, J. Liu, "A high-speed railway network dataset from train operation records and weather data." Scientific Data 9, 244 (2022) 10.1038/s41597-022-01349-8 Sci-Hub

Upstream URL OK

https://github.com/CoMuNeLab/adaptsys_train-delays
https://github.com/CoMuNeLab/adaptsys_chinese_train_delays

Upstream license

MIT License (Italian network); none stated (Chinese network), https://github.com/CoMuNeLab/adaptsys_train-delays/blob/8104d02cd15560dd1eccdd572018ca5c6d28757d/LICENSE

Networks

Tip: click on a column header to sort the list, or hover over it for a description.

Column legend

$\left<k\right>$: Mean degree
$\sigma_k$: Standard deviation of the degrees
$\lambda_h$: Largest eigenvalue of the Hashimoto (non-backtracking) matrix
$\tau$: Random walk mixing time
$r$: Degree assortativity
$c$: Global clustering coefficient
$\oslash$: (Pseudo-) diameter
$S$: Relative size of the largest component
NPs: Node properties
EPs: Edge properties

Name	Nodes	Edges	$\left<k\right>$	$\sigma_k$	$\lambda_h$	$\tau$	$r$	$c$	$\oslash$	$S$	Kind	Mode	NPs	EPs	gt	GraphML	GML	csv
italy	2,415	8,556	7.09	7.24	20.67	310.62	0.19	0.29	16	0.98	Undirected	Unipartite	name pos osm_name railway region pop capacity rain delay_q10 delay_q50 delay_q90 delay_mean	length	213 KiB	274 KiB	269 KiB	254 KiB
china	721	2,404	6.67	4.92	10.90	202.52	0.04	0.44	16	0.97	Undirected	Unipartite	name pos english_name name_zh osm_name osm_name_en city province pop rain osmid capacity	length line_name line_name_en railway highspeed usage service electrified maxspeed maxspeed_designed	80 KiB	102 KiB	99 KiB	96 KiB

Ridiculograms^*

^* These are automatically generated force-directed visualizations, and can be quite meaningless for networks both big and small. They should not be taken seriously as sources of scientific insight. See here for a discussion.