Mobile phone data to study the impacts of a severe local flood and social events on human mobility (in Switzerland)

Introduction

This Zenodo repository contains datasets related to the detected and inferred positions of individuals within a study area (approximately 15 km by 20 km) surrounding the city of Zofingen, Switzerland. The data provide aggregated and anonymized positions of individuals at a fine-grained level, corresponding to road segments or railway tracks, and span a 44-day period from June 17th to July 30th, 2017. This period includes a severe local flood event as well as social events. The datasets also include the postal codes of individuals' municipalities of residence who traveled to the study area. The data were acquired by the University of Bern from Swisscom, at a total cost of 32400 CHF (around 35796 USD as of December 2024). The related peer-reviewed research article will be available soon.

Network components

The network topology within the study area surrounding the city of Zofingen, Switzerland, is defined by two JSON files: "nodes.json" and "edges.json". These files respectively contain the attributes of all nodes and edges within the study area.

The JSON file for the nodes is an array of nodes that contains the following information:

internalID: This is the unique ID used internally by Swisscom to identify the edges.

externalID: This is the unique ID used by the data source from which Swisscom retrieved that particular node. For example, if the node coincides with a node from OpenStreetMaps, then this will be the ID used by OpenStreetMaps.

nodeType: This indicates the data source from which the edge was retrieved. Usually OSM for OpenStreetMaps.

lat: The latitude as per the WGS84 coordinate system

lon: The longitude as per the WGS84 coordinate system

The JSON file for the edges is an array of edges that contains the following information:

start: The node internalID corresponding to the beginning of the edge. An edge has no particular direction and Swisscom arranged them such that the starting node ID is smaller than the ending node ID.

end: The node internalID corresponding to the end of the edge.

edgeType: The type of infrastructure represented by the edge, typically road or rail. Swisscom also uses a virtual type that we use to connect the road to the rail network.

path: A list of node internalIDs that are present along the edge. Indeed, an edge can be quite long. This is the case for example for highway portions between two entries/exits. Even though Swisscom does not compute metrics down to the intermediate nodes, they use them to display smoother paths on maps.

POSACT and PATHACT datasets

These datasets provide estimated counts of people observed along road and rail networks, within a roughly 15 km by 20 km area surrounding the city of Zofingen, Switzerland. These estimates rely on the Swisscom's market penetration and are calculated using two methodologies called as the datasets (i.e. POSACT and PATHACT), and explained in the "explanation_of_POSACT_and_PATHACT.pdf" file. To safeguard the user privacy, estimates for edges with less than 20 detected users are not included (missing data for an edge is assumed to represent less than 20 observations).

Both POSACT and PATHACT datasets have the following headers:| edgeStart | edgeEnd | hourOfDay | estimatedCount |

An explanation of the headers:

edgeStart: The node internalID of the beginning of the edge.

edgeEnd: The node internalID of the end of the edge. Together with the internalID of the edge start, this uniquely identifies an edge.

hourOfDay: The period of aggregation, which is 4 hours for POSACT and 1 hour for PATHACT. The number shown here is the beginning of the aggregation period. For example, if the aggregation period is 4 hours, the number 0 corresponds to 0h00 -> 3h59, the number 4 corresponds to 4h00 -> 7h59, and so on.

estimatedCount: the Swisscom estimate for the number of people detected close to, or passing by, that edge.

PLZs dataset

This dataset provides estimates of the number of users traveling from their municipality of residence to the local area of study.

The "plzs" dataset has the following headers:| municipalityID | locationName | hourOfDay | estimatedCount |

An explanation of the headers:

municipalityID: Postal code used for the municipality of residence.

locationName: The city name associated with the zip code, provided just for reference.

hourOfDay: The period of aggregation, which is 1 hour for PLZs.

estimatedCount: Estimate of the number of people who travel from their municipality of residence to the local area of study.

Disclaimer

The information presented here on Zenodo is based on the details originally provided by Swisscom in their dataset package. The dissemination of this information on Zenodo strictly adheres to the contractual agreements in place between Swisscom and the University of Bern, as well as to the explicit consent granted by Swisscom via email, dated January 31, 2024. This consent specifically authorized the publication of this dataset package, including its data descriptions and explanations.

Contributions of the authors

S.L. conceived the idea of using mobile phone data, initiated and established the collaboration with Swisscom, uploaded the data to Zenodo, and wrote the data descriptions (based on the information provided by Swisscom). M.K. and A.Z. secured funding to support the data acquisition efforts.

Corresponding person

Simone Loreti