Capturing Cross-Border Mobilities of People:
A Twitter study
Håvard Wallin Aagesen, Olle Järv, Ate Poorthuis
Hi, my name is Håvard Wallin Aagesen, and I am a PhD candidate the Norwegian University of Life Sciences (NMBU) and the Digital Geography Lab at the University of Helsinki.
Today I'm going to present a paper that is still work in progress, so any feedback is highly appreciated.
I am going to talk about my work on cross-border mobilities of people using Twitter data, that I'm doing together with Olle Järv from Helsinki and Ate Poorthuis from KU Leuven. And although this might not necessarily be specifically urban mobility, as this session is called, but I think the approach that we are using is applicable in different spatial scales as well.
Mobility &
150 million people live close to borders
Border regions mostly omitted in spatial research
Country-specific, as silos
Less attention on people beyond migration
And in the European context, these cross-border regions are mostly top-down defined.
Knowledge Gap
Who crosses borders & why?
Where & when borders are crossed?
How (un)expected events like COVID-19 influence?
How mobility of people form functional border regions?
Lack of data
And this leads to the knowledge gap that we are focusing on; trying to understand who crosses borders and why, where and when borders are crossed, how unexpected events like COVID-19 influence border regions, and how the mobility of people form functional border regions.
And one of the reasons for this knowledge gap, is the lack of data on cross-border mobility, especially what can be characterized as frequent mobility. And there are many reasons for that; national bodies are mandated to collect data on what happens within the country, and the sharing of data between countries have proven difficult and GDPR-regulations also made it more difficult. So that's why we using social media data, as it is a source of data that is not limited by borders.
First studies indicate the feasibility of the approach
So we have already used our approach in two publishd papers. The first paper focused on the Greater Region of Luxembourg, which is a small area with a lot of frequent commuting across country borders. and then the second paper focused on cross-border mobilities in the Nordic countries in light of the COVID-19 pandemic, where we tried to capture how the pandemic affected mobility across the borders. Both studies show that it is possible to use mobile big data to study cross-border mobilities of people.
Scaling up the research
Examine the feasibility of the approach to study border regions at European level
Characterize border regions from the perspective of C-B mobility
So what we are doing now, is that we are scaling up the analysis to the European level.
Methodology
Data
Geolocated Tweets in Europe
2012-2022
Ca 14 million users
Ca 4 billion Tweets
So what do we actually do?
Data
Geolocated Tweets in Europe
2012-2022
Ca 14 million users
Ca 4 billion Tweets
Movement detection
Max duration 45 days
Max distance 300km
~ 3 million C-B movements
And then we detect movements based on the geolocated Tweets by connecting two consecutive Tweets with a maximum duration of 45 days and a maximum distance of 300 km. This is to ensure that the data represents movements in the border areas, and not long-haul flights, and to try and make sure a level of somewhat frequent mobility. At this point we group all the movements into pairs of neighboring countries, so only movements that cross the border to a neighboring country. So we ended up with around 3 million cross-border movements. Which we then use to analyse temporal rhythms and spatial patterns.
Characterizing border regions
First off, we look at temporal characteristics and how the country pairs cluster. And so here our data clusters into four clusters, ranging from more dominated by weekend mobility, so often leisure related, to more weekday mobility, often work-related commuting and so on.
Characterizing border regions
Visually, it looks like this. So here we see patterns of country pairs that are characterized by more weekend/leisure related mobility, like in the Baltics and Andorra, and to some extent the Nordic countries here, except from where we are now, between Denmark and Sweden, which is characterized by more weekday(work-related) mobility.
Example:
So to give you an example of the process, here you see the border region between Denmark and Sweden.
And when we combine them we are left with the functional area of the border region, so where people actually move within the region. So the top 50% of the movements are colored in purple, and the top 25% in dark blue.
And then we are now doing this on the European scale, combining all these cross-border functional regions and country pairs. To see where they are, and how they differ from the top-down defined border regions.
Take home messages
Mobility approach is feasible to characterize border regions in Europe
Provide the dynamic perspective of people
Simple and robust methodology, but how reliable are the findings?
So to sum up, we have shown that the mobility approach is feasible to characterize border regions in Europe. We have developed a simple and robust methodology, but we still need to assess how reliable the findings are. Especially on the European level, we have been able to confirm the reliability for some regions, but still need to compare to ground truth for more regions.
Future steps
Adding place of residence and directionality
Comparing with socio-economic factors
Implications to policy and planning of border regions
Some of the things we are looking to advance going forward, is to look into adding place of residence of the users. We are developing methods to do so, but have not run this for the current dataset yet. And we also want to add directionality to the movements, so that we amongst others can compare the regions with socio-economic factors, and then how this links to policy and planning of border regions.
Resume presentation
Capturing Cross-Border Mobilities of People: A Twitter study Håvard Wallin Aagesen, Olle Järv, Ate Poorthuis Hi, my name is Håvard Wallin Aagesen, and I am a PhD candidate the Norwegian University of Life Sciences (NMBU) and the Digital Geography Lab at the University of Helsinki.
Today I'm going to present a paper that is still work in progress, so any feedback is highly appreciated.
I am going to talk about my work on cross-border mobilities of people using Twitter data, that I'm doing together with Olle Järv from Helsinki and Ate Poorthuis from KU Leuven. And although this might not necessarily be specifically urban mobility, as this session is called, but I think the approach that we are using is applicable in different spatial scales as well.
havard.aagesen@nmbu.no
+
haavardaagesen
