Data from 130 million commuters reveal US ‘megaregions’
- Maps published in PLOS ONE using five-years of commuting data reveal geography of US megaregions
- New approach to age-old empirical problem of how we define geographical regions
- Research has potential use in transit planning, infrastructure development, real estate and retail analysis.
The daily commutes of more than 130 million Americans have been used to identify commuter-based megaregions in the United States for the first time, a new paper published in PLOS ONE revealed today (30 November 2016).
While the division into 50 states is how many think of the US, geographers have for the last 50 years also studied networks of closely connected metropolitan areas, known as ‘megaregions’, which often cut across state lines.
Previously megaregions have been typically identified by an interpretative method that links large metropolitan regions through similar environmental and infrastructure systems, economic links and cultural similarities. These approaches are often based on a ‘best guess’ kind of approach, and do not rely on the analysis of large datasets.
Now Dr Alasdair Rae and his co-author Dr Garrett Nelson have developed an empirical approach to identify megaregions using a dataset of more than 4 million ‘commuter flows’ involving the travel to work patterns of 130 million Americans.
The data comes from five-years worth of data from the American Community Survey between 2006 and 2010. The yearly nationwide survey of 3.5 million employees asks where they worked ‘last week’.
Using algorithmic ‘community partitioning’ software developed by the Massachusetts Institute of Technology (MIT) and cloud computing powered by Amazon Web Services, these commuter flows were mapped out and revealed massive labour market areas across the US that form distinct megaregions.
Dr Alasdair Rae, Senior Lecturer, said: “We often know how countries are divided up geographically; into officially defined states, regions or cities, but this doesn’t normally reflect underlying economic linkages.
“Garrett and I therefore wanted to explore the economic geography of the United States from a functional economic perspective, to arrive at a different understanding of how the country functions in relation to journeys to work - a significant foundation of any economy.”
A visual interpretation of the data was also carried out to compare it against the algorithmic findings. Whilst this interpretation discovered similar megaregions were identifiable at a national scale, it was not sufficient for real-world applications where statistical accuracy was required.
“We first mapped more than 4 million commuter links and then explored these visually and then using more complex algorithms. Our aim was to demonstrate how the country functions from an economic point of view, but also show that you can get different results depending upon the choices you make - put simply, the algorithm isn’t always right.”
Dr Rae added: “In addition to identifying broad US megaregions, we also conclude that there is no replacement for a common sense interpretation of any results generated through computational approaches. We believe the megaregions we identify are a true reflection of the economic geography of the United States but of course they need to be tested and validated in the real world for them to have real use.”
It is hoped that the research will provide a basis for further empirical approaches to studying megaregions and allow policymakers in areas including infrastructure, transit planning and electoral geography to evaluate how they approach the limits areas.