Study Based on 4 Million Commutes Draws Up New U.S. Megaregions

An ever growing part of the world’s population lives in what are known as megaregions – clusters of interconnected cities. Researchers Garrett Nelson and Alasdair Rae have attempted to map the megaregions of the contiguous United States by studying the commutes of American workers.

A commuter flow-based regionalization of the United States. ILLUSTRATION BY GARRETT DASH NELSON AND ALASDAIR RAE, PLOS ONE

As megaregions grow in size and importance, economists, lawmakers, and urban planners need to determine their boundaries in order to be able to coordinate policy at this new scale. But because megaregions are defined by connections – such as interlocking economies, transportation links, shared topography, or a common culture – they find themselves running into the same problems geographers and cartographers have always encountered when trying to delineate conceptual areas.

To solve this geographical problem, Garrett Nelson of Dartmouth College and Alasdair Rae of the University of Sheffield used census data on more than 4 million commuter paths and applied two different analyses, one based on a visual interpretation and the other rooted in an algorithm developed at MIT.

All commutes of 50 miles or less in the greater San Francisco Bay Area. Shorter, higher volume commutes are in yellow, with longer, lower volume paths in red. ILLUSTRATION BY GARRETT DASH NELSON AND ALASDAIR RAE, PLOS ONE

But where should planners draw the edges of a megaregion encompassing this activity? To answer this question, Nelson and Rae turned to an algorithm-based tool designed by MIT’s Senseable City Lab to mathematically recognize communities. The algorithm only considers the strength of connections between nodes (more than 70,000 census tracts in this case), ignoring physical locations, so the results required some cleanup and iterations – such as eliminating superlong commutes between places like New York City and Los Angeles and excluding nodes with only very weak connections – to produce a coherent map of plausible megaregions. The difference between the visual and mathematical approaches can be seen in the map below of the Minneapolis-St. Paul area.

The Minneapolis-St. Paul megaregion looks different when using a visual approach (left) or an algorithmic approach (right). ILLUSTRATION BY GARRETT DASH NELSON AND ALASDAIR RAE, PLOS ONE

These algorithmic megaregions are easier to interpret using the map at the top of the post, which shows the connections. However, it gives some seemingly odd results like the sharp boundary that follows the New York-Connecticut state line or the small, splotchy green megaregion that floats between Birmingham and Dallas. Clearly, the statistical method didn’t get everything right, so Nelson and Rae combined the two methods to draw the country’s megaregions. They started by drawing lines around the dots on this map:

An algorithm assigned individual census tracts (dots) to 50 megaregions (colors). PHOTOGRAPH BY GARRETT DASH NELSON AND ALASDAIR RAE, PLOS ONE

Then they overlaid those shapes on the flow map at the top of the post and reinterpreted the boundaries to eliminate outliers and account for geographic continuity. The result is the map below, which eliminates some of the visual oddities.

Combining visual and mathematical approaches yielded this map of U.S. megaregions. ILLUSTRATION BY GARRETT DASH NELSON AND ALASDAIR RAE, PLOS ONE

The researchers hope their approach helps us gain a better understanding of the country’s economic geography. Clearly, the map is a work in progress and some areas still don’t look quite right, but we think they are getting there.

Source: PLOS via NG


Please enter your comment!
Please enter your name here