Steven Cramer

Where can you fly? (2019 data)

I knew that New York was well connected to Europe. I knew that LA was a gateway to Asia. I'd read it in airline industry publications, seen it referenced in news articles. But I'd never seen it.

The time was 2021, the middle of the COVID lockdowns. I hadn't traveled anywhere in a while. I love planes, I love thinking about how they connect the world, and I wanted something to stare at. I discovered that the US government publishes data about every commercial flight that touches a US airport, and I turned it into an airline-agnostic route map of where people could fly in 2019.

Each line means at least one large plane per day worth of seating capacity flies between the two airports. Thicker lines mean a stronger connection: bigger planes or more planes per day. The top 50 connections are in red, the rest are in blue. See the full map here.

The data

The US government publishes data about where commercial planes fly when those routes touch the US.

This data match takeoffs and landings when one of those occurs in the US. It does not show what a person intends to fly. Consider three ways of getting between Chicago and Paris, and how that appears in the data:

  1. Chicago ORD -> New York JFK -> Paris CDG, the data records two flights: one for ORD->JFK, and one JFK->CDG.
  2. Chicago ORD -> Frankfurt FRA -> Paris CDG, the data records one flight: ORD->FRA. The flight from FRA->CDG is not included because it doesn't touch the US, even if it was purchased on one ticket.
  3. Chicago ORD -> Paris CDG, the data records one flight: ORD->CDG.

While airline marketing departments have taught us to think of these flights as the same because they all start in Chicago and end in Paris, this reality of routing matters: beyond the obvious differences like scheduled travel time, some of these routes are more resilient to bad thunderstorms or overrunning meetings than others. This data reflects the deep-pocketed possibilities of a savvy passenger in our air network, not the marketing-department illusions.

Also, the data reflect seats available rather than passengers flown. So if a 300 person plane flew between ORD and JFK once, it wouldn't matter in this data whether it was fully packed with 300 people or virtually empty with 3. We're mapping deep-pocketed traveler possibilities, not airline profitability.

Visualizing the data

While a Wikipedia airport page shows what routes are possible, we focus on which routes are common and plot it on a map.

Commonality matters. For example, it's fun that one can fly direct from NYC to Singapore, but only 58,443 seats made that trip in the entire year. That's enough for one (small) planeload per day. Conversely, 1,800,000 seats flew between New York JFK and London Heathrow in that same year, enough for one (big) planeload per hour between 6AM and 9PM every day of the year. We want to see that there's a sky train between NYC and London, and we'll leave that direct flight to Singapore for the trivia books.

Maps matter. We see that some regions are better connected than others, and maps allow us to quickly see that.

First takes from the top 50 air routes

On the map, the top 50 routes are shown in red.

New York - London outranks almost every domestic route. More seats fly between London LHR and New York JFK than all but three US airport pairs (LAX-SFO, JFK-LAX, LAS-LAX).

Western Europe and East Asia are well connected. Large international hubs like Paris, Frankfurt, Amsterdam, Tokyo, Hong Kong, and Seoul all have thick lines to multiple American cities.

Our connections south are thin. This is partially an artifact of how we project maps, but our routes to South America, Central America, Africa, and Australia are strikingly sparse compared to the dense web going east and west.

NY goes to Europe; LA goes to Asia

Clicking a line on the map highlights all routes from both endpoints, which lets you see what each airport specializes in. Airline industry coverage often talks about "gateways", and this lets you see them firsthand.

New York JFK is a gateway east. Dense connections deep into Europe, but comparatively fewer to Latin America and Asia.

Los Angeles is a gateway west. Much better connections to Australia, Asia, and Hawaii, with thinner lines to European second cities.

Gateways north and south

Using the word "gateway" is more common with airlines as they talk about their hubs away from the top world cities like NYC and LA.

Miami is a gateway south. American Airlines calls Miami "its primary gateway to Latin America and the Caribbean." The map confirms it: connections to Latin American second cities that even JFK doesn't have, fewer lines to Europe, nothing to Asia.

Seattle is a gateway north. The airport brands itself as "The Gateway to Alaska," and sure enough, Seattle–Anchorage is a top-50 route. (Delta also built Seattle into an Asia gateway over the past decade, and you can see those routes too, though they're still much smaller than LAX's.)

Conclusion

I made this map because I was stuck at home and missed the world. I ended up staring at it for hours not to plan any trip, just to watch the lines and think about where they go.

There's a kind of knowledge that only comes from looking. I could have told you before I started that JFK was a European gateway and LAX was a Pacific one, but it's different to see it.

Appendix: Making the map

The hard part of this was finding the dataset.

The underlying data comes from the Bureau of Transportation Statistics T-100 Segment (All Carriers).

The key fields are:

Then I essentially ran a SQL query to get seats between airports...

select ORIGIN, DEST, sum(SEATS)
from t_100_table
where ORIGIN < DEST -- only get one direction of the flights 
group by ORIGIN, DEST 
order by 3 desc

...joined that result with the coordinates of different airports, and turned it into a line for plotting in JavaScript.

I then limited it to only show the routes that have as many seats as between Chicago ORD and the small commercial airport closest to where I grew up.