Mapping how railroads built America
A new look at antique US railroad maps reveals how cities grew over the past 200 years. The FT's Alan Smith and Steven Bernard trace how cities, people and the economy spread from coast to coast
Produced by Tom Hannen. Featuring data from the HISDAC-US Data Archive.
You can enable subtitles (captions) in the video player
Welcome to Maptastic. Now you might think that you know the US. But we're going to look at some maps that show it in a completely different light, looking at some new data that shows us how the US has built up over the last two centuries from East Coast to West Coast.
Now cartographers love nothing more than new data to put on maps. And today, I'm with my colleague, Steve Bernard, who is our master cartographer at the FT, to look at some exciting new data recently released by some academics at the University of Colorado.
And the first thing that I did with this data was load it up quickly into our computerised mapping software, our GIS. And I zoomed in to Phoenix, Arizona, just to have a little look at what the data says about how Phoenix has developed over time.
And as you can see, in the early 20th century we start to see these little black dots, which are these individual pixels showing that that land has been built up, right? And as you can see, as we go through the 20th century, you get this amazing picture of sort of growth.
And this city structure, the city shape of Phoenix, really starts to come out. And massive expansion in recent years. So this was a very crude first attempt at just looking at this data and thinking: does it offer us anything useful? Now I showed it to you, Steve. And you got similarly very excited about it.
Yeah, the first time I saw it, I was looking over your shoulder at this. And I just thought, this is going to be great to do for the whole of the US.
And this is all of this data showing how the US has been built up since 1810 running as an animation now. It's almost impossible to see to start with because you have these very small little yellow pixels starting to light up on the East Coast. What just happened there?
So this is the railways flashing up as they were developed. I got the data from the Library of Congress, which had these amazing old maps, which they had digitised. And I took these into the mapping software and essentially traced them, every single railroad from 1830s to 1890s.
And I needed to do them in stages so I could show the development of the railroad from the East Coast all the way to the West Coast.
So the yellow dots are the populated...
The yellow dots are the populated areas.
And the railroads are what you've just flashed up on there. That's interesting by itself because now we're looking at two different data sets. That's one of the powerful things about this GIS, this computerised mapping software, is the way that you can layer information on top.
So this might be a story that people are familiar with. But this is the first time that they've seen it.
See it actually growing, yeah. Then there's this whole pivotal moment in the late 1860s when the first railroad connected the West Coast to the East Coast. And then you'll see the explosion of population along the West Coast from that point on. San Francisco, Los Angeles, further north into Seattle, they are growing exponentially from that point on.
And the amount of railroads which are developed is phenomenal. Sort of unrelenting for the 60 years or so from the 1830s to the 1890s.
When people think about travelling around the US now, invariably people think about internal flights. But back in this period, the railroad, not only was it essential for getting about, just from looking at these maps, we can see that the railroad actually helped to define the geography of the US.
So all of the areas that are building up are building up based on the connections that the railroad is making. I mean, one of the things I thought was very, very interesting looking at some of these old maps was just how important it was for people to have maps of these connections.
I mean, some of these maps here are really, really beautiful maps. But they're all talking about the connections. This map here, Baltimore and Ohio Railroad, is actually titled Shewing the Connection, this kind of joining up of the cities. It was an incredibly important part of America's economic expansion.
This one here, the map of the Canal and Railroads of the US, it's almost like a still from your animation.
It was painstaking tracing all of these because obviously each map was slightly different at slightly different projection, which means I had to what's called georeferencing, which basically turns this map into a rubber sheet and allows me to stretch it inside the mapping software to get it exactly into the projection that we're using in the final animation.
After 1956, the US highways decided to build what is now known as all the interstates in the US and which again just enhanced the connections between the cities and made it easier for people to mobilise and...
We're seeing highways on the animation where although it's still allowing us to go east to west, it's actually connecting up more of the country. So the interstates are going to places that not even the railroads necessarily were going.
So again, another layer of geography telling us something a little bit more about how the US was developing at that time. The really interesting thing at this stage as well, as you can already see, this is so looking at 1970, just that difference in density between the east and the west.
And then that kind of the open road is really sort of the Midwest and out west. You can see that. And they look very, very sparse and isolated, the roads. Whereas a lot denser connections over on the East Coast - very, very stark.
A lot of that has to do with farmland areas in the Midwest, which a lot of the area is given over to agriculture as opposed to large cities. So this is just showing exactly what areas of land have been developed since 1810 through to 2015.
I added another layer showing when each parcel of land was actually developed.
So before, the yellow just meant it's been built on.
It's been developed, yeah.
Now the colour is showing us when. OK, so let's just interpret that colour ramp a little bit for us here. So the very earliest settlements are coloured yellow. So we can see that's over on the East Coast there. Massachusetts lighting up, for example.
Then it becomes red, orange-red, and through to purple. So the more purple you see on the map that's recent development.
Yeah, so a lot of the southern states, you can see, are a lot more purple than obviously the East Coast, Massachusetts, Boston, New York. But still some of the orange on the West Coast where Los Angeles, San Francisco, were connected by those early railroads.
Just it's worth pointing out there is still a lot of this data set - that uncertainty is still a part of any big data set. But this one in particular, there's still big chunks represented by this teal sort of colour there.
Lots of a bit... Michigan.
I guess some ambiguity. I don't think that stops us thinking that this is a very interesting way of looking at the US. But it's kind of important to know that no data set is perfect even when you put it on a map. The thing that's fascinating for me about using these computerised mapping systems is that you can zoom in and zoom out freely.
And so we're going to go live into the GIS now to have a look at some of this data.
So this is the application I used to create the animation and the maps is called QGIS, which is an open source software, which anyone can use for free. One thing I looked at initially was 40 metro areas, which had populations over a million, and seeing how they had changed individually over the 205 years that we have in the data set.
Chicago just grabbed my attention immediately. A, because it's sort of a famous city that everyone's fairly familiar with. But the way that the city has grown in those early years, you can really see very clearly, if I just turn on the railways over the top, how the early developments and settlements were following along the railroads.
So this is using the same colour scheme that we were looking at before. So the yellow areas are the older areas, the purple being more recent. So that is fascinating. You get this kind of...
Sort of spidery.
A spider's web sort of structure building out of the centre of town. That's instantly recognisable. Absolutely. Where else was interesting?
So what we're going to see here is going to be the population growth from 1810 to 2015 in the downtown area. It's the whole metro area of Phoenix-Mesa because the census data doesn't cover that whole area because obviously it's expanding all the time.
The map might show us when it was built. But it doesn't actually show us an awful lot about what's going on there. The population data is going to show us actually when did the people arrive.
Yeah, exactly. Well, you'll get a sense obviously in the beginning. As in your map, there's not a lot going on...
Dormant period for Phoenix.
The odd thing pops up in the late 1800s. But after that, it's just an...
Now we start to see it go, yeah.
Carefully at the population map at the bottom chart, that's at the bottom. It really does grow exponentially in the past 40 years or so.
So the really interesting thing looking at animation is actually there was a lot of growth in the built area before you started to notice a real increase in the population. And in fact, most of the area had been built on before the population density presumably really ramps up.
So that's fascinating, being able to see the sort of geographical spread of the city relative to the sort of density of population living in that area.
The population has pretty much tripled since the 1970s, which is quite an explosion.
There are other cities as well...
There are indeed.
...worth looking at. And in fact, the population charts that you've created there are really, really interesting because one of the things we see when we look at all of the US, those major metro areas, is that there's an interesting pattern repeated across several cities.
Indeed, yeah. A lot of the new cities like Phoenix, San Antonio, San Diego, San Jose, they're the success stories of this urbanisation. But there are also some of the larger cities that are being left behind. If you look at this set of charts here, we're basically showing cities in red that have seen a decline in recent years.
And there is none more dramatic than Detroit, which was at one point the fourth most populous city in the US as recent as the 1940s. And is now seeing their population drop from 1.8m down to just over 700,000 at the last census.
That's a huge proportionate drop and in absolute terms a big drop. The interesting thing looking at the chart there is that decline has been going on for a long time. So the peak was back in the middle of the 20th century. And there's been a long-term decline since then. How can we map that back onto the city maps?
Looking at Detroit... well, I've got a map on now is the bottom 20 per cent of income. So the poorest families and also the highest vacancy rates, where they coexist at the same time. And you'll see a big hole appearing in the downtown area of the city.
So these are areas with vacancy rates that are quite high. So a large proportion of sort of empty buildings and deprived areas. So what's really interesting about doing that and showing it like this is that it is almost perforates that map that we have just built up.
So you have this period of growth and expansion. And then in places like Detroit, it's actually possible to visualise on the map how you... this has not gone back to obviously kind of pristine or virgin land. But it's just...
People have just vacated it. And they haven't come back. This data is 2010 census data. There is a regeneration in Detroit underway. And people are now developing the downtown area, which should attract more people. But it is fascinating to see that a lot of these highly vacant areas and poor areas in the cities across the Rust Belt are predominately in the older part of town.
If you look at the orangey, this whole area is basically where most of the deprived and poor areas are. The newer, more recent urbanised areas are not so badly affected. And that's the story which is true across a lot the Rust Belt cities. So here we have Pittsburgh and Cleveland, Columbus, Indianapolis.
So if you look at Cleveland, it's a very similar story. In the older part of town, the poorest and most vacant areas are in the older parts of the city. It could be due to it being too expensive to develop these old buildings.
So the pattern that you just mentioned that is really consistent is that those yellow areas, which are the older areas, are the ones that are more likely to be zapped by our perforated data. They're the ones that are the oldest. But they are also, in some senses, the most challenging for policymakers today to deal with.
And in fact, that leads me back to the value of these maps in general because just as these old maps were so incredibly valuable during that massive expansion period in the 19th century as the US built out westwards, these maps are equally valuable to people who are making decisions today about what to do.
So you can imagine if you were a city planner in Cleveland or Detroit, this view of how the cities developed and the way that it is now is not something we've been able to do before. But this is not exactly what you'd want to see if you were looking to make interventions to turn things around.
That's the good thing about sort of seeing, taking in, say, this old data in terms of the historical value to it but also being out to map new data sets on top of it to sort of get a lot more value from it.
And I guess the other thing with 200 years' worth of data here more or less, you can slice through that as well to really understand more recent things. So if we ignore the two centuries of change, we can do things like use this data to say, right, show me the only... only show me the areas that have been developed in the 21st century. So where is America growing right now?
This is since 2000.
So now the yellow pixels that we're seeing are parcels of land in the US that have been developed...
...for the first time since 2000. That is really fascinating. Can we see any patterns at a city level in here? I mean, how are some of our old friends like Phoenix faring on this data? So for Phoenix, this looks like a story of continuing growth, just looking at the extent of land that's showing up here.
Absolutely. Look at that. That's really fascinating. So this general pattern that we've seen as cities grow is still happening in Phoenix. There's still big expansion outwards. Now there's data linked behind this map. That's what this software is doing it's using the data that we've got to draw the maps for us.
Can we actually see what the data is telling us there?
So our magic inspector tool.
OK, so this is the tool that when you click anywhere on the map, it's going to tell you what data is behind that point.
To go to the exact layer.
There we go.
So this will tell us where we click on the map, what year it was first developed.
And so this data set works in five-year age bands, if I recall correctly. So it's not something that's...
It's not day to day, month to month. It's every five years. Steve, people might be looking at this video and seeing you having so much fun clicking and enjoying and exploring this data. How do people get started if they want to play with this themselves?
Well, the data set is freely available from the Harvard website. So you can download it. And you'll be able to have the same starting point that I did. The first thing you just need to get hold of is QGIS, which is an open source software, which allows you to take this data in and map it immediately.
If you want to learn QGIS, I've actually done this series of 31 videos, which allow you to sort of slowly get to grips with how to use the software. It's quite a vast piece of software. But you normally only need to use probably about 10 per cent of it to get to the stage where you can map something like this. You don't need to know everything about it.
The Library of Congress maps are free to download. Yeah, the highways... the Census Bureau has so much data on it. It's a vast resource population data, deprivation data, vacancy data. All of this is freely available on the Census Bureau website.
Fantastic. Thanks, Steve.