Posts about Maps

The main challenge in the map's design is how to show 3 lines all running together. Until the Silver Line, no track segment had 3 lines. When there are 2 lines, the map shows a small dot in between the two. But what to do with 3?

In our 2011 contest, people tried a lot of solutions, like much thinner lines (like most transit systems), striped lines, pairs or triples of dots, or just bigger dots and much more.

Metro's first draft used little "whiskers" on each side of the circle. A few people liked them, but most hated them and pushed for "pill" or "capsule"-shaped station symbols instead, or thinner lines.

Metro now has a new version that incorporates those suggestions. It shrinks the line width by 24%, which still leaves fatter lines than in other transit systems, but much slimmer than the current map. In this option, the stations with 3 lines now use the "capsule" shapes. They also created a new version that keeps the "whiskers" but cleans up the map in other ways.

What works, and where there could be a few more tweaks

The capsule version is much better than the previous versions. The curves are very tight and clean. The thinner lines look better, and the capsules are superior to the "whiskers."

It seems to me that for consistency with the circles, the capsules should be as large on the rounded ends as the small circles are today—basically pulling 2 halves of one of the current circles apart and putting a rectangle in between. Or as Cameron Booth put it, "an elongated 'pill' symbol with the same cap radius and the normal circle." That would make them much larger in overall area, though. On the latest draft, they're much thinner. What do you think?

Booth had very harsh words for the current map (redesigned last year). He pointed out many technical errors, like the way the parking P and hospital H icons didn't line up with the text at all. Metro has corrected at least some of these, like one Booth pointed out:

Metro also abandoned an idea of abbreviating words like "Ctr" and "Hgts" in station names (another choice Booth panned), but they are abbreviating "Rd," "St," "Ave," and "Blvd" for all stations. The original map abbreviated some but not all road types.

As the map goes through iterations, some have repeatedly pointed out that there is considerable parkland east of the Anacostia, including right along the river, but none appears on the map. Given that even the Pentagon (a large office fortress with parking lots and highways around it) gets to be inside a "park" space on this map, it seems reasonable to put some green along the east bank of the Anacostia.

It continues to mystify why Metro doesn't want to put the "Farragut Crossing" out-of-system transfer on the map. If it did appear, that could entice some casual users to take it instead of crowding trains through Metro Center.

While Metro is adjusting lines a little, it also would be smart to move Metro Center and Gallery Place closer together, so that fewer tourists take the Red Line one stop and then transfer, and put Union Station nearer the Capitol, because it's the station closest to the Senate.

Overall, the capsule map seems best, and the map overall is moving definitively in the right direction. Especially compared to the pre-Rush Plus map, where curves were all uneven, some labels were not even at a 45-degree angle, and everything was just a mess in so many tiny ways, the map has gotten far more professional.

Update/note: The image at the top does not show the legend and other information that's at the top or bottom of the map. You can click on an image to see the full map including header and footer.

Update 2: The original version of this post said that Metro has a new version with capsule station symbols, but in fact they have 2 new versions, one of which has capsule symbols and the other with whisker symbols. I misunderstood the whisker version Metro posted as being the old one rather than a revised whisker one. The post has now been updated to show both new versions as well as the correct previous one and the current map.

Ivey also maps which candidate won the most votes in each precinct.

Ivey also notes that looking at the overall amount of ink for each candidate doesn't necessarily reflect reality. The peripheral areas where Bonds was strongest, for instance, are also less densely-populated areas of the city. He says,

The map can be misleading in the same way typical U.S. presidential election maps are, since the area of a precinct is not proportional to the number of voters there. A candidate who wins in densely populated, high-turnout areas will often look worse on the map than a candidate who wins in less dense or low-turnout areas.

One observation is that you can't really detect Rock Creek Park on the Silverman map. Rock Creek forms a bright line on the other maps, but not Silverman's. On the other hand, the Anacostia River is a bright line on everyone's map.

Their article by Ryan Sibley also shows many other interesting statistics, such as who got money from outside the region, the balance of corporate and individual contributions (Anita Bonds and Michael Brown got only about half individual contributions, while it's nearly 100% for Silverman), and more.

Sibley also notes that while DC's Office of Campaign Finance releases computer-readable data files with contribution information, some data is not in those files, like which candidate goes with a campaign committee. That's in PDFs, but PDF data isn't usable in mash-ups without human work.

What do you notice?

With 9.8 million people, Los Angeles County is larger than North Carolina, the 10th most populous US state. (It's also geographically larger than Rhode Island and Delaware, combined).

Baltimore County (which doesn't include Baltimore City) is more populous than Wyoming, Vermont, North Dakota and Alaska; Prince George's is also larger than South Dakota; Montgomery County larger than Delaware; and Fairfax County has more people than all of those as well as Montana and Rhode Island.

Thanks to Dan Malouff for the tip via Twitter.

This is actually a problem I hear often. A family friend moved to DC a couple of years ago, for a job at PriceWaterhouseCoopers in Tysons. The Silver Line was still a few years off, but he wanted to live in a vibrant, urban neighborhood. Where should he go?

The bus maps are daunting to decipher. It took me a couple of hours to really puzzle through the combinations and cross-reference it with my general knowledge of housing prices in various neighborhoods.

Boston-based AutNo tries to help by putting rental listings and trip planning together in one interface. You can view available rentals (it doesn't have places for sale, yet), click on one, and see transit directions to your office or another location you specify.

The about page reads:

AutNo is the first apartment search designed and developed specifically for people without cars. For the first time since the automobile was invented, the percentage of Americans who drive to school or work is on the decline. Gas prices are skyrocketing and automobile carbon emissions are contributing to global warming. Commuting and living without an automobile is the way of the future for many people. AutNo is dedicated to helping these people find apartments.

It will also show driving routes to work, too, if you want them.

You can narrow down results by price and number of bedrooms. A future feature that would be helpful is to also let people restrict the searches by travel time. That way, you could say that you want a place under $2,000 a month that's no more than a 45 minute trip to work, or whatever.

Basically, combine this with Mapnificent:

And, at the risk of sounding like a broken record: this is why open data is valuable. A transit agency might build a great app, but they're never going to build a mash-up of real estate data and transit data. When it's easy to put transit routing into an app, you not only can build apps that give people transit routing, but tools and apps that combine transit routing with almost anything else.

Update: I hadn't know it, but WalkScore actually has this exact Mapnificent-style feature. You can filter apartment listings by transit distance to a point:

However, when you click on an apartment, WalkScore does not show you the transit routing with trains and buses you would take, while AutNo does. Without that information, people won't as easily learn which buses might work best for them or be able to judge whether a location is really likely as acessible from transit as the system says.

It would be best to have both at once on the same site; as it is now, I'd recommend that people use a combination of both tools for their search.

2. Los Angeles: Despite its reputation for sprawl, LA compares favorably to the densest cities after New York. Its peak density of 94,000 ppsm is well above DC's.

2000

2010

3. Chicago: Home to probably the single densest census tract in America, a 508,000 ppsm anomaly that's so small it's not visible at normal scale. Besides that tract, Chicago tops around at about the same level as LA.

2000

2010

4. Miami: Thanks to more narrowly-drawn census tracts along its high-rise coast, Miami's peak density shot up from 38,000 ppsm in 2000 to 77,000 ppsm in 2010, but the actual change wasn't as significant on the ground.

2000

2010

5. Philadelphia: At 64,000 ppsm, Philadelphia's peak is about the same as DC's, but Philly's rowhouse neighborhoods extend farther out.

2000

2010

6. Dallas: Dallas' density dropped significantly. It has fewer dense tracts in 2010 than in 2000, and its peak is down to 44,000 ppsm from 57,000 ppsm.

2000

2010

7. Houston: Unlike Dallas, Houston appears to be densifying. Oddly, its densest area is not the core.

2000

2010

8. Washington (with Arlington & Alexandria): Washington is has more dense neighborhoods and a higher peak than in 2000. The numbers shown on these maps are slightly different than those on Michael Rodriguez's map, which used a different map projection to calculate area. These census numbers are official.

2000

2010

9. Atlanta: Not only is Atlanta shockingly sparse, its densest tract fell from 41,000 ppsm in 2000 to just 21,000 ppsm in 2010. The explanation? A downtown public housing complex was demolished, erasing the population of the densest 2000 tract.

2000

2010

10. Boston
One of only 4 cities with a tract above 100,000, Boston has a single tract that reaches 110,000 ppsm.

2000:

2010

11. Detroit: Detroit's peak density of 18,000 ppsm is about the same as in 2000, but the number of mid-density tracts in the 10,000-20,000 ppsm range declined significantly as the city continued to empty.

2000

2010

12. Phoenix: Central Phoenix didn't change much, and tops out at 23,000 ppsm.

2000

2010

13. San Francisco: San Francisco has more tracts above 100,000 ppsm than any city except New York. It tops out at 161,000 ppsm.

2000

2010

14. Seattle: With a peak of 51,000 ppsm and a small but significant core, Seattle occupies a middle ground between the older denser cities and newer sparser ones.

2000

2010

15. San Diego: While downtown San Diego densified compared to 2000, and its 50,000 ppsm peak is higher, some of its other denser neighborhoods are sparser in 2010.

2000

2010

16. Minneapolis: Minneapolis' changes were minor compared to most other cities. Its peak was 25,000 ppsm in 2000, and it still is in 2010.

2000

2010

17. Tampa: By far the sparsest city on this list, Tampa's peak of 13,000 ppsm means it has no tracts in the 3rd or 4th categories, and precious few crack even into the 2nd.

2000

2010

18. Denver: Like a smaller Minneapolis, Denver looks much the same. Its peak of 23,000 ppsm is respectable for a mid-sized non-coastal city.

2000

2010

19. Baltimore: Baltimore's lone tract in the densest category is an impressive 86,000 ppsm, but that tract is down from a whopping 176,000 ppsm in 2000. What happened?

2000

2010

20. Saint Louis: Saint Louis' losses have been less drastic than Detroit's, but they still hurt. Its peak is down to a Tampa-like 13,000 ppsm, from 15,000 ppsm in 2000.

2000

2010

I made all these maps using American FactFinder on census.gov, which has data for every county in the United States. I couldn't have done it without Geoff Hatchard, who walked me through the laborious census.gov process. If you'd like to make your own maps, I documented step-by-step instructions. Godspeed.

Cross-posted at BeyondDC.

Thse two maps show DC neighborhood density at the time of the 2000 census (top) and 2010 census (bottom). I made the 2000 map using census.gov sometime after the 2000 census. Michael Rodriguez created the bottom map just recently. Unfortunately the two maps use different scales, but they're still informative.

In 2000 the densest census tract in the DC region was in northern Columbia Heights, between Spring Road and Newton Street. It had 57,317 people per square mile (ppsm). In 2010 that tract is up to 59,209 ppsm, but that's only good enough for 2nd place in DC, and 3rd regionally.

The densest tract is now southern Logan Circle, between Rhode Island and Massachusetts Avenues. It's boomed and is now a whopping 67,149 ppsm.

The rest of central Northwest, from Mount Pleasant down to Massachusetts Avenue, varies from around 30,000-50,000 ppsm. Capitol Hill is in the 20,000-30,000 ppsm range.

Meanwhile, in Alexandria, the tract at the corner of I-395 and Seminary Road is up to 59,886 ppsm, 2nd densest in the region after Logan Circle. There hasn't been any new development in that tract since 2000, but the suburban-style apartment towers in it may have fewer singles and more families, which could account for the increase. Crystal City is 45,448 ppsm, and Ballston is 43,788 ppsm.

Suburban Maryland's densest tract is in Langley Park, at 49,354 ppsm. Downtown Silver Spring is 34,816 ppsm, and downtown Bethesda is around 11,000 ppsm.

Cross-posted at BeyondDC.

The latest Census data show that Montgomery County reached 1 million people, a statistic that has gotten a lot of worthy attention. Still, let's remember that jurisdiction boundaries are pretty arbitrary. As commenter AlanF also pointed out, DC, Arlington, and Alexandria (the "core jurisdictions") have just about reached 1 million as well (999,662 as of the latest Census estimates).

Michael Rodriguez decided to analyze "inside the Beltway" as if it were its own city. Given the way the Beltway separates communities, it's a good natural boundary and means more than the artificial lines between counties or between DC and Maryland.

Inside-the-Beltway would have about 1.7 million people. in 423 square miles. That's a little smaller than Los Angeles and only about half the density of people per square mile.

Update: Commenter npm points out that Rodriguez's table appears to be incorrect, and "Inside the Beltway's" density may be more like 80% of Los Angeles' rather than 50%.

Update 2: A reader with access to GIS systems has estimated the land and water area of "Inside the Beltway." Plugging in those numbers, and assuming that the other numbers on the table are correct, the table would look like this.

Update 3: Rodriguez has updated his post and fixed the errors in the DC and "Inside the Beltway" numbers. I've updated the table to reflect them.

Geography	Total area (sq. mi.)	Water area (sq. mi.)	Land area (sq. mi.)	Population	Density (Pop./sq. mi.)
Inside DC Beltway	266	10	256	1,725,686	6,749
District of Columbia	68	7	61	632,323	10,298
New York City	469	166	302	8,336,697	27,541
Los Angeles	503	34	469	3,792,621	8,087
Chicago	234	7	227	2,707,120	11,920
San Francisco	232	185	47	805,235	17,177

The lower density than Los Angeles comes because most of the land inside the Beltway is actually not very dense, except for central DC, Capitol Hill, along Georgia Avenue, the Rosslyn-Ballston corridor, bits of Silver Spring and College Park, and a few other places.

Also, if "inside the Beltway" were a city, metonymy in the national press would be even more severe than it is today.