Detail - Dot Density Map of U.S. and Canada showing one dot = one person.
OK this is taking dot density
mapping to another level altogether! Brandon
Martin-Anderson, who does something called “Computational Urban Planning” at
the MIT Media Lab, created a dot density map, but it’s not your normal dot density
map, where 1 dot equals 1,000 people or what-have-you. His map has one dot for each and every person
in Canada and the United States. That’s
right - over 340 MILLION dots on one map (plus location labels!) How did he do it? Well, as the students in the Programming for
GISc class last term will appreciate, it was done with Python scripts and a lot
of fancy footwork. You have to zoom in
really, really far to see how the individual dots look (they look like very
very very very small dots). If you look
at the map displaying the entire landmass, you see population centers more like
smudges rather than dots.
Why did he do it? You might well ask. Here’s what he says on the website: “I wanted
an image of human settlement patterns unmediated by proxies like city
boundaries, arterial roads, state lines, etc.
Also, it was an interesting challenge.”
Well, that last bit is undoubtedly true (the challenge part) but the map
IS mediated by proxies, it’s just that the proxy is very small (the census
block, the smallest geographic unit whereby the Census Bureau aggregates
individual data. There are generally from
0 - 600 people in a census block. Yes,
some census blocks have zero population.
In Wyoming, for instance, 64% of the census blocks have no population. The minimum size of a census block is 0.69
acre. There is no maximum size.
The main thing in his map that is
unmediated is the one person, one dot concept. Each dot on his map is not a proxy for multiple people. And since in many parts of the continent the census
block is such a small area, randomly assigning the dots within the block is
reasonably accurate depiction of where people live. This breaks down a little bit in the less
populated parts of the U.S., where the census block covers a much larger
territory, since the geographic size of the block varies from place to place,
but the population size within the blocks stays fairly constant. (For a nice history of the evolution of U.S. census
enumeration districts, see http://www.census.gov/geo/www/GARM/Ch11GARM.pdf
)
In any event, the dot map, while
looking not too much different at first glance from the traditional dot density
maps using 1 dot = many, is actually quite unique. For hard-core Python-ists, he gives the codes
as well as all the raw data he used. I
have to admit, I was a bit lost here.
Don’t even know what a “quadkey” is although he states it is a kind of
cheapo spatial index (used to sort the dots so you can generate the tiles?)
See the interactive map at:
http://bmander.com/dotmap/index.html
Read how he put it all together at: http://bmander.com/dotmap/methods.html
And, according to one of Brandon’s Tweets, “Barack Obama and the
First Family are the five dots in the center of this map:” http://bmander.com/dotmap/index.html#lat=38.896602&lon=-77.036613&z=14
Some other cool dot density maps in the blogosphere recently:
Commuter maps of Seattle: http://uxblog.idvsolutions.com/2013/01/people-dots-seattle-area-commuting.html
Really nice cartographic techniques, beautiful looking maps, check out the
legend on the “Seattle Commute Times in Minutes” map.
Thanks, Natalia, for sending me the original link to the
ultimate dot density map of all times!
Also, see the following posts about ways of mapping density:
