Citizen Cartography

Maps are, and always have been, wielded as instruments of power; until the mid 20th century, most maps were created in the service of the state. However, the past few decades have witnessed an unprecedented increase in free and accessible geospatial data and mapping tools. As a result, the production and transmission of geographic knowledge has become democratized. Indeed, cartography is no longer a “science of princes”, but rather, an increasingly popular, and effective, vehicle for community-driven activism and self-representation. Today, practically anyone can create maps that challenge, and perhaps even overturn, established systems of knowledge production and representation.

I like to consider myself one of these 21st-century citizen cartographers. In my spare time, I produce thematic maps that illuminate (or at least attempt to illuminate) interesting demographic, political, and social patterns in urban spaces. I like working with city-level data, because at this scale, it is highly granular and precise. The following map of Boston, for example, uses contour lines to cast a simplified picture of the distribution of bicycle accidents in 2012: 

Underlying this visualization is a geospatial dataset that contains thousands of individual bike accidents, each with its own lat-long coordinates and attribute information. To generate the contours, I used QGIS—a free mapping program—to calculate the number of crashes that occurred within 400 meters from each pixel. Then, I extracted the contour lines by automatically connecting adjacent pixels with similar crash densities. I also imported several reference layers into QGIS—e.g., roads, parks, and water shape files, all obtained freely from OpenStreetMap—to provide visual context. 

After manipulating the spatially sentient data, I imported the graphic files into Inkscape, a free vector-based graphic design program. Although this move necessarily forfeits a lot of the actual data—for example, the bike accident records no longer know where they exist in geographic space—it also offers incomparable control over map aesthetics. In Inkscape, I assigned colors to individual colors, tweaked sizes and thicknesses, and added text. 

So, what does this map tell us? For one, most of the high-density crash hotspots appear to align with either highway entrances/exits (e.g., the Cambridge St. crossing of I-90 in the northwest quadrant of the map), busy multi-lane intersections (the Commonwealth Ave./Massachusetts Ave. intersection in the center of the map), or surface-level public transit corridors (the MBTA green line Huntington Ave. corridor, immediately east of the 'Brookline' label). And this map also illuminates the relative un-safety of the streets surrounding the Boston Common. This particular relationship has interesting political implications, as the Boston municipal government recently redoubled its efforts to ban cyclists from the Common.

(As an aside, shortly after submitting this map to a well-trafficked data visualization blog, I received a handful of letters from Boston-area cyclists who asked to use the map for advocacy purposes. I also received two exceptionally vitriolic pieces of hate mail, both of which accused me of misrepresenting the data. The fact that this map elicited emotional responses from both camps is, in my opinion, pretty cool, and definitely indicative of maps’ rhetorical power.)

Or consider the following crime map of Washington, DC:

This one uses a “binning” technique to aggregate individual crimes into equal-area polygons (in this case, hexagons). This cartographic technique is useful for simplifying dense point data and/or for visualizing multiple variables simultaneously. In this map, hexagon size represents the number of crimes that occurred within the area, and hexagon color represents the proportion of those crimes that were violent. Again, I used QGIS to generate the hexagonal lattice, to count the number of crimes occurring within each hexagon, and to calculate the violent crimes proportion. I also used QGIS to convert a black-and-white digital elevation mode into a semi-three-dimensional terrain layer. All of the data was free. In Inkscape, I added the bivariate symbology (size and color), added labels, and stylized the reference layers.

This map reveals some interesting patterns in crime distribution. For instance, few crimes were reported on Capitol Hill and in downtown DC , presumably due to a fairly visible law enforcement presence in those areas. Moving northwest from the city center, crimes appear to increase in frequency, although most of the crimes reported in these culturally diverse but rapidly gentrifying neighborhoods were deemed non-violent. Relatively few crimes are reported in the low-income neighborhoods east of the Anacostia River, but nearly a third of all crimes that occurred in these neighborhoods were described as violent. I chose to highlight a few hexagons in which many violent crimes were reported. Interestingly, these pockets of distributed relatively evenly throughout the city. Ultimately, this map shows that while many people associate crime with physical imperilment, most crimes reported in DC do not actually involve violence.