I’ve long been frustrated with the state of transportation data in the US*.  Last week’s post was about the weaknesses in the data the National Household Travel Survey (NHTS) releases at very fine scales.  You can still feel confident about the overall trends in the NHTS, but as you slice away degrees of freedom (like income levels or gender, you start to get less reliable data summaries.  And yet that is some of the best data about biking and walking that there is.  The Census has been learning to count walkers and bikers  since 1960, but they recently changed their survey methods in ways that challenge comparison between 2000 and 2010.  What’s more, the census is about aggregate behavior at the block group level, and only for the journey to work.  The hapless sample of households that gets the long form wouldn’t be nearly as likely to give up their entire travel day for the federal record if they knew they could;t hide among 1,500 of their neighbors.

I’m fine with that.  I’m not fine with the paucity of data on walking and biking compared to traffic and transit.  It makes sense that we would track the latter two a lot better than we do the first two, because they are more expensive and more dangerous.  Transit and even traffic lanes cost millions  of dollars a mile to build and thousands of dollars a year-mile to maintain, while walk and bike lanes cost tens of thousands to build and hundreds to maintain.

Another thing we don’t know is just how much walk and bike lanes we have, and how much they cost.  Sidewalks are built by default for certain classes of roads, such as residential or parking lot circulation, often by private developers.  Many arterials and highways don’t even have them, because what walker in their right mind would walk a mile on the side of a freeway?  Some low traffic, narrow roads are built without any sidewalks at all; a reasonable choice considering the traffic volume.  Bike lanes come in all sorts of forms, from a string of desultory signs linking one stretch of shoulder with another, to a stripe of paint between traffic and the curb, to fully separated paths with their own paving and maintenance schedule.  Whether they actually get repaved and maintained on schedule is a function of how many people use them**.

This is why I was so excited to learn about OpenStreetMap (OSM) a couple of years ago. A crowdsourced approach to mapping, it contains as much geographic detail about the world as the world is willing to give it, in a freely available and customizable interface.  There are bicycle mapping sites for OSM  It took me until this year to actually sign upend begin editing the stuff.

What’s exciting about OSM is that its a shared Geographic information System (GIS).  The data is not very deep, and there’s little call for analysis, but the point here is mapping.  Mapping the whole world.  While Google, Bing, and Mapquest already do this, their data is not openly accessible or usable.  With OSM, I can edit a map today and have it in GIS tomorrow, to use in geographic and demographic analysis.

OSM is a spectacular open source data set, with quite a lot of information on parking, land uses, and building footprints.  There’s even a fairly comprehensive set of bicycle trails and lanes.  I have no way of field-truthing all of them, but from the places that I am familiar with, OSM seems to have the goods.  There’s even an OSM mapping interface with bike trail features highlighted. But no sidewalks.

The official line is that sidewalks are assumed to be a part of city streets and assumed to not be a part of freeways.  The classifications between those two extremes are what interest me, however.  There are 4.5 million miles of roadway, and 2.5 milli0n miles of local roads, in the US.  How many of them have sidewalks on one side, on both sides?  This information interests me, as I would like to be able to stand walking up next to traffic, transit, and biking and make comparisons.

There is an officially sanctioned technique in OSM for assigning sidewalks to traffic ways, by editing the tags of the lines in the OSM interface.  This has three problems:

  • Sidewalk presence/absence is not apparent in the OSM editing interface
  • Sidewalk tags do not make it into GIS
  • Crosswalks are not explicitly described

It has been a learning experience to go to different cities and try to pick out their sidewalks.  I got tired of drawing sidewalk lines in Washington DC before I realized that the best foundation for walking geography in a street map full of traffic ways is crosswalks.  Crosswalks are the only legal way for walkers to enter the streets, although they offer scant legal defense for the deceased in many states.  The form and frequency of crosswalks indicates who many people are walking in an area, and how many would have the gall to request occasional passage though the traffic way. One feature of the suburbs is the three legged crosswalk.  At a four way intersection on a traffic arterial, the event of a walker is presumed to be so rare that one side of the crosswalk is left open to the flow of traffic.  Wouldn’t want to slow down traffic commuters with mewling walk signals.   The majority of intersections outside any downtown are devoid of crosswalks, leaving walkers to use common sense and their squirrel sense to make it across the road.  Millions do.  Thousands don’t

The form and scale of crosswalks tells a lot about the esteem of walking in a place.  The average speed of walking is 4.5 feet per second, or 11 seconds across a 4 lane road.  The elderly and young are slower, of course.  Small crosswalks and skinny streets are less hazardous than broad and complex crosswalks.  Sub-crosswalks across slip ramps for the hurried right-turning driver’s benefit are especially dismissive of the walker.  What OSM lets me do is tour all this for any town in the US, from New York to Flowery Branch.

I’m mapping crosswalks in OSM to learn about walkable cities at first scan, but eventually to link sidewalks into blocks.   It’s a great way to explore the variety of American cities and how we build, or don’t, for walking.  Check it out, and if you like, add some crosswalks, sidewalks, bike paths, stairs, utilities, cafes, groceries, or even buildings.


* Though its apparently better than European data in may fields
* How many people use them is a function of how much they connect