I am a bicyclist/attorney/programmer in Oakland, California. On a lark, and to learn about how OSM works, I made a tileserver/leaflet for California, which is at http://bikewater.org. The tiles are styled to show amenity=drinking_water prominently. Based on what I've seen on that map, I’m now interested in improving coverage for that tag, by creating a website to collect public-domain geo-located photos of drinking fountains/taps as a datasource for OSM. Before I spend time writing any code, I want to run it past the community for your thoughts and input, to help me determine if it is something worth my time. My overall goal would be to make it super-simple to get information on drinking fountains from people who are otherwise not into mapping, GIS, or OSM. In particular I'm hoping bike clubs might get into this, since it is data that helps them, and bicyclists cover a lot of territory but can stop easily to take a photo.

Here's what I've done and what I'm proposing, in more detail:

(1) Current OSM Drinking-Water Data. As a bicyclist, I want to know where I can fill my water bottles on a ride. On long rides or tours, I've run out of water in remote places; at that point, the location of the nearest water tap becomes a the most important datapoint. On shorter rides, or just riding around town, it's helpful to know where you can get a drink. So, using the switch2osm guide, I made the map at bikewater.org to shows amenity=drinking_water icons with more contrast and starting more zoomed-out (10) than the OSM default (17). I haven’t told anyone about the site, except this list. It is California-only and it is on a weak Amazon EC2 server, so it may be slow or fail to render some tiles. There was a $.99 sale on .org's while I was working on it; it's not an actual organization or anything.

What I think it shows, though, is that OSM mapping of drinking water sources varies widely, and seems to correlate with income or education level. Berkeley, Palo Alto, and San Francisco, have dozens of amenity=drinking_water nodes. By contrast, California’s central valley from Lodi to Bakersfield – an area where millions of people live – has about five. Around where I live in Oakland, I’ve started taking pictures of drinking fountains when I find them and I’d say about half aren’t OSM-mapped. I've also read about how a lot of municipalities and nonprofits are interested in drinking water data to reduce disposable plastic bottle waste, and reduce soda consumption. So that motivated me to think about how to get the amenity=drinking_water data more complete.

(2) Description of the Proposed Website. My proposed website would have a form where the user uploads the photograph of the drinking fountain, checks a box for a public domain (CC0) dedication, completes a CAPTCHA to reduce spam/abuse, and optionally enters text data correlating with commonly-associated tags for OSM's amenity=drinking_water, such as access restrictions (e.g., hours), description, bottle=*, indoor=yes, etc. After that, I'd show the uploader the photo's position (from EXIF data) on the bikewater.org map so user can preliminarily check if the site is already in the OSM database, and whether that location is accurate.

The data path from photo to OSM database I’m a little less sure about, but I think the initial method would be for me (or other volunteer reviewers) to open the picture in JOSM to check that the location is reasonable, check again that it’s not a duplicate, enter the tag data, and commit it. Obviously there's room for improvement there, like integrating with OSM accounts to allow direct edits, but I want to start simple.

There's other projects that are sort of like this, WeTap.org and Mapillary seem to be the closest. WeTap is water-focused and I’d probably reach out to them before doing anything, but right now seems to be iOS only. Mapillary, although it's new and hot, I've only looked at briefly; but it seems pretty complex and generalized, focused mostly on map enthusiasts for developing street view. I think what I'm proposing would be different enough, but if there's something that really already does this let me know. It sounds sort of like the "old" OpenStreetView but for one purpose; from reading the wiki article on that it sounds like it more or less worked. I'd probably start with just California and then add more data if the idea seems to have traction.

(4) Questions. Is this worth doing? Would pictures like this be good enough as a datasource for OSM? I don’t want to spend the time on it if folks don’t think those are "yes" answers.

A second question is, if it is worth doing, how much information do I need about the person uploading for it to be valid data? Do I need to keep and collect names and emails of uploaders? I’d rather not be collecting their personal information, and having the barrier to entry of creating an account, if it is not necessary. If there is a photograph of a drinking fountain, from an anonymous person who has indicated they took that photograph and it is public domain, and someone OSM-knowledgable has reviewed it, is that good enough to use as a basis for adding a node?

Let me know where else I should run this past people, as well, like maybe the imports mailing list or if I should make a wiki page or anything else before proceeding.



Talk-us mailing list

Reply via email to