I'm a big fan of this sort of everyday practical data visualization! To anyone who hasn't already seen this feature: when you select walking or biking directions on Google Maps, it renders a line in the sidebar to show you elevation change over the route. You can mouse over positions on that line, (or over the route line on the map) and it will render an indicator on the other line.
In this example, the elevation trace reveals a very steep section on Gregory Canyon Rd, between Baseline Rd and Flagstaff Rd:
Meh. Not to be a debbie downer, but that's the most basic courtesy you can offer to cyclists, and even that isn't telling half the story. In central Europe anyway, Google Maps is unusable for creating cycling tracks or directions.
There are much better alternatives, such as komoot or Strava or GPSies (I'm not affiliated.)
As important as (and sometimes more important than) the profile in a section is the gradient. The Google map doesn't tell me if there's going to be a 10% or 20% piece, and I can't glean it from the curve, because depending on the length of the track, the relations change.
Additionally, it's very important to know what kind of road you're on, and what kind of surface that road has. The thing only Google could do here, and doesn't, is to tell me how busy (with cars) the roads are and offer me a way around busy roads. Komoot at least shows me the type of road or street and its surface.
the hill in the middle looks scary, but the line along its ridge tells me the gradient isn't steep. Yellow is 8% (it tells you on hover). It tells me what percentage (and where) the road is a bike road, or larger road, etc.
Strava on the other hand has heat maps: https://www.strava.com/heatmap which tell you where local cyclists like to go. One caveat is that it shows you where local road cyclists and mountain bikers like to go. Those folks can have vastly different understanding of what viable road conditions are from each other, and from you.
Unfortunately, planning cycling tracks is still a chore, and getting it wrong can be dangerous or at least unpleasant. The market is there, but it's nowhere near as huge as for cars.
(I'm the author of the 2014 article posted here - really pleasant surprise to see it surface on HN.)
Fully agree with your comments. It's not just Google that can tell you about busy roads, though - in some countries that's available as open data.
I run https://cycle.travel/map , which uses open traffic data (in the US, UK and a couple of other places) to influence its OpenStreetMap-based bike routing. So, for example, in the Home Counties (broadly the London commuter belt) the A-class roads are generally busy and unrideable, whereas A-class roads in the Scottish Highlands are often (not always) beautifully quiet. cycle.travel will happily route along the latter but not the former.
The Strava approach is interesting, but Strava users are biased towards athletic cyclists who generally have a higher tolerance for busy roads. For tourers and other leisure riders I'm less convinced that the presence of other cyclists is a good signifier of pleasant cycling. (In London, for example, Strava shows the Euston Road as a busy cycling thoroughfare, which I'd never dream of riding.)
Thanks for chiming in! cycle.travel/map is really good! I like how responsive the interface is, and its routing algorithm seems on par with other ones I've tried. Do you take into account traffic info for Germany? I couldn't find good open sources for that.
Yeah, Strava tends to overestimate the rideability of bigger roads. Keep in mind that a lot of groups use Strava, and it's easier to ride a busy road when you're in a group. In Germany, for example, you even have a different legal status on the road. I would love it if I could somehow filter the global Strava data, but they don't make their data available; not that I would expect them to.
No, I've not found national traffic data for Germany, unfortunately. I think it exists for one or two areas (NRW maybe?) but certainly not nationally, and the proprietary sources tend to be prohibitively expensive. I'm intrigued as to whether it might be possible to count cars in imagery as a possible alternative...
I cycle along the Euston Road fairly often - it has bus lanes, which to me makes it pretty much fine. I wonder if that's another dimension of variation which should be considered.
I do avoid the underpass, though, that's terrifying!
That komoot visualization is nice. I'm glad to see that others are doing it better than Google. I thought my comment might bring out replies like this and the Open Street Map equivalent mentioned elsewhere in the thread -- happy to learn about these alternatives.
For those who prefer OpenStreetMap-based solutions, OsmAnd and OpenRouteService[1] do this as well, and GraphHopper (available on osm.org and on [2]) shows total ascend+descend. Not sure what they base the data on, though, as most roads aren't tagged with a very accurate inclination, so I guess it uses a height database that is not OSM-based.
Thanks for sharing those. The popup on hover says "Elevation: [...]" so I think you're right that they're approximating grade by using a height database. It could be OSM-based if enough contributors are supplying lists of (lat, lon, elev) GPS coordinates from journeys they take.
It's also very possible that they're using the public domain dataset from the Shuttle Radar Topography Mission (SRTM)[1] in 2000 -- probably with some cleaning to eliminate large spikes due to trees and buildings.
This map uses shading to indicate how steep the gradient of each individual city block is, but interestingly the shading doesn't show the direction of that gradient, so you don't immediately know whether it will be uphill or downhill. But the map also includes contour lines, so if you look closely, you can figure it out from the contours even if you're unfamiliar with the neighborhood.
On the other hand, if you're generally familiar with the neighborhood, the shading might be helpful by itself because you might have an intuition about when you're going to be going uphill and when you're going to be going downhill.
I agree that replicating the cool visualization method from the linked article might be tricky in San Francisco for the exact reasons you mention.
I have family living in a vilage shown in the snapshot of the map. It's deep black forest so, while I think displaying the steepness is a neat idea, there, you just know that over the mountain is steep, and if you want to avoid it, you stay in the valley :)
In this example, the elevation trace reveals a very steep section on Gregory Canyon Rd, between Baseline Rd and Flagstaff Rd:
https://i.imgur.com/NWiRmdK.png