No announcement yet.

OpenStreetMap - crowd-sourced mapping.

  • Filter
  • Time
  • Show
Clear All
new posts

  • #76
    As per my explanation in the message I sent you (on OSM), shortly after Strava released a highly-detailed version of its Heatmap (very useful for map editors) it pulled the plug because it inadvertently revealed details about known and secret US military bases around the world. Some personnel forgot all about OpSec and uploaded their daily runs to Strava, thereby revealing to everybody not only the base's location but its outline. Strava eventually re-released it but without high-magnification.

    During that short window in 2017, I managed to use the high-res version to refine many of OSM's High Peaks trails (their meta-data contains a note indicating if they were refined using 2017 Strava Heatmap). To repeat that trick today, you'd need to get your hands on more than a dozen GPS tracks (and I'm being conservative because popular routes in Strava Heatmap may represent hundreds of tracks). That's why when I see one of these tracks being modified today, I have to ask how is the new reference data superior to 2017 Strava Heatmap.

    FWIW, I have my system set for recording a waypoint every 2 meters or 5 seconds (i.e. time and distance). That's for hiking (it's different for walking, driving, cycling, etc).

    For completeness, I'll add the 'zigzag' I had referenced in our OSM discussion. The red line is the original version of the East Trail (to Giant), based on 2017 Strava Heatmap, and the green line is your modified version. I circled the zigzag because it represents a significant deviation from the original route.

    Looking for Views!


    • #77
      Hi Taras,

      I was simply following the public GPS traces in OSM; snip attached. Yes, there are only two.

      What data would be better to use?

      Click image for larger version  Name:	Capture.JPG Views:	0 Size:	17.8 KB ID:	503541

      Edit to add: Open data needs to be reviewable and verified by the open community; relying on unavailable data isn't sustainable in the long run.


      • #78
        Originally posted by toeknee View Post
        Edit to add: Open data needs to be reviewable and verified by the open community; relying on unavailable data isn't sustainable in the long run.
        Its lack of availability doesn't mean it has lost its authority. It represented the best data available to any OSM map editor back in 2017. If a section of a trail's routing has changed since 2017, then the new measurements become more authoritative. In this case, it's possible that this section of the trail has either changed since then or simply represents two people who followed a slightly more circuitous route compared to the mapped trail. That's the problem when you only have two GPS tracks; it's sufficient to draw a trail that hasn't been mapped yet but not enough to challenge one that was based on far more than 2 tracks.

        I will offer an alternative explanation: maybe I made a mistake. Imagine those two lines appearing as a 3/4" wide red line. That's about how wide the Heatmap would appear at that magnification level. In other words, the very sharp inverted V would be difficult to discern when drawn with a wide brush. I measured the base of that inverted V and it's less than 10 meters wide. It's possible that I misinterpreted the Heatmap and drew the trail across the base of inverted V as opposed to following its outline. Maybe. I'd like to see more than 2 GPS tracks in this instance. In addition, the GPS tracks should be recorded outside of winter because snow obscures paint-blazes (and more) and allows for easier travel outside of established routes.

        Just in case you think I'm being obstinate, all that I'm trying to achieve is impress upon you to the challenge of maintaining the integrity of the existing trail data. If something is based on many measurements and someone comes along and changes it based on just two measurements, you can see how, if left unchallenged, can lead to diminished accuracy. FWIW, in the grand scheme of things, your changes are minor compared to another editor who recently made sweeping changes that have rendered some trails invisible (and they aren't responding to my messages).
        Last edited by Trail Boss; 01-12-2020, 10:42 PM.
        Looking for Views!


        • #79

          I recommend you review the many GPS tracks posted on AllTrails. They do not provide an aggregated view (like Heatmap) so you'll have to review each one individually. Nevertheless, it shows the variance in quality of the many recordings. This is one of the better ones (recorded this month) and supports the map's version of the trail. Unfortunately, this particular track stopped before the disputed area (i.e. the "inverted V").

          Click image for larger version  Name:	Screenshot from 2020-01-12 23-04-32.png Views:	0 Size:	54.4 KB ID:	503545

          Here's another, also showing inbound and outbound tracks, but shows the disputed bare-rock area. Notice how both tracks align with the trail precisely where you altered it to appear as an inverted V.

          Click image for larger version  Name:	Screenshot from 2020-01-12 23-13-09.png Views:	0 Size:	48.9 KB ID:	503546

          Nevertheless, it is possible to find an example or two where people's tracks display your inverted V. I even found one showing the direct path in one direction and the inverted V in the other directions.

          Click image for larger version  Name:	Screenshot from 2020-01-12 23-28-09.png Views:	0 Size:	52.4 KB ID:	503547

          So what does it suggest? Some people followed the depicted trail, representing the most direct route across the bare-rock area, whereas others followed a more roundabout path. When I had used 2017 Strava Heatmap, the center-line of the aggregated tracks ran along the dashed line shown above (i.e. the average of all available tracks).

          Given that satellite imagery shows this specific area is bare-rock, devoid of alpine vegetation to circumvent, my preference is to retain the direct route shown by the original trail.
          Last edited by Trail Boss; 01-13-2020, 12:40 AM.
          Looking for Views!


          • #80
            Thank you, I appreciate all the information and for your patience.

            I do not think you are being obstinate. Getting the trails on OSM accurate is important. There are enough people following OSM-based maps on their devices (including me at times when breaking trails) that where we draw the map will cause a physical impact on the ground.

            I agree that two GPS tracks is a minimal and generally insufficient basis for changes if current maps are based on better data. And good point about winter tracks. I didn't think to look at AllTrails; it's not a site I use often. I reverted my changeset 79297429.

            I still feel we have an open issue in that much of the OSM data is based on data that is no longer available, and we're in some weird "don't touch it or you'll break it" mode. At a philosophical level it violates the principles of community data, OSM, Wikipedia, and even general scientific inquiry when people are forced to say, in essence, that the maps are based on data no one can review, so no one contribute to improving or even reviewing the maps. Another way of saying this is that unavailable data does not have authority anymore because the open collaborative community can't use it to verify the OSM.

            All of us should be able to review the OSM data and the data on which it is based. We should be able to review each others' data. I upload my GPS traces to OSM before I use them to edit OSM. (Though I must admit when I have a multiple personal traces of the same trail, and I use the multiple traces, I have only been uploading one to OSM. I'll do better in the future.)

            Guess I'm going to go do that hike a few more times. It was sweet having extra people, two cars, and being able to do the traverse, I don't have that option often.


            • #81
              You've made several valid points about the nature of the data used to define the map. However, this specific situation is not unique. I challenge you to find the sources used to create most anything found in OSM. Most of the time the map editor overlooks to mention what was used as reference material.

              Although Strava Heatmap is no longer available to OSM's map editors at high magnification levels, it remains available at lower magnification in OSM. Therefore the situation isn't nearly as dire as you've described it: "... violates the principles of community data, OSM, Wikipedia, and even general scientific inquiry...". If you have a Strava account, you can still magnify the Heatmap to a high degree, just not within OSM's map editors (like JOSM and ID). Having said that, I know of efforts to circumvent Strava's imposed limitations on OSM. I've tried them myself with mixed results (mostly failure).

              The upshot here is that if the trail's tag indicates it was refined using 2017 Strava Heatmap, the trail is more than likely to be sufficiently accurate for hiking navigation. If a section of trail has been re-routed then it stands to reason that section should be revised in OSM using as many measurements as possible. Otherwise, attempts at further refinement are questionable due to the lack of copious amounts of aggregated data (like Strava Heatmap at high-magnification) or the editor's unwillingness to manually aggregate many, many GPS tracks.

              If there is anyone who still believes one or two GPS tracks is sufficient to change a trail, that was based on 10+ tracks, they should peruse examples in AllTrails or Wikiloc (or other source of GPS tracks). The examples demonstrate the degree of variability in tracks (see the examples in my previous post); each track looks like a "drunkard's walk". You need the average of many, many tracks to get an idea of the trail's true location. Strava's Heatmap was very helpful for defining the trail through Avalanche Pass where individual tracks often varied widely due to difficult GPS reception caused by the high rock walls in the pass.

              There are a few High Peaks trails that I did not refine using 2017 Strava Heatmap because I ran out of time (i.e. when Strava blocked access) or even the Heatmap's data was clearly sketchy. The Preston Ponds Trail is an example of that. Refining it is a challenge because it sees little traffic and so there are very few publicly available GPS tracks. The trail's meta-data clearly states the available information was sketchy and it could benefit from further refinement.

              If you, and others, are keen about improving OSM's ADK trails, I invite you to add the location of ladders and staircases. This thread lists the ones I have not mapped yet (due to a lack of data)
              Looking for Views!


              • #82
                I even found one showing the direct path in one direction and the inverted V in the other directions.
                That might actually be my recording from Oct 13 - unfortunately, I don't recall what made me go one way and then another - I can only say it was not intentional.

                BTW Alltrails does have a heat map overlay option with the paid version. From what I see, it looks like it supports the inverted V.


                • #83
                  Post a screenshot of it and any other details about their Heatmap

                  I have a free Strava account therefore I can magnify their Heatmap to, what they call, "street level". I can't access it in OSM's map editors but I can view and compare it to other maps. In the following screenshot:
                  • The top image is from Caltopo showing the original East Trail as I had drawn it, two years ago, using 2017 Strava Heatmap as a reference.
                  • The bottom image is from today's (2020) Strava Heatmap. I've set the magnification level of both images to be as identical as it allows (the Heatmap is slightly more enlarged).

                  Click image for larger version

Name:	Caltopo and Strava Heatmap.png
Views:	51
Size:	31.5 KB
ID:	503562

                  1. The trail that was mapped using 2017 Strava Heatmap still closely matches 2020 Strava Heatmap.
                  2. There are many "outliers" (note the purple tracks) but the bulk of them agree.
                  3. The white line's width is on the order of 10 meters which corresponds to the rated accuracy of consumer-grade GPS receivers.
                  4. The alleged location of the 'inverted V' (marked with an arrow) does not show a corresponding sharp kink in the Heatmap.
                  Looking for Views!


                  • #84
                    I magnified Strava's Heatmap and adjusted the colors and discovered the following interesting result. You can see a unique GPS track highlighted within the aggregated tracks. I don't know why it's visible but it may represent the most recent track uploaded to Strava. What's of greater interest is that this track contains the so-called inverted V and, when compared to the aggregated tracks, is clearly an outlier.

                    Click image for larger version  Name:	Annotation 2020-01-13 145203.png Views:	0 Size:	40.1 KB ID:	503565
                    Looking for Views!


                    • #85
                      It's not easy to make out, but the kink is more pronounced in the Alltrails heat map data (red lines) than on the current trail (dotted line):
                      Click image for larger version  Name:	RPR.PNG Views:	0 Size:	31.1 KB ID:	503569
                      Attached Files


                      • #86
                        The screenshot is small but I can still see that only outliers form the inverted V ("kink"). The majority agree with Strava Heatmap and show a simple directional change (~90-degree change to the southwest if you're traveling east to west).

                        Anyway, the real deal is for someone to hike the trail and take note of what the route does at that exact spot, preferably when there's no snow cover (that may obscure any blazes or self-evident signs of regular passage). I've been through there four times but honestly can't recall that spot. Perhaps there's a ledge that people skirt when travelling one way and climb it when traveling in the other direction.
                        Looking for Views!