I often add tracks from satellite imagery, and also do occasionally add missing tracktype tagging from that imagery.
The tricky bit is to understand that there are no physical barriers which are not visible on the imagery, but the tracktype, surface and smoothness are often easy to guess correctly, with the proviso that, when in doubt I put lower quality indication tags. I do that only in areas which I generally know, i.e. I have used many tracks in the area, and know from that how they typically look like. So, I use some kind of NA (Natural Intelligence) learning process to interpret the satellite images. >From my user experience (on bicycle) I find it important that a barrier-free track exists, that I can follow at my own risk as far as surface quality is concerned than not to have it on the map. Barriers (gates, fences, access forbidden signs) are the real risk in this. I often consult GPX tracks and also the Strava heat map to get an idea of which tracks are used. I would not, however, apply this approach to an area which I do not generally know. Doing this on a world-wide scale is most likely a bit more complex, but I could imagine that a person who is doing that continuously develops a capability to interpret well the satellite photos regarding tracktype. But, by my own experience, it is much more common to encounter access problems than to encounter tracktype-related problems. In summary I would not advocate a revert unless there are at least some real examples of errors. Volker (mapping often in the treeless and flat agricultural country side of the Po valley in Italy)
_______________________________________________ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
