For example in this view the road that ends in the middle of the woods (tile) does not really end there, but goes to the village and connects to the main road near the pond (tile). I am sure it is there and how it looks, but I cannot draw it precisely without GPS trace. Should I draw it as a straight line and correct later? There is valuable information added (points are connected and that is the important part for navigation). And how about GPS traces with limited accuracy? asked 08 Jul '11, 00:06 LM_1 |
Regarding the "general question" of minimum expected accuracy - I'd tend to expect that any edit should leave a map that is overall more accurate than it had been before the change was made. In this case, a section of road is/was completely missing, so anything that indicates the presence of the road is improving the overall accuracy of the map. Where the accuracy is substantially different from then level of accuracy of surrounding objects, I'd expect a fixme= tag, but if a significant number of ways in an area are approximate, the source= tag should be sufficient. That then leaves open the question of how to properly determine the accuracy of the surrounding data when adding data that you can't tell whether it's more or less accurate than what's already there, but discussion of that would belong in another venue... answered 09 Jul '11, 16:17 banoffee Improved accuracy of map after edit seems like an easy-to-follow policy.
(09 Jul '11, 23:31)
LM_1
|
If you do add the roads, make sure to use a 'FIXME' tag. For example, 'fixme=Approximate route, needs survey'. Several mapping tools will highlight such 'fixme's and alert nearby mappers that the road needs improving. answered 08 Jul '11, 12:22 Richard ♦ 2
... or even more concisely "fixme=resurvey", as suggested at http://wiki.openstreetmap.org/wiki/Fixme and widely used.
(08 Jul '11, 13:55)
banoffee
|
Feel free to "guess" the road in your case, and revisit the geometry when you have better sources available. But I would add a caution to this - a couple of years ago someone added straight line connections between railway stations across a country in Europe. This was completely inappropriate given that these cut straight through areas where the roads were already well mapped. Instead, I would say the "minimal expected accuracy" should be based on the existing accuracy, and the density of real-world features, of the surrounding area. So a straight line through a forest might be appropriate where a straight line though an urban area is not, and a straight line through an unmapped area may be appropriate where a straight line through a carefully mapped area is not. answered 08 Jul '11, 10:51 Andy Allan 1
We had a similar discussion in France about subway mapping in Paris. Some people mapped underground features very approximately (no GPS, no aerial imagery, just personnal feeling) where everything on the ground and above is very, very accurate (cadastre, aerial imagery).
(08 Jul '11, 15:40)
Pieren
|
I would not hesitate to "guess" the road in your case. You are right in saying that the topological information (what connects to what) is more important than the geometry. You can always add answered 08 Jul '11, 00:27 Frederik Ramm ♦ Jonathan Ben... |