|
Hi I am making my first edits on OSM. I am concentrating on POIs in my local area at this stage. I have a few questions. Firstly, I am currently using Potlatch and it seems to me that the background satelitte imagery is everso slightly off from the data - probably only by a metre or so. My question is, if I am thinking about tracing the imagery to set data for buildings for example, is this acceptable? Building position data that already exists in my area seems to have come from surveys. Secondly, I am noticing that in my immediate area there is variation in the way features are described. For example one feature might be described using a node inside a closed way, e.g. a node is used to describe a shop, and this node is inside a closed way describing a building. But then just around the corner a closed way is used to describe a shop with no internal node, and no data on the closed way describing it as a building. (I hope that makes sense). I am wondering if there are guidelines about how POIs such as shops should be described? I'm guessing it might depend on whether each POI is in a sepertae building for example? I guess I am wondering if I should be changing the way objects are described, or whether I should just be leaving the existing nodes and ways, and editing their properties (to update the data on what they actually are, e.g. change name of shop/restaurant etc). Cheers Richard |
|
Welcome, it is better to ask one question at a time. I will answer a couple of your points. The imagery may be off, you can move it but you will need several gps traces as a reference, then spacebar and drag will move it if required. As OSM as lots of contributors we do not always map and tag the same way, the wiki and this forum may lead you to the majority idea. or look around the map and see what others have done in your area. good luck |
|
Bing imagery can be out by up to a few metres. In part, that may be because, to correct the perspective errors in an aerial photograph, it is necessary to know the camera's exact height above the ground, and height data of the required accuracy is difficult to obtain. In my area (Fareham, England), the Bing imagery seems to be about two metres adrift at the highest zoom level, but the second highest zoom level appears to be very accurate. As Andy Mackey says, the best way to use the imagery is to slide it until it agrees with something reliable (e.g. several GPS traces in the same area, preferably taken at different times). You can then map things like building outlines from the imagery, although I would still try to confirm the line of a public road or path with a GPS trace of my own. You should also be aware that GPS units need a bit of settling time before they achieve their best accuracy. Ideally, leave the device switched on for at least a minute, in a place where it has a good view of the sky. 1
bing (as many other aerial imagery sources) can be substantial misaligned, offsets of up to 50 meters are not uncommon. It is really important when tracing from imagery to
(19 Oct '12, 23:18)
SimonPoole ♦
I hadn't realised that the imagery could be that far out. Could you show me a place where the georeferencing error approaches 50m?
(20 Oct '12, 11:01)
Madryn
Madeira used to be offset by a tens of meters. It's not now; it looks like Bing have fixed their imagery (it now aligns with our tracks). This question refers obliquely to the problem.
(20 Oct '12, 12:22)
SomeoneElse ♦
|
|
My answer would be to map the POIs to the fullest data (and time) available. So, if there is reliable information about which buildings which POI is in, and there is good aerial imagery, and you have enough time to trace it reliably, then do draw the building and tag it with the POIs attributes. But you could only add a node for the POI also and it would not be incorrect, but would be much faster. According to the wiki, the three forms of describing a POI are correct, but personally I use only the "node at the center" and "building with attributes" approaches and do not use the "node at the center of the way with attrbutes" approach. |
