Hello,
I am Ernst (de_pelgrim) a Dutch guy always interested in mapping, but never started the right way. Now I am doing some reading about mapping and planning to practice mapping during my holiday in Lesbos Greece.
After walking around Petra (Lesbos) looking at what could be added I found a lot of mapping work that can be done.
But my first remark is that the building drawn not always match the location on bing satellite images and I am not able to shift the satellite images to correct this. So is Bing not a good reference image or was an other reference image used, which one? Or was this done by some automated tool and needs editing.
An example is a pier that is about 10mtr off compared to the satellite images and has no connection to the road.
I’m using Android based tools like osm go, every door and vespucci on phone and tablet, no desktop available at my holiday address.
I hope someone can help me start, by explaining what’s best to do.
Best regards,
Ernst.
There’s no super simple solution, but the issue is typically caused by the aerial imagery not being properly (fsvo) aligned. This could potentially be past versions of the imagery available or current.
The good news is because you are there it should be reasonably easy to determine what is off by both using existing and generating new GPS traces.
Then you should align the available imagery as good as possible with the traces (it could well be that this isn’t actually necessary but then you know that it was simply old imagery that caused the issue). All editors worth their salt will allow you to adjust the background and store the offset(s).
If after all of that the buildings are still substantially misaligned then you can consider moving them. I wouldn’t do that for differences under a meter or two, but would for 20 to 50.
Thanks SimonPoole, so I was thinking wrong about satellite imagery, but for now it is what it is.
I am going to walk to an open spot (inside or at the edge) of the village with a building I recognize on the satellite imagery and Gps mark that point in vespucci walk a few hundred meters to the next open spot and make a second mark, and maybe do 1 or 2 more in a line or square. Repeat it a couple of times to check gps variations. Now I can align the satellite imagery on those 2-4 marks and see what the result is compared to osm data. If it’s fairly oke (within 5 mtrs) I don’t move it, but I can add new buildings. This I repeat along the village, luckily the village is small. Sadly it isn’t allowed to use a drone inside the village for better contouring buildings, but this my first tryout.
Ernst
A couple of things to keep in mind.
- If the off nadir angle of the satellite image is large, buildings, particularly tall buildings, can appear to “lean.” The location where the base of the building appears is likely to be more accurate, and is the one we want to capture.
- The horizontal offset in satellite (and aerial) imagery can vary from place to place. There are a couple of reasons for this. First, the imagery has been orthorectified using a digital elevation model (DEM), the DEM may be relatively course, e.g. 30 meters and of course has its own inaccuracies. As the elevation changes across an area it may result in different offsets (even though orthorectification is supposed to correct for this, it can’t do a perfect job). Second, What we consume is really a mosaic of different images perhaps captured at different times from different angles. Each part of the mosaic may have a different offset.
- The horizontal offset in satellite and aerial imagery can vary over time. The provider may update all or part of the mosaic over a give area at any time. Because it was captured under different conditions, the new imagery is likely to have a different offset from the previous imagery.