0

I trying to create buffers (50km) around many points all over the world (airport locations) in QGIS. I will then overlay these buffers with a world-wide raster dataset so I can extract information from that. The point layer is in WGS84 now, I am wondering what is an appropriate projected coordinate to use (one that does not give me too much distortion)?

I have searched this site and the internet a lot and did not find an answer because I think this cannot achieved with PostGIS or just by using a projected coordinate system for one country.

2

There's a lot of things to say here. First, there is no really one projected coordinate system which can cover the whole world without any distortion. Secondly, the units of WGS84 is degrees, so you won't be able to compute buffers in kilometers accurately.

In my opinion, the best option is to use the geography type in PostGIS as explained here: What is the Difference between Geometric and Geographic columns?.

  • So I guess I need to import both the raster dataset and the point dataset into PostGIS, then do some kind of buffer operation to do the calculation? Is this the way to go about it? – Nathan Apr 21 '16 at 7:02
  • Yes I think it is. You'd use the Geography type, which is always in WSG84 and so, is unprojected. – pzijd Apr 21 '16 at 13:46
1

Its fairly easy with a bit of spherical geometry to work out the exact coordinates of a 50km circle centred at any lat-long coordinate. So you don't really need to use a buffer algorithm.

On a sphere, every 50km circle is the same, just shifted to a new centre. So compute the coordinates for a 50km circle at the North pole (easy) and then apply a rotation to centre it at your point coordinates.

If you can't do the above maths yourself, then for each point convert its coordinates to any cartesian coordinates centred on itself (eg zenithal, looking straight down at each point), compute the circle (use a buffer algorithm) and transform back to lat-long. Look out for points 50km from the poles since they'll go screwy when you back transform.

  • I think that makes sense to me. Since I don't expect any points to be within 50km from the poles I don't have the worry about the caveat. Do you suggest doing this in QGIS or PostGIS? – Nathan Apr 21 '16 at 7:08
  • isn't it a translation not a rotation? – Ian Turton Apr 21 '16 at 7:51
  • 1
    @iant I suppose its a translation on the metric space of the surface of the sphere, or a rotation in 3d space in which the surface of the sphere is embedded... – Spacedman Apr 21 '16 at 10:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.