[geotk] Circle Polygon

[theuns]

Ok, Thanks for the reply Martin

I also agree with what Adrian explained.

I displayed a world shapefile using the EPSG:3395 projection.
and draw 4 circles with radius 50km.

This is the places(more or less) where i draw them , and i measured the
diameter across the with and height, of the circle.

1. Equator , width +-100km , height +- 100km 
2. Cape Town (SA) , width +-82km , height +- 82km
3. Southern Tip of South America , width +-57km , height +- 57km
4. Northern Part Greenland , width +-14km , height +- 14km

So the circle's diameter does take off further away from the equator,
and like Adrian said, it should be egg shaped (eclipses), but the circle
stay round?	

I need to draw a circle,to plot radar coverages. The largest circle i
would have to draw is 500km.
So i would take Martin's suggestion to play around with different
projection's according to the need of , what to preserve.

Only thing that bother's me is the that the circle does not display
correctly, in different parts of the world with the EPSG:3395
projection?


On Tue, 2009-09-15 at 11:08 +0200, Martin Desruisseaux wrote:
> Hello Thuns
> 
> As Adrian said, its look like the deformation we would expect from a Mercator 
> projection. A difference of 1° in latitude will appear larger and larger in 
> metres as you approach a pole. At the pole, the value reach infinity.
> 
> If we put the raisoning in the opposite way, having a fixed length like 2 km 
> converted to degrees will appear smaller and smaller as we approach a pole. At 
> the pole, any circle will appear to have a radius of 0 degrees of latitude.
> 
> It is not possible to project a sphere on a flat surface while preserving every 
> geometric properties. This is why there is so many projections. You need to 
> choose which geometric properties you want to preserve (angles, surface, etc.).
> 
> If you need to preserve the circle radius while moving north-south, and if you 
> don't need to move in the east-west direction more than 6°, try a UTM (Universal 
> Transverse Mercator) projection.
> 
> If you prefer to stay with the classical Mercator projection, try to set the 
> "Latitude of 1st standard parallel" parameter value to the central latitude of 
> your area of interest. This the latitude where the scale is "true" and play the 
> role of the correction factor you mentioned.
> 
> 	Martin