You could use the proj4 library to describe a circle using the great-circle distance.
For example, here's 3000km radius from Edinburgh, Tokyo, Cape Town and Quito in wgs84/Equirectangular. Only Quito is vaguely 'round', due to its proximity to the equator. I've also added in a single densified spoke line at an azimuth of 36 degrees (approx NE)
If we change to an azimuthal equidistant projection centred on Edinburgh, you'll see the radius around Edinburgh resolve to a circle...
On Mercator (like your web app), you see more distortion as you move away from the equator, but the buffers are more elliptical.
The following python code does that (requires pyproj and shapely)
from shapely.geometry import Polygon, MultiPoint, LineString
def geodesicpointbuffer(longitude, latitude,
Creates a buffer in meters around a point given as long, lat in WGS84
Uses the geodesic, so should be more accurate over larger distances
:param longitude: center point longitude
:param latitude: center point latitude
:param segments: segments to approximate (more = smoother)
:param distance_m: distance in meters
:param geom_type: shapely type (e.g. Multipoint, Linestring, Polygon)
:return: tuple (proj4 string, WKT of buffer geometry)
geodesic = pyproj.Geod(ellps='WGS84')
coords = 
for i in range(0, segments):
angle = (360.0 / segments) * float(i)
x1, y1, z1 = geodesic.fwd(lons=longitude,
# makes a great circle for one spoke.
example = geodesic.npts(longitude,latitude,x1,y1,1000)
coords2 = 
for xx,yy in example:
coords2.append((x1,y1)) # make sure we include endpoint ;-)
flight = LineString(coords2)
ring = geom_type(coords)
return "+init=EPSG:4326", ring.wkt
# example : Cape Town. 3000km buffer.
spec, wkt = geodesicpointbuffer(18.4637082653, -33.8496404007, 2000, 3000000.0, Polygon)
if __name__ == "__main__":
You can paste the WKT output into QGIS using the useful QuickWKT plugin.
You could use other methods - as coneypylon mentioned, you could create a circle on a custom equidistant projection in meters, centred on your starting point. I find though that for large distances an error creeps in (only a few km at 2000 km, but for intercontinental distances these errors can mount up)
From memory, the mmqgis plugin allows buffering in km. I'm not sure which method it uses, though.
Note that you might have problems rendering polygons in QGIS that cross the antimeridian if you're starting in Asia - ogr2ogr with the -wrapdateline option can help here. You might find this is less of a problem with openlayers/leaflet, IIRC they allow longitudes greater than 180 and less than -180.
There's a good writeup about geodesic buffering here on the esri blog.