Changes between Version 1 and Version 2 of Ticket #12427, comment 12


Ignore:
Timestamp:
2016-01-24T15:53:55+01:00 (10 years ago)
Author:
cmuelle8

Legend:

Unmodified
Added
Removed
Modified

  • Ticket #12427, comment 12

    v1 v2  
    1313> Why not simply use `greatCircleDistance` and multiply the mean radius?
    1414Because the method I use is documented elsewhere (see javadoc) and curvature radii functions in class Ellipsoid are there for a reason :)
     15
     16EDIT:
     17The problem interpolating LatLons you refer to (esp. at the poles) appears if you measure their distance on the mercator projection directly - this is what {{{euclidianDistance}}} does, and you can see the effect easily by drawing a straight line from (80,-170) to (80,170) and then hitting {{{Ctrl+I}}}.
     18
     19{{{curvatureDistance}}} does '''not''' do this. It simply uses the interpolation to get another point on exactly the same line. ''All'' distances measured use a very similar method to the one in greatCircleDistance, with the exception that for the east-west bound integration another "great circle" radius is used than for north-south bound integration. I've explained this thoroughly in the javadoc to the function, just like I've explained the "mistakes" {{{euclidianDistance}}} deliberately makes.