MagneticField fails at the poles

[Gary E. Miller]

I tried using your MagneticField, but I get bizzare results.

For example. If I am at the North Pole (90N), then the longitude
does not affect my location. Yet the calcualtor gives me wildly different
results based on the longitude?

$ MagneticField 
now 90N 0W 
1.05 88.08 1893.9 1893.6 34.7 56588.6 56620.3
now 90N 90W
-88.95 88.08 1893.9 34.7 -1893.6 56588.6 56620.3
now 90N 180W
-178.95 88.08 1893.9 -1893.6 -34.7 56588.6 56620.3
now 90N 90E
91.05 88.08 1893.9 -34.7 1893.6 56588.6 56620.3
now 90N 180E
-178.95 88.08 1893.9 -1893.6 -34.7 56588.6 56620.3

Those vary from -178.95 to 91.05, but should be identical.

Similar for the South Pole.

RGDS
GARY

[Charles Karney]

GeographicLib does not ignore the longitude at the poles. The
longitude is used to define the directions of north and east. See if
you can make sense of the results assuming the latitude 90N is
interpreted as (90-epsilon)N with epsilon -> 0+. Bug me again if you
still have questions.

[Gary E. Miller]

"GeographicLib does not ignore the longitude at the poles."

Agreed. Clearly obvious from the results.

But it computes wrong. The magnetic variation can not be all those different values at the exact same location.

What do you think the magnetic variation is at the Poles?

1.05, -88.95, -178.95, 91.05, or -178.95?

I want to use this to create a 5x5 degree table for bilinear interpolation. The same way geoid separation is done in most GNSS receivers. I can't do that when cells that should contain identical data do not.

[Gary E. Miller]

Ah, never mind. I just dre myself a map, maybe I get it now. Sorry for the confusion.