[Proj] MGRS/USNG support?

[Oscar van Vlijmen]

>>> Let the projection code calculate what the nearest central meridian is,
>>> merely based on the input of the longitude of each point.

> From: Gerald I. Evenden
> UTM allows for a 1/2 degree overlap or extension beyond its normal
> longitude range.  Given that factor, how do you know which CM to use?
That's a shortcoming, but to calculate a 'regular' CM would be the best
thing to do.

> Similarly, if the point lies exactly on the division between zones, which
> CM do you use?
This is a matter of choosing one and documenting what the code does.
I believe the usual choice is zone 1 for lon=-180, zone 2 for lon=-172, ...
zone 31 for lon=0, zone 32 for lon=6, etcetera.

> Also, some zones in the Scandinavian region extend beyond the normal
> +/-3 degree limit, so again, explicit specification of the zone is required.
This is NOT required! It can be done based on latitude and longitude alone.
Others have done it, I have done it, so it can be done again.

Admittedly, usage of tmerc/utm in this fashion would be rare, but it
certainly is requested every now and then.
Some people do have lat/lon data without zone designations. If the UTM or GK
setup of a mutizonal region is highly regular -- without overlaps,
ambiguities, singularities etcetera -- an algorithmically derived CM can be
used without any problem.

> From: Frank Warmerdam
> And of course, there is no
> way of doing the same thing when going the other way.

This conversion is of course asymmetrical.