Sounds good John,
We'll need to discuss the approach here. I think there is probably a
great opportunity for some cross-discipline collaboration.
It may not be as simple as refactoring the transformations to handle
meteorological -> geographic grids. I have yet to get a firm answer to
this, but I have a suspicion that in some cases, it may make the most
sense to use the current implementation.
My assumption is that the forcings and observations used by
meteorological modelers were not necessarily converted to spherical
coordinates. Calculations and model runs are performed using spherical
math though. I am not certain about this, but if it is the case, the
most appropriate approach would be to assume the positions are
actually from the original datum of the populating data or model forcing.
Do you know, or have a contact who could definitively address this?