[CF-metadata] CF surface variable names

[Jonathan Gregory]

Dear Karl

> Alternatively, a scalar dimension could be defined giving the height 
> above the surface from which visibility was measured.
For other quantities, we have a "surface" name and a name without surface.
This avoids having to include some rather contrived coordinate such as
height = 0 m to indicate the surface.

> 2. In the suggestions concerning direction, should we be explicit about 
> whether the direction is "toward" or "from" (as in winds from the NW or 
> winds toward the SE)? We could have 2 standard names (direction_toward 
> and direction_from) Also presumably direction would have to have units 
> something like "degrees east of north".

I think degrees should be OK. The definition of direction will state that
it means degrees clockwise from north. We are already using direction in
this sense in other standard_names.

> 3. Jonathan, I note that we have precipitation and snowfall amount, but 
> not rainfall amount.  I guess it's the difference between precip. and 
> snowfall.  I think (in the standard name table) our "help" comments 
> should indicate that precipitation amount, rate, etc. includes all forms 
> of precipitation, just to make sure that no one stores rainfall under 
> precipitation.  Do you think this is a good idea?

Yes. I will note that we should add that remark. We should also include
rainfall_amount anyway.

Cheers

Jonathan