[CF-metadata] water level with/without datum

From: Jonathan Gregory <j.m.gregory>
Date: Tue, 23 Feb 2010 14:33:00 +0000

Dear Jeff

Thanks for your email. I appreciate your arguments, which are very reasonable,
but I don't agree with them so far.

> Replacing 'sea_' with something else seems like it would
> break much existing code. Adding some names should be mostly harmless.

Yes, adding names is better. We can resolve this by keeping "sea" to mean
"sea", and adding some other ones to refer to "sea, lake or river", as
requested. We agree on that point.

> For the specific case of water-level measurement devices, I think the term
> water_level_* is better, applied generically regardless of oceanic,
> lacustrine or riverine environment, for the following reasons.
>
> * There is no other generic name, and inventing one like SLR or
> sea+lake+river seems contrived.

Contrived, yes, but sea+lake+river is certainly explicit and self-explanatory,
isn't it? Standard names are contrived to explain what they mean, rather than
being the terms used most commonly (although some of them are common terms).
The term "name" is a bit misleading. They are not names, in most cases. They
are answers to the question, "What does that mean?", when a term is used.

> * Every use of sea_level I can find in the CF name list refers to sea level
> as a semi-constant reference point rather than as an instantaneous
> measurement. Indeed, a comment used repeatedly in the CF table is that
> "sea_level means mean sea level, which is close to the geoid in sea areas."
> Therefore, separate names for sea_level_*, lake_level_ and river_level_* do
> not seem appropriate.

Yes, sea_level refers to a fixed level, like geoid, but the quantity you are
referring to as water_level is more like sea_surface_height, which is a
time-varying level, and is referred to a fixed level. I am proposing
of sea+lake+river_surface_height_above_X for your water level, if it's
not the sea.

> * Yes, there is water in the atmosphere and underground. CF already seems
> to qualify those uses with terms like 'atmosphere', 'cloud' and 'in air.'
> Therefore, it does not seem necessary to qualify 'water' as being on the
> surface--simply retain the existing non-surface qualifiers. If you later
> need a name to refer to the level of water in an underground aquifer or
> something, then create one.

That gives a special status to water on the surface. CF names attempt to deal
even-handedly with all geophysical quantities. Such an argument would mean
that, in the first place, we would have used the plain name "temperature" to
mean air temperature (since we started with atmospheric models, mostly), and
later added sea_water_temperature. That would have been inconsistent and I
believe that such inconsistencies would make standard names less satisfactory
in practice. In general, we have tried to include some context in standard
names.

> Regarding the Use Case of measuring temperature in the sea and then
> continuing the trajectory upriver: there currently do not exist
> water_temperature names other than sea_water_temperature, so the Use Case
> is not satisfied at present. Perhaps in future CF could define additional
> generic names like water_temperature that may have specializations such as
> sea_water_temperature and fresh_water_temperature, but this can be done
> later and independently of the existing actual use case of water level
> measurements.

I think Roy's example is a relevant use case. Although he has not made a
proposal, his data set requires either a new name of river_water_temperature,
or a name which can be used for both sea and river. The existing name of
sea_water_temperature is not sufficient for the case he described.

Best wishes

Jonathan
Received on Tue Feb 23 2010 - 07:33:00 GMT