[CF-metadata] new standard name request for pH

From: Jonathan Gregory <j.m.gregory>
Date: Mon, 11 May 2009 18:29:30 +0100

Dear Nan and Roy

I think it is a bit arbitrary whether one regards them as different quantities
or different methods of measurement, but now I am tending in the direction of
the latter because of what John last said about it. It is actually somewhere
between, because the answers are known to be systematically different with the
different methods. However, it turns out not to be possible to describe in a
phrase (even a long phrase) what the difference is. That is, if asked the
question in a seminar, the expert would say, "The quantities differ because of
various details in the way they are measured." If the difference is so
detailed and hard to summarise, it seems to me that it really is a matter of
technique, and not of different geophysical quantities. I would expect it to be
possible to say, in a phrase, what the *essence* is of the difference between
geophysical quantities, whereas that cannot necessarily be done for methods.
Consequently, I would say that if we introduce a new attribute to identify
methods of measurement, we would not expect it to be as self-explanatory as
standard names. Whereas it is reasonable to expect an average scientist to
understand what is meant by a standard name, it is not necessary or reasonable
for different measurement/calculation methods. They depend on domain expertise.
Does that sound sensible to you?

A new attribute would also be able to distinguish between cloud fraction from
cloudsat, calipso or isccp, which is the other example we've had. We've agreed
these intend to be the same geophysical quantity, but it is known or expected
that the retrievals from different instruments will not give exactly the same
answers. However, it is not possible to summarise simply what these differences
are or to give them a geophysical explanation. So this is again a matter of
methods of measurement, like pH scales, it seems to me.

A further purpose for this attribute, if we designed it generally enough, could
be to identify the raw quantities which we have discussed in another thread.
The standard name for something could be a measured voltage, for instance,
and the other attribute says what function this quantity has in a sensor. That
definitely refers to methods of measurement.

Regarding Nan's question, I think pH by all the definitions is a dimensionless
quantity. It is a logarithm, so it must be dimensionless. However, that
does imply that the quantity whose logarithm is being taken must be in certain
units, to get the right answer, numerically. That has to be recorded in the
definition of the quantity. I agree, that is awkward. There is some precedent
for this. For instance, we have some standard names which specify units of
decibels. These quantities are logarithms too, and involve a reference level,
with specified units, in their definition.

Best wishes

Jonathan
Received on Mon May 11 2009 - 11:29:30 BST