Dear Alison,
Thank you, that is a good summary. I agree that it makes sense to treat "covariance" and "correlation" in the same way as "product".
Although these two terms appear in the guidelines, they don't yet appear in the standard name table itself, so the usage "covariance_over_longitude_of_northward_wind_and_air_temperature" will be a first for "covariance". Can we include a sentence in the description saying that "Covariance refers to the sample covariance rather than the population covariance"? I don't think it is likely that we will need to deal with population covariances in the CF standard name list, but, if we do, it would make sense to have a separate term for them.
Jonathan asked whether it would make better sense to use "covariance_wrt_..." rather than "convariance_over_". I think this would be an improvement: the covariance is an integral of a product and so using the same syntax as used for integrals makes sense here.
regards,
Martin
________________________________
From: CF-metadata <cf-metadata-bounces at cgd.ucar.edu> on behalf of Alison Pamment - UKRI STFC <alison.pamment at stfc.ac.uk>
Sent: 30 May 2019 14:12
To: cf-metadata at cgd.ucar.edu
Subject: Re: [CF-metadata] order of product in standard name
Dear Jonathan and Martin,
I would like to make sure that we have fully resolved this issue as it affects names that are currently under discussion in two threads:
covariance_over_longitude_of_northward_wind_and_air_temperature (latest comment:
http://mailman.cgd.ucar.edu/pipermail/cf-metadata/2019/020922.html);
also:
* product_of_eastward_wind_and_lagrangian_tendency_of_air_pressure
* product_of_northward_wind_and_lagrangian_tendency_of_air_pressure
* product_of_lagrangian_tendency_of_air_pressure_and_specific_humidity
* product_of_lagrangian_tendency_of_air_pressure_and_air_temperature
* product_of_lagrangian_tendency_of_air_pressure_and_geopotential_height
(latest comment
http://mailman.cgd.ucar.edu/pipermail/cf-metadata/2019/020927.html).
I agree with Jonathan that for covariance and correlation names it is better to have covariance_over_Z_of_X_and_Y, (rather than putting over_Z at the end as stated in the current guidelines). This is consistent with the way we now write integrals as integral_wrt_X_of_Y.
The current guidelines for correlation and covariance say simply that X and Y should be ordered alphabetically, whereas for product we have additional guidance: 'If X and Y are both scalars or both components of vectors, they are put in alphabetical order. If one of them is the component of a vector, it is put first i.e. the vector component is X, the scalar is Y.'
I think we are agreed that the same guidelines should apply to the ordering of X and Y in standard names for all commutative properties such as product, covariance, correlation, etc.
The rule about putting vector components first means that the covariance name as written above is in the correct order.
Regarding the lagrangian tendency names, I agree with Martin's argument that lagrangian_tendency_of_air_pressure should be regarded as a vertical component of the air velocity vector, rather than a scalar. Therefore it is correct to write the specific_humidity, air_temperature and geopotential_height names as above (this is the opposite way round to what was originally proposed).
I also agree with Martin's suggestion of ordering vector components as X=horizontal and Y=vertical when horizontal and vertical components are the operands. This is a useful general convention and gets around the problem of having product_of_eastward_wind_and_lagrangian_tendency_of_air_pressure and product_of_lagrangian_tendency_of_air_pressure_and_northward_wind which would arise from the alphabetical rule. (Of course if we are combining two horizontal or two vertical vector components we would still need to fall back on the alphabetical rule).
It is on my 'To Do' list in the next couple of months to conduct a thorough review of the standard names Guidelines document and the sections of the Conventions document covering standard names and units. They are all somewhat out of date. If we can agree these changes they can be included as part of that process.
Best wishes,
Alison
-----------------------------------------------------------------------------------------------------------
Alison Pamment Tel: +44 1235 778065
NCAS/Centre for Environmental Data Analysis Email: alison.pamment at stfc.ac.uk
STFC Rutherford Appleton Laboratory
R25, 2.22
Harwell Oxford, Didcot, OX11 0QX, U.K.
-----Original Message-----
From: CF-metadata <cf-metadata-bounces at cgd.ucar.edu> On Behalf Of Jonathan Gregory
Sent: 10 May 2019 15:08
To: cf-metadata at cgd.ucar.edu
Subject: [CF-metadata] order of product in standard name
Dear all
I should have noticed that we already have a guideline about product_of_X_and_Y, which should also apply to covariance, correlation, and any other commutative function of X and Y:
If X and Y are both scalars or both components of vectors, they are put in alphabetical order. If one of them is the component of a vector, it is put first i.e. the vector component is X, the scalar is Y.
Cheers
Jonathan
_______________________________________________
CF-metadata mailing list
CF-metadata at cgd.ucar.edu
http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
_______________________________________________
CF-metadata mailing list
CF-metadata at cgd.ucar.edu
http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
Received on Mon Jun 03 2019 - 09:25:22 BST