[CF-metadata] Putting the units in a CF standard name: area_fraction

From: Martin Juckes - UKRI STFC <martin.juckes>
Date: Thu, 31 Jan 2019 16:32:44 +0000

Hi Steve,


In the interests of clarity, could you say why the option I've proposed is not in your list?


I'm not convinced that adopting your choice 2. will promote common language among disciplines. It is a laudable aim, but I fell it would be better taking into account the way in which are standard is perceived by others.


regards,

Martin




________________________________
From: Steven Emmerson <emmerson at ucar.edu>
Sent: 31 January 2019 16:02
To: Juckes, Martin (STFC,RAL,RALSP)
Cc: CF-metadata (cf-metadata at cgd.ucar.edu)
Subject: Re: [CF-metadata] Putting the units in a CF standard name: area_fraction

Martin,

So, it would seem like the potential solutions to the problem you perceive are

  1. Not use the standard name "fraction" in variable names to accommodate people who are confused when the values are given in percent; or
  2. Use the standard name "fraction" and expect people to learn.

I favor #2 because it promotes a common language amongst disciplines.

Regards,
Steve Emmerson


On Wed, Jan 30, 2019 at 3:40 PM Martin Juckes - UKRI STFC <martin.juckes at stfc.ac.uk<mailto:martin.juckes at stfc.ac.uk>> wrote:
Hi Steve,


The issue is more that CF allows more freedom in the choice of units than many people expect from a "fraction".


A second problem, I think the problem is that I didn't explain the issue clearly. In the CMIP data request we are specifying that variables with standard name "area_fraction" should be given as percentages. This is allowed by the CF convention: an "area_fraction" can be 0.5 or 50%. The reason that percentages are being used is because "area_fraction" is being used like the proportion of land covered in grass, and people are used to having these as percentages rather than fractions. It is all perfectly correct as far as the convention goes, but people often interpret the use of "area_fraction" for a percentage as an error.


Given that we have the framework of allowing flexibility in the choice of units, I feel it would be better to avoid having the term "fraction" in the standard name, given that it is often interpreted as implying a specific choice for the units.


regards,

Martin


________________________________
From: Steven Emmerson <emmerson at ucar.edu<mailto:emmerson at ucar.edu>>
Sent: 30 January 2019 21:37
To: Juckes, Martin (STFC,RAL,RALSP)
Cc: CF-metadata (cf-metadata at cgd.ucar.edu<mailto:cf-metadata at cgd.ucar.edu>)
Subject: Re: [CF-metadata] Putting the units in a CF standard name: area_fraction

On Wed, Jan 30, 2019 at 12:54 PM Martin Juckes - UKRI STFC <martin.juckes at stfc.ac.uk<mailto:martin.juckes at stfc.ac.uk><mailto:martin.juckes at stfc.ac.uk<mailto:martin.juckes at stfc.ac.uk>>> wrote:

I'm afraid I don't understand your comment. When I search for "fraction" in the NIST document I find it defined as being a ratio, which is inconsistent with the current CF usage. The CF standard name concept "area_fraction" is not what NIST or others understand as a "fraction". I'm suggesting a change to remove this inconsistency.

Unless we're talking past one another, I'll have to disagree. The NIST unit for "mass fraction" is "1" -- even though it's a ratio. A fraction can be represented many ways. "1:2", "1/2", and "0.5" all represent the same fraction, for example.

Does the CF convention require a particular representation for a fraction?

Regards,
Steve Emmerson
Received on Thu Jan 31 2019 - 09:32:44 GMT