Standard names: Adding reference epoch sentence to anomaly terms

[feggleton]

Proposer's name Francesca Eggleton (not original proposer)
Date 20-10-2020

In #188 it was mentioned that we should add a sentence about using the reference_epoch into the definition of terms which are anomalies or related. The following terms were suggested:

air_pressure_anomaly | Pa | | 26
air_temperature_anomaly | K | | 25
brightness_temperature_anomaly | K | |
geopotential_height_anomaly | m | | 27
ratio_of_sea_water_potential_temperature_anomaly_to_relaxation_timescale | K s-1 | |
ratio_of_sea_water_practical_salinity_anomaly_to_relaxation_timescale | s-1 | |
sea_water_temperature_anomaly | K | |
surface_temperature_anomaly | K | |

change_over_time_in_sea_water_absolute_salinity
isccp_cloud_area_fraction
sea_water_absolute_salinity

The comment can be seen here: #188

Sentence suggested: "To specify the reference (baseline) epoch to which the quantity applies, provide a scalar coordinate variable with standard name reference_epoch."

Please add any other names or comments about this below.

[taylor13]

Although "climatology" is used in isccp_cloud_area_fraction, it simply expands the acronym: ISCCP = International Satellite Cloud Climatology Project. There is no climatological period needed or used in defining isccp_cloud_area_fraction.

Similarly, the use of climatology in the description of change_over_time_in_sea_water_absolute_salinity and sea_water_absolute_salinity may not imply a climatological period is needed in defining these quantities. Perhaps an ocean expert can comment.

[larsbarring]

@taylor13 I added those three standard names to the list based on a simple search for the term climatology in standard name descriptions, just so that any such standard name should not be accidentally overlooked. The intention was by no means to imply that they ought to have a reference period specified but rather to give the opportunity for others to step in and say, as you just did, that this is not necessary/relevant. As a matter of understanding (and not arguing with you!): as the specified "climatology" represents the "ISCCP epoch" could there be a value for data users to have this specified in the metadata even though the project as such is widely known?

@feggleton I think the suggested sentence to add is well crafted, but from @davidhassell's comment I got the impression that adding such a sentence to the definition of a standard name may have implications for the existing usage of the term. Is that right?

[taylor13]

Understood. Thanks for identifying all the variables that warranted some review.

Regarding the "ISCCP epoch", my understanding is that although the algorithms developed to determine the isccp_cloud_area_fraction were developed during a particular time, these algorithms do not depend to any significant degree on the climate of that period. They can be applied to any cloud field.

[larsbarring]

Thanks for clarifying, much appreciated! /Lars

[martinjuckes]

Does the reference_epoch have to be a scalar? E.g. I could have 3 time series in an array for 3 different reference times, float tanom(time,series);

float series(3):
    series:standard_name = "reference_epoch" ;
    series:flag_values = 1,2,3;
    series:flag_meanings = "tar_reference", "ar4_reference", "ar5_reference" ;

where TAR, AR4 and AR5 refer to different IPCC assessment reports, which used different reference periods.

Could the rule here be the same as for any other coordinate variable: the dimensions of the reference time for variable X should be a subset of the dimensions of X.

The definition of air_pressure_anomaly says that it is the anomaly relative to a climatology ... which implies that the reference could have 12 monthly values. This would require a coordinate variable with a climatology attribute. A note saying that bounds or climatology attributes should be used to specify the extent to the reference period(s) could be added.

[github-actions]

This issue has had no activity in the last 30 days. This is a reminder to please comment on standard name requests to assist with agreement and acceptance. Standard name moderators are also reminded to review @feggleton @japamment

[larsbarring]

@martinjuckes: I think that your examples/use cases make a lot of sense. As does the rule you refer to

... the same as for any other coordinate variable: the dimensions of the reference time for variable X should be a subset of the dimensions of X.

The description of teh reference_epoch standard name should then be expanded to reflect this wider usage.

[sethmcg]

I have a few additional standard names that take a reference_epoch:

keetch_byram_drought_index uses in its calculations the annual average precipitation, which is defined over some reference period.

modified_fosberg_fire_weather_index uses KBDI as an input, so it should inherit the reference_epoch from it.

nfdrs_severe_fire_danger_index is based on percentile values of ERC and BI, which need to be defined with regard to some (common) reference period.

What do we need to do to add some reference_epoch boilerplate added to the descriptions of those standard names?

[efisher008]

Hello @sethmcg @taylor13 @larsbarring,

I am coming back to this conversation after the migration of currently active and closed issues relating to standard names to the new vocabularies repo on GitHub.

The first possibility I could see would be adding reference_epoch to the CF phrasebank, which would mean it appears in the suggested text when adding a name containing that term to the CF editor. The obvious problem with this is that none of the standard names mentioned in this issue explicitly contain the term, meaning it would not be automatically picked up. Nevertheless, I have added reference_epoch to the phrasebank with the description:
"To specify the reference (baseline) epoch to which the quantity applies, provide a coordinate variable with standard name reference_epoch. As when specifying any other coordinate variable: the dimensions of the reference time for variable X should be a subset of the dimensions of X."
In the case that it will be of use when constructing a future name. Please note the 'scalar' descriptor has been removed here as per @martinjuckes comment.

I would suggest adding the above description to the names @feggleton has mentioned from issue #188 (I believe @roy-lowry and @larsbarring were involved in finding these):

• air_pressure_anomaly
• air_temperature_anomaly
• brightness_temperature_anomaly
• geopotential_height_anomaly
• ratio_of_sea_water_potential_temperature_anomaly_to_relaxation_timescale
• ratio_of_sea_water_practical_salinity_anomaly_to_relaxation_timescale
• sea_water_temperature_anomaly
• surface_temperature_anomaly
• change_over_time_in_sea_water_absolute_salinity
• isccp_cloud_area_fraction
• sea_water_absolute_salinity

as well as @sethmcg's names in the comment above this one:

• keetch_byram_drought_index
• modified_fosberg_fire_weather_index
• nfdrs_severe_fire_danger_index

Are there any objections to this? Happy to continue the discussion if I've misunderstood/oversimplified.

Best regards,
Ellie

[JonathanGregory]

Dear Ellie

Thanks for reviving this issue. I agree with your approach. In your proposed text, it would be helpful to mention the word "climatology" as well, since it's commonly used e.g. "To specify the reference (baseline or climatological) epoch ...". I think the "As specifying" sentence should include the possibility of there being a scalar reference epoch, which is the commonest case. I suggest, "... provide a coordinate variable with standard name reference_epoch, which could be either a scalar coordinate variable or have a size-one dimension, for a single reference epoch. The bounds of the reference_epoch variable provide the start and end of the reference period. It is possible to have more than one reference epoch, by giving the variable a dimension of size greater than one. As when specifying any other coordinate variable, the dimension of the reference time variable (if it is not a scalar) must also be a dimension of the data variable."

The text is very likely to apply to any standard name including anomaly, like most of those you list. I don't think we need to add this text to ratio_of_sea_water_potential_temperature_anomaly_to_relaxation_timescale ratio_of_sea_water_practical_salinity_anomaly_to_relaxation_timescale change_over_time_in_sea_water_absolute_salinity isccp_cloud_area_fraction sea_water_absolute_salinity, because none of these require a climatology to be interpreted. They're probably in the list because their description mentions "climatology".

Best wishes

Jonathan

[taylor13]

Regarding Jonathan's last paragraph immediately above, I don't think changes are needed to isccp_cloud_area_fraction; some of the descriptions for the other standard names Jonathan lists appear to me to not be perfectly well defined regarding whether they may or may not be relative to some climatology that hasn't already been fixed. Does anyone know for sure? [I'm not an ocean measurements expert.].

[sethmcg]

For the phrasebank, I think rather than saying "To specify the reference (baseline) epoch to which the quantity applies", I would say something like "To specify the reference (baseline) epoch used in deriving the quantity". An anomaly doesn't apply to a baseline, the baseline is used to calculate the anomaly.

Do we need the last sentence? I think it's a little confusing, and may not be correct: "As when specifying any other coordinate variable: the dimensions of the reference time for variable X should be a subset of the dimensions of X." The reference_epoch standard name says in its definition that it's a period of time, so if it has dimensions, they should always be [time, bounds], right? Are there any cases where you'd be using a different period of time at different points in space? Also, I can imagine cases where, in the process of calculating an anomaly, you lose the time dimension, so you might have a case where you have anomaly[lat, lon] and epoch[time, bounds], and that would still be meaningful...

[github-actions]

This issue has had no activity in the last 30 days. Accordingly:

• If you proposed this issue or have contributed to the
  discussion, please reply to any outstanding concerns.
• If there has been little or no discussion, please comment
  on this issue, to assist with reaching a decision.
• If the proposal seems to have come to a consensus, please
  wait for the moderators to take the next steps towards
  acceptance.

Standard name moderators are also reminded to review @feggleton @japamment @efisher008

[TomLav]

Hello. I am trying to revive this issue, not sure why it stopped.

As far as I can see, standard_name reference_period exists:

The period of time over which a parameter has been summarised (usually by averaging) in order to provide a reference (baseline) against which data has been compared. When a coordinate, scalar coordinate, or auxiliary coordinate variable with this standard name has bounds, then the bounds specify the beginning and end of the time period over which the reference was determined. If the reference represents an instant in time, rather than a period, then bounds may be omitted. It is not the time for which the actual measurements are valid; the standard name of time should be used for that.

However, there is 1) no invitation for data providers to use this standard_name when providing _anomaly variables, 2) no documentation (practical examples) for how to use reference_epoch.

I think this thread tries to address item 1) by adding mentions of reference_period from the definition of some standard names (mainly those ending with _anomaly, plus others). But it stopped.

For item 2) we would probably need 1-2 good examples in the CF Convention document.

I note that there are other on-going discussions for improving the way CF handles anomalies (e.g. https://github.com/orgs/cf-convention/discussions/305), but the one here (using reference_epoch) seems to be the current accepted way (although it is not fully documented).

[larsbarring]

Thanks for reviving this @TomLav.

When now quickly reviewing the comments above I do agree with @JonathanGregory and @taylor13 to exclude the following standard names:
ratio_of_sea_water_potential_temperature_anomaly_to_relaxation_timescale
ratio_of_sea_water_practical_salinity_anomaly_to_relaxation_timescale
change_over_time_in_sea_water_absolute_salinity
isccp_cloud_area_fraction
sea_water_absolute_salinity

[taylor13]

(The edits I made immediately above were simply entering line breaks to the list of names for readability.)