⇐ ⇒

[CF-metadata] [lc-cci] proposal for new CF standard names for land cover observation and classification

From: Jonathan Gregory <j.m.gregory>
Date: Wed, 28 Mar 2012 09:17:19 +0100

Dear Martin

I don't think that anyone has responded on the CF email list to your detailed
email at the end of Jan, have they?
(in the archive at:
http://mailman.cgd.ucar.edu/pipermail/cf-metadata/2012/027738.html)
That is a pity and I hope that there will be other responses. Thank you for
your carefully considered proposals. It might be there haven't been replies
because there is a lot to consider at once! Here are my comments on some
of your points:

> standard name: surface_reflectance
> definition: fraction of incident
> radiation<http://en.wikipedia.org/wiki/Reflection_%28physics%29>
> reflected by a surface, wavelength dependent

Is this different from surface_albedo, which we already have? surface_albedo
could have a coordinate of radiation_wavelength to make it spectral.

> standard name: probability_of_fire_occurrence
> definition: probability that the observed pixel shows a burned area
> unit: none (a value between 0 and 1)

If I remember correctly, there is an approved proposal for burned area
with a different standard name, that is going to be added to the table.

> standard name: probability_of_snow_occurrence
> definition: probability that the observed pixel shows snow or ice
> unit: none (a value between 0 and 1)

I think this would be a bit imprecise because it doesn't say exactly what
"snow" means (snowfall or lying snow) and it doesn't say what the threshold
is. I would suggest something more explicit such as
  probability_of_surface_snow_amount_above_threshold
with a coordinate or scalar coordinate of surface_snow_amount (kg m-2) to
specify the threshold; this could be a multivalued coordinate variable if
there are several thresholds. There are other possibilities such as
lwe_thickness_of_surface_snow_amount (m) - it depends what you want. Such
names would be consistent in style with existing ones like
  number_of_days_with_air_temperature_above_threshold

> standard name: probability_of_water_occurrence
> definition: probability that the observed pixel shows water
> unit: none (a value between 0 and 1)

I am not sure what this means, to be honest. Again, it would be good to make
it more precise by building a probability name from an existing (or a new)
standard name for the quantity on which a threshold is set.

> For the existing standard name land_cover which is a synonym of
> area_type we propose to allow for external vocabularies as valids.

Yes, use of external vocabularies is fine if they are standardised, available
online and well-maintained. However, as you say, CF maintains its own area_type
table. Therefore would it be better to introduce a new standard_name
specifically for land cover types of the UN Land Cover Classification System?
I presume that system has some governance arrangement of its own. If we had
a different standard name for it we would avoid conflicts and overlaps.

> standard name modifier: confidence_level
> definition: applied to some discrete variable, denotes the probability that the assignment of the value to the variable is correct

Yes, I think that would be an appropriate use of a standard_name modifier.

> standard name modifier: consistence_level
> definition: applied to some variable, with relation to some other
> variable, denotes level of consistency within the dataset

I am unclear about this, since it is not a quantity I have encountered before.
Does it have a precise definition?

> standard name modifier: source_flag
> definition: applied to some variable, denotes information source
> used to derive the value

Yes, I think this too could be a standard_name modifier, but I would suggest
"source", since there is an attribute of that name, with a similar intent.
Logically variables with a source modifier would be strings, but as usual
they could be encoded as flags with flag_values and flag_meanings.

Additions to the list of standard_name modifiers in the CF standard Appendix C
is a modification to the CF convention, not the standard name table. To propose
to modify the CF convention, you should open a new trac ticket
http://cf-pcmdi.llnl.gov/trac/query?status=new&status=assigned&status=reopened&status=closed&order=id&desc=1
and specify in the ticket the modification to the text of the standard. There
is a template for trac tickets to modify the convention. If you haven't got a
CF trac account, I believe Jeff Painter could help (painter1 at llnl.gov).

Best wishes

Jonathan
Received on Wed Mar 28 2012 - 02:17:19 BST

This archive was generated by hypermail 2.3.0 : Tue Sep 13 2022 - 23:02:41 BST

⇐ ⇒