[CF-metadata] Request for standard names: sea ice age and fraction of first year ice

From: Thomas Lavergne <t.lavergne>
Date: Thu, 26 May 2011 07:38:05 -0000

Dear Laurent, Bruce, Jonathan, and all,

May I step in?

Laurent expresses a need for new standard names related to sea ice type, from a modellers point of view. I would like to take the opportunity to also discuss standard names for such quantities, but from a satellite observation point of view.

The reason I step-in, and not start a new discussion is because I guess both Laurent and I will use the same definition of what is first-year-ice (by opposition to multi-year-ice). The way we use these definitions to build our gridded products is, however, different.

Some satellite products we distribute via the MyOcean project (SIW TAC) are observations of sea ice types. However, unlike what Laurent's model computes, the satellite product is a classification where each non-land pixel in the grid is either "open_water", "first_year_ice", "multi_year_ice", or "ambiguous". The latter is for cases where we do not feel confident for deciding if the (ice) grid cell is first- or multi- year ice. In essence, the satellite product is a classification, very similar to e.g. a land-cover product, where each cell contains a code, that code corresponds to a class (with associated label and definition). All classes are exclusive from all others.

In my case (sea ice type classification), I wonder:
1) if an appropriate standard name could be sea_ice_type_classification.
2) Do you have previous experience with 'classifications' in CF?
3) if my 'classification' could re-use (in a more generic way) the structure of CF 'flags', with "flag_values", "flag_masks" and "flag_meanings". Since my data would be a 'sea ice type classification' (and not 'flags'), I would fancy using 'sea_ice_type_values', and 'sea_ice_type_meanings' (I would not use 'sea_ice_type_masks').

In the case of Laurent, Jonathan proposes:
> first_year_sea_ice_area_fraction

The main point here, is probably to agree if the variable holds a "first year ice fraction of total ice concentration", or a "concentration of first year ice" ("first year ice fraction of total cell area"):
first_year_sea_ice_area_fraction + multi_year_sea_ice_area_fraction = sea_ice_area_fraction
or
first_year_sea_ice_fraction * sea_ice_area_fraction = firt_year_sea_ice_area_fraction
and
multi_year_sea_ice_fraction * sea_ice_area_fraction = multi_year_sea_ice_area_fraction
(first_year_sea_ice_fraction + multi_year_sea_ice_fraction = 1)

Concerning the discussion if the standard name should define "first year sea ice age" only (always less than... 1 year!), or the "age of sea ice" (from 0 to X years), I guess both standard names will proove handy in the end. Fowler et al (at NSIDC) deliver a sea ice age product, based on sea ice motion estimates from satellites. Their gridded values are from 0 to 3 or 4 years, as far as I remember.

I hope this helps, and does not broaden too much the focus of Laurent's original request. We can start a different thread on CF generic "classifications" if you deem it more appropriate.

All the best,
Thomas
Received on Thu May 26 2011 - 01:38:05 BST