[CF-metadata] CMIP5 cryosphere standard names

From: Siobhan.O'farrell at csiro.au <Siobhan.O'farrell>
Date: Mon, 27 Sep 2010 17:28:27 +1000

Dear Jonathan and Alison,

I have been working through Alison's 10 points.

1) Agreed,

2) I think Jonathan's sub categories where snow_on_sea_ice and snow_free_sea_ice approach helps here, we definitely want the surface temperature over the ice portion of the grid cell rather than over the whole grid cell (using cell methods). Also the interface temperature between snow and ice is important to account for, as it is much warmer than the surface temperature due to the thermal insulation on snow so I am no two worried about the case when there is no snow cover but again the snow_on_sea ice (area type) would help to prevent the confusion.

3) Agreed

4) Agreed

5) Agreed

6) It makes sense to separate the water flux from snow and rainfall as the impact differently on the surface using the cell methods would indicate them onto the sea ice portion of the cell. Some rain may penetrate into the ice lattice if it s very porous Jonathan or contribute to melt ponds in the Arctic summer, but I don't know of a sea ice models IN CMIP5 that sophisticated to allow for that effect.

7) Agreed use ice floes

8) Sea ice thermal energy, I am a bit confused on the history on this one there were two requests originally made in my 2008 document sent To Karl and circulated through the cryosphere community.


The first was
Snow_thermal_energy_content Snth new Jm-2 month 1 all

For the land surface component. It was in an earlier version of Karl's document (I had saved 16/6/2009) but is not in the latest 17/9/2010.

The sea ice heat content was given as


integral_of_sea_ice_temperature_wrt_depth_expressed_as_heat_content hcice New Jm-2 Month 1 All


It looks like they retained the hcice name but changed the units.

So I am puzzled as well, I think it should be Jm-2.

The issue of whether the heat content of the snow over the ice should be included is highlighted, it is possible, but it would be preferable as separate variable, as the thermal properties are very different form the ice, and it could confuse interpretation of result if they were merged into one effect (Unless you are using the UM, Jonathan where the atmosphere model treats the two media as if they are combined).

9) I still support the need for a sea ice albedos, my comments on whether it is total or bare still stand, again it should be defined on the sea ice portion of the grid. Alison's suggestion of an area type snow_free_ice could help distinguish it is bare, but if the snow depth is also known, then that would be clearer, though there will be transition months as the ice becomes snow free.

So is one total sea ice albedo sufficient?

10) Agreed if they are all measured on the sea ice portion of the grid box (cell methods.


Best wishes
Siobhan


-----Original Message-----
From: Jonathan Gregory [mailto:jonathan at met.reading.ac.uk] On Behalf Of Jonathan Gregory
Sent: Saturday, 25 September 2010 4:58 AM
To: alison.pamment at stfc.ac.uk
Cc: cf-metadata at cgd.ucar.edu; O'Farrell, Siobhan (CMAR, Aspendale); taylor13 at llnl.gov
Subject: [CF-metadata] CMIP5 cryosphere standard names

Dear Alison

> Therefore, I think we
> should introduce a new standard name of
> surface_downward_heat_flux_in_snow (Wm-2). The CMIP5 output can then use
> a cell_methods of "area: mean where land". OK?

I agree.

> (a) Long name 'Surface Temperature of Sea Ice' with units of K and the
> following explanatory comments: 'When computing the time-mean here, the
> time-samples, weighted by the area of sea ice in the grid cell, are
> accumulated and then divided by the sum of the weights. Report as
> "missing" in regions free of sea ice. Note this will be the surface
> snow temperature in regions where snow covers the sea ice.'

> (b) Long name 'Temperature at Interface Between Sea Ice and Snow' with
> units of K and the following explanatory comments: 'When computing the
> time-mean here, the time-samples, weighted by the area of snow-covered
> sea ice in the grid cell, are accumulated and then divided by the sum of
> the weights. Report as "missing" in regions free of snow-covered sea
> ice.'

It seems to me that (a) is surface_temperature for area: where sea_ice,
and (b) is sea_ice_surface_temperature for area: where snow_on_sea_ice,
in which I have invented a new area_type. Both are time: mean.

> 6. Water flux names
>
> I think we should use the
> existing standard names of
> rainfall_flux; kg m-2 s-1
> snowfall_flux; kg m-2 s-1
> and supply a cell_methods attribute of "area: mean where sea_ice over
> sea" for each of them. Does that seem more sensible?

Yes, I agree. That seems to be exactly what the CMIP5 description says. Thanks.

> 8. Sea ice thermal energy
>
> According to the CMIP5 document, the unit is J and the long name is 'Sea
> Ice Total Heat Content'. The explanation says, 'Ice at 0 Celsius is
> assumed taken to have a heat content of 0 J. When averaging over time,
> this quantity is weighted by the mass of sea ice. Report as "missing"
> in regions free of snow on land.'
>
> I find this rather confusing - if it is supposed to be a sea ice
> quantity, then surely it should always be reported as "missing" over
> land and open sea. It doesn't sound as though latent heat is included.
> I don't think we can call it a 'content' because it isn't a quantity per
> unit area. Perhaps Siobhan can help to clarify this quantity further.

It is unusual to report a quantity extensive in area, isn't it. Things are
usually intensive i.e. m-2.

> 9. Bare Sea Ice Albedo
>
> Perhaps it would be more
> accurate to use the existing name 'sea_ice_albedo' and introduce a new
> area_type of 'snow_free_ice'.

I think that would be a correct description, yes. This area-type is the
counterpart of snow_on_sea_ice which I needed above. I think yours should
be snow_free_sea_ice.

> 10. X and y components of stress on sea ice surface and base
> We already have existing names
> surface_downward_x_stress; Pa
> surface_downward_y_stress; Pa
> which we can use for the first two quantities.
>
> I now propose that we introduce two new standard names
> upward_x_stress_at_sea_ice_base; Pa
> upward_y_stress_at_sea_ice_base; Pa.
>
> I think also that for CMIP5 these four quantities should be accompanied
> by a cell_methods attribute of "area: mean where sea_ice".
>
> Is this OK?

I would say so, yes.

Best wishes

Jonathan
Received on Mon Sep 27 2010 - 01:28:27 BST