Arctic/Antarctic sea ice extent (seasonal cycle) REF

[axel-lauer]

This recipe adds a seasonal cycle diagnostic for the Arctic and Antarctic sea ice area as well as time series for the Arctic September and Antarctic March sea ice area.

The new recipe (#3891) requires:

• the updated OSI-450 (aka OSISAF/CCI) sea ice concentrations: #3963

Pull request: #3891

[TomLav]

Dear ESMValTool community,

I am the PI for the ESA CCI Sea Ice Concentration data (#3821) and the OSISAF/CCI Sea Ice Concentration data.

I am trying to get these two datasets (with a priority on the OSISAF/CCI data) into the REF. So far, I have been told to upload my data through Obs4MIPS and ESGF and I will continue this effort. At the same time, it is now clear to me that the two sea-ice metrics of REF will be implemented through ESMValTool, so it is relevant for me to find out what is already in place and how will the metric work from my files in practice.

I have some questions:

The OSISAF/CCI Sea Ice data is longer and more up-to-date than the ESACCI-SEAICE data and would be a more natural choice for REF. Both have the same file format (except the grid is 25km, not 12.5km). Could the CMORizer for ESACCI-SEAICE be re-used for the OSISAF/CCI dataset?

Once the CMORizer has run on my data, could they be published as-is via Obs4MIPS on the ESGF or are they for internal use in ESMValTool?

I am happy to move this dicussion to emails or another github issue if preferable.

I am looking forward to help if I can
Thomas

[axel-lauer]

So far, we thought about using the ESACCI-SEAICE dataset ("High(er) Resolution Sea Ice Concentration Climate Data Record Version 3 (SSM/I and SSMIS) from the ESA CCI+", version 3.0). If the format of the OSISAF data is the same (except for horizontal resolution), adapting the existing reformatting script should be quite simple.

Is there a preferred download option for these data? I could then give this a try.

The data reformatted for ESMValTool is a "CMOR-like" format that is already quite close to the obs4MIPs format. We would have to check what is needed to get the metadata to fully comply with the obs4MIPs specifications. At the moment, ESMValTool uses the obs4MIPs CMOR tables v2.1 (I believe), so that would need to be updated to v2.5. @ESMValGroup/technical-lead-development-team would such an update of the obs4MIPs CMOR tables be easy to do?

[TomLav]

So far, we thought about using the ESACCI-SEAICE dataset ("High(er) Resolution Sea Ice Concentration Climate Data Record Version 3 (SSM/I and SSMIS) from the ESA CCI+", version 3.0).

The ESACCI-SEAICE dataset will be a great choice when wanting to assess higher (spatial and temporal) resolution simulations. But it is currently limited to 1991-2020 with no committed extension. Ideally you will keep related software in the repo for future use beyond REF.

The OSISAF/CCI covers 1978-present and is continuously updated. Its previous version was also used in CMIP6 and AR6.

If the format of the OSISAF data is the same (except for horizontal resolution), adapting the existing reformatting script should be quite simple.
Is there a preferred download option for these data? I could then give this a try.

Thank you. I appreciate that you want to give this a try first. The data can be accessed either via anonymous ftp:
ftp://osisaf.met.no/reprocessed/ice/conc/v3p0
or through opendap/thredds:
https://thredds.met.no/thredds/catalog/osisaf/met.no/reprocessed/ice/conc_450a_files/catalog.html

The OSISAF/CCI dataset has both daily .nc files (similar to the ESACCI-SEAICE) but also monthly SIC mean files (under monthly/). This choice will depend on what the ESMValTool will do the metric : compute daily sea-ice area then monthly sea-ice area, or compute monthly sea-ice concentration, then monthly sea-ice area.

Let me know if I can help further.

The data reformatted for ESMValTool is a "CMOR-like" format that is already quite close to the obs4MIPs format. We would have to check what is needed to get the metadata to fully comply with the obs4MIPs specifications. At the moment, ESMValTool uses the obs4MIPs CMOR tables v2.1 (I believe), so that would need to be updated to v2.5. @ESMValGroup/technical-lead-development-team would such an update of the obs4MIPs CMOR tables be easy to do?

I understand. I certainly don't want to put extra work on your shoulders right now. I was asking out of curiosity as I am a bit lost between CMOR, ODS, obs4MIPS, etc...

[TomLav]

Also, if I may, in order to help the community keep track of the different SIC datasets produced in Europe (no only sea-ice concentration), I would suggest that the "ESACCI-SEAICE" dataset is renamed to "SICCI-HR-SIC" in ESMValTool. SICCI-HR-SIC is how we (the data producer) refer to this dataset in publications and reports.

SICCI-HR-SIC:
"SICCI" = "Sea Ice CCI", "HR" = "High(er) Resolution", "SIC" = "Sea Ice Concentration".

Thank you.

[bouweandela]

@ESMValGroup/technical-lead-development-team would such an update of the obs4MIPs CMOR tables be easy to do?

Yes, we would just need to copy an up to date version of the tables into ESMValCore and make a release. To use a newer version of the tables without changing ESMValCore, you could use the cmor_path variable in the config-developer.yml file to point to a newer version of the tables.

[bouweandela]

We would have to check what is needed to get the metadata to fully comply with the obs4MIPs specifications.

I expect we will mostly need to add the ability to add the required global NetCDF attributes to the files. I started working on facilitating that in #3960.

[bouweandela]

At the moment, ESMValTool uses the obs4MIPs CMOR tables v2.1 (I believe), so that would need to be updated to v2.5.

Our copy of the obs4MIPs CMOR tables appears to be rather dated (6 years old), see here.

[axel-lauer]

I updated our CMORizer for the OSISAF/CCI datasets (aka OSI-450) to handle version 3 (#3963). These data are now used as reference data in the basic sea ice diagnostics for REF (#3891).