DMS in AM3 - update ancillary file

Atmosphere AM3
1

Hi AM3 team, (and @Matt_Woodhouse, @BecJackson @alanah.chapman @ZhangchengPei )

I’ve just been poking around the AM3 ancillary files.

It looks like AM3 (N96 with interactive aerosol) is using the Lana et al. DMS climatology. This climatology is the standard, though a new one was released a few years ago now and my testing showed that it did make a bit of a difference to the cloud/radiation bias over the SO in AM2, and that some of the negative impacts that it produced were moderated by turning on the PMO aerosols (they are already on for AM3). In addition to my own paper ( ACP - The ACCESS-AM2 climate model underestimates aerosol concentration in the Southern Ocean; improving aerosol representation could be problematic for the global energy balance ) there is a bit of othr literature around showing that this update performs well.

Anyway, I’m wondering what a pathway could be to test/get the new DMS climatology (Hulswar) into AM3? I’ve got the 1x1 deg file - which I have regridded to the n96 grid, but perhaps the NRI team can do a better job at regridding than I did… :slight_smile:

I’m happy to run a few tests, or is it better of this is done in house?

2

ALLLSO - I’ve found similar improvments when I updated the chlorophyll fields to something realistic that @dougiesquire sent me. But that was in rAM3 not AM3 - so hasn’t been tested for global impact. But perhaps something that could be done simultaneously?

3

Hi Sonya,

We have the CCI-Ancillary-Suite which has been used to create ancillaries for experimental AM3 runs e.g. W21C’s n512e runs. You wouldn’t need to run the whole suite, but take a look at the ancil_clim_dms task- that’s the method used to create the DMS ancils for AM3 thus far. Should be able to take the python script there and use it with your DMS file and the existing AM3 landmask.

5

To add to what @lachlanswhyborn said, for the moment, the n96 suite is using CMIP6 ancillaries. We will be updating to CMIP7 ancillaries in the coming months but we are also interested to hear from the community on the datasets they are interested to use.

@sonyafiddes , if you don’t mind doing some tests using the python script Lachlan mentioned that would be great. I’ve opened issues about the DMS data source and the chlorophyll fields.

Just to make sure, is the chlorophyll ancillary file the same as the sea colour?

6

Hi Claire & Lachlan,

Thanks for that - I’ve cloned the CCI suite and have made the changes to the DMS and clim sea tasks, however, running into issues from the get go to start the suite.

The default: GRID_FILE="/scratch/rp23/lw5085/access-om3-025deg-ESMFmesh.nc"is stored on rp23 - which isn’t listed as a requirement here: GitHub - ACCESS-NRI/CCI-Ancillary-Suite: A Cylc version of the CCI ancillary suite. Admins: @whyborn and @bschroeter · GitHub - could it be moved into one of the other projects, or should rp23 be added to the list of required permission (or should I just request it)? Is this (along with GRID_SOURCE="ocean_mesh" recommended for AM3 ancillary generation? I’ve changed to running with ‘land_cover’ and using the grid file - but running into problems with that too:

For the ancil_lct task: I’m getting straight away

/scratch/jk72/slf563/cylc-run/CCI-Ancillary-Suite/run1/share/data/preproc/ESACCI-LCCS-and-Water-Bodies.nc does not exist.So it looks like it’s missing a pre-processing step

When I looked in the flow.cylc I see:

                        {% if GRID_SOURCE == "ocean_mesh" %}

                                (landseamask_from_ocean_mesh & preprocess_CCI_water) => ancil_lct

            {% elif GRID_SOURCE == "land_fractions" %}

                (landseamask_from_land_fractions & preprocess_CCI_water) => ancil_lct

But nothing calling preprocess_CCI_water for land_cover.. And I’m not sure where the landseamask would come from for the land_cover option…?

Any - I’m thinking I have missed something here in my setup? Any advise to get this running?

Thanks!

7

The ancillary suite is very much a work in progress- I’d recommend just taking the land mask (the target qrparm.mask from install_ancil) from your simulation and your DMS file and use ancil_general_regrid.py. The mistake in the Cylc graph has been fixed, and the ocean mesh target has been changed to a recent OM3 mesh on vk83.

8

Ok thanks for that Lachlan - I will give that a go!

9

Hi @sonyafiddes, not sure if you were also generating new chlorophyll ancillaries, but the workflow had an error in the chrolophyll task- it was using the land mask, rather than the sea mask (they’re not just inversions of eachother, due to mixed land/ocean points). It’s now been fixed (thanks to @ashjbarnes for picking that up).

10

thanks for speedy fix @lachlanswhyborn !!

11

Ah ok thanks Lachlan - I did reproduce the chlorophyll ancillaries - probably with the wrong mask. Will regenerate them :slight_smile: You can see what I produced here- will update when I get to it. Chlorophyll field · Issue #45 · ACCESS-NRI/CCI-Ancillary-Suite · GitHub