search

JMMP-Group / GO_coordination

Repository to coordinate the joint development of GOSI configurations
6 stars 0 forks source link

GOSI10: Improve initial conditions #3

Open oceanaguiar opened 10 months ago

oceanaguiar commented 10 months ago
atb299 commented 10 months ago

I've just recalculated the 1995-2014 climatology and regridded onto the eORCA025 grid. It looks like the pattern of strange zeros is not in EN4 (v4.2.2), but was introduced during the regridding (maybe a vertical regridding issue, or was there a check performed that set temperatures below freezing to zero?). It is present in both the model initial conditions and the dataset used by marine_assess.

The other undesirable features (warm coastal seas, Canadian Archipelago and Hudson Bay) are part of EN4 itself, so I think we should switch to using WOA23 and/or patch in the PHC climatology for the Arctic.

The sss climatology that forced models are restored towards should also be included in this task.

Original: Image

Recalculated: Image

jdha commented 10 months ago

RE short spin up - for ref the code we're using in SE-NEMO to spin up the ice is currently here for 4.0.4: https://github.com/NOC-MSM/SE-NEMO/tree/master/MY_SRC

dtatsd.F90 in_out_manager.F90 iom.F90 istate.F90 par_oce.F90

or here for 4.2.1: https://github.com/NOC-MSM/SE-NEMO/tree/115-nemo-4.2/MY_SRC/4.2.1

it uses a namelist flag ln_reset_ts to reset the temp and salt to initial conditions when a restart is called.

oceandie commented 5 months ago

According to OMIP-1 (Griffies et al. 2016) and OMIP-2 (Tsujino et al. 2020) protocols, OMIP simulations are initialised with temperature and salinity taken from WOA13.v2:

Potential temperature and practical salinity are initialized in the upper 1500 m using January observational-based climatology from version 2 of Locarnini et al. (2013) for temperature and Zweng et al. (2013) for salinity. We refer to these data as WOA13v2. The January fields from WOA13v2 stop at 1500 m. We filled deeper depths with the mean from January/February/March (winter in the north and summer in the south). Initial conditions are provided at the OMIP website, with both 1 ◦ and 1/4 ◦ versions. Interpolation should be made to the respective model grid. Conversion to Conservative Temperature and Absolute Salinity should be made for models based on IOC et al. (2010).

The initial temperature and salinity fields are therefore each a composit of two different variants:

1) upper 1500m: January mean (averaged decades between 1955-2012) 2) below 1500m: seasonal mean (Jan-Feb-Mar; all decades)

The Geomar CMIP6-OMIP/OMIP-input repository provides the workflow and the code to generate an initial ocean state in agreement with OMIP protocol.

oceandie commented 4 months ago

HI @AlexM62 , could you please post here the plots you were showing today at the SE-NEMO catchup regarding the EN4 trend? I thought they were quite interesting and it would be usefull to have them ...

AlexM62 commented 4 months ago
AlexM62 commented 4 months ago

And here is the other one I showed, with the mean Atlantic overturning streamfunctions for GS1p0, GS1p1 and GS1p2, along with the Hovmoller plots of the streamfunction anomalies at 1046m. SE-NEMO AMOC vs z.pdf

oceandie commented 4 months ago

Hi @AlexM62, thanks, but I was meaning the ones on EN4 water column trend ...

AlexM62 commented 4 months ago

OK, here you are: EN4 T & S drifts from 1958.pdf

atb299 commented 4 months ago

@AlexM62 , I was just looking at your EN4 T & S drifts from 1958.pdf. I think you have mis-labelled the plots. The left two look like they are Global and the right two are Atlantic (not top/bottom as currently labelled). Also, are they meant to be drift from 1958? If so, something doesn't look quite right - none of the plots start from 0 on the LHS. Have you subtracted the time mean instead?

AlexM62 commented 4 months ago

Thanks, Adam - your eyes are keen for a Friday afternoon! I realised I was using the first year of the EN4 dataset, which was 1950, rather than 1958, and the diagonal figures were indeed swapped.

Here is a corrected version with corrected reference year. EN4 T & S drifts from 1958.pdf

oceandie commented 3 months ago

A new initial condition for our GOSI configurations based on WOA13v2 and the OMIP protocol has been created with the code versioned in GO-iniTS repository.

Info about how this data were generated, how to replicate the generation process and how to use the new initial condition to initialise GOSI10 can be found in the README file.

An experiment using GOSI10p1.1 with the new initial condition has been launched (suite id = u-dg878@289604).

ukmo-cguiavarch commented 2 months ago

GOSI10p1.2 has been released using WOA13 initial conditions. WOA13 salinity is also used for the SSS restoring. The largest difference between WOA and EN4 concerned the salinity in the surface layers in Arctic. There are also differences in bottom temperature and salinity in the Red Sea and in the Persian Gulf. These are very local diffrences responsible of the large difference seen in the T/S section in the North part of the Indian Ocean domain.