Surface density transformation

[eguil]

Compute density bined buoyancy flux, transformation and formation (a la Walin 1982) Create specific proc surface_transf.py Will require E-P (wfo) and net downward heat flux over oceans (fields TBD)

[durack1]

Perfect, I had a placeholder I was filling out titled "Surface flux transformation to density (E-P, heat flux to density)" however, this is much more descriptive.. I haven't read the Walin contribution, so will do. This is very useful background to me: http://doi.org/10.1111/j.2153-3490.1982.tb01806.x

This will be very insightful..

[eguil]

you may also want to read this one: Kevin Speer and Eli Tziperman, 1992: Rates of Water Mass Formation in the North Atlantic Ocean. J. Phys. Oceanogr., 22, 93–104. doi: http://dx.doi.org/10.1175/1520-0485(1992)0220093:ROWMFI2.0.CO;2

[durack1]

Great, and this one will also be useful when we're considering the Southern Ocean:

Kevin Speer, Stephen R. Rintoul, and Bernadette Sloyan, 2000: The Diabatic Deacon Cell*. J. Phys. Oceanogr., 30, 3212–3222. doi: http://dx.doi.org/10.1175/1520-0485(2000)0303212:TDDC2.0.CO;2

[durack1]

Yeah, the git web submission for issues breaks the URL.. Trying this through the email parser (which seemed to work for you):

http://doi.org/10.1175/1520-0485(2000)0303212:TDDC2.0.CO;2

[eguil]

ok - I know this paper. We wrote a relate one with Kevin in 2000: http://onlinelibrary.wiley.com/doi/10.1034/j.1600-0870.2000.01060.x/abstract I've also found a couple of papers by Groeskamp et al (JPO 2014) who extend this approach to the combined T and S space. More complex but quite interesting as well.

• http://journals.ametsoc.org/doi/abs/10.1175/JPO-D-13-0213.1
• http://journals.ametsoc.org/doi/abs/10.1175/JPO-D-14-0039.1

[eguil]

Hi Paul, From the CMIP5 CMOR table, I need the total downward heat flux: hfds which should be the sum of: rlds (LW), rsntds (net SW), hfls (LHF) and hfss (SHF) On /work/cmip5/historical/ocn/mo I only find 3 of them: rlds, hfls and hfss. There is a rsds (downwelling SW) but for one model and not usable as such.

Hence my question: do we have either hfds or rsntds on the ocean grid ? Or do we have to look for this on the atmosphere grid (but we will miss the heat flux underneath the ice) ? Cheers, Eric

[durack1]

Good questions.. I'll probably have to add another variable to the xml creation list.. If indeed they exist.. Do these exist in the document: http://cmip-pcmdi.llnl.gov/cmip5/docs/standard_output.pdf Not sure what the implications are if we need to mix and match atmos and ocean grid variables..? What will the regridding do to it?

[durack1]

I'll poll what we have and what we need and append to the list for atmos and ocean vars.. The new run should kick off on Sunday, so we should know what exists by mid next week

[durack1]

FYI, the listing of variables is at: https://github.com/durack1/cmip5/blob/master/make_cmip5_xml.py#L161-175

[eguil]

ok, let me know what you can find. Interpolating atmos fields is a pain as you have to remove any land value first (via some nearest neighbour extrapolation).

On 12/9/14 17:56, Paul J. Durack wrote:

I'll poll what we have and what we need and append to the list for atmos and ocean vars.. The new run should kick off on Sunday, so we should know what exists by mid next week

— Reply to this email directly or view it on GitHub https://github.com/eguil/Density_bining/issues/18#issuecomment-55422838.

[eguil]

The net heat flux requires: rlds, rlus, rsds, rsus, hfss, hfls which are all in your list of atm_vars. It would be quite something if the net heat flux on the ocean grid was not available ! Definitively something to add on the list asked by the WGOMD and Anna Pirani...

On 12/9/14 18:53, Paul J. Durack wrote:

FYI, the current listing of variables is at: https://github.com/durack1/cmip5/blob/master/make_cmip5_xml.py current lines 390-400

— Reply to this email directly or view it on GitHub https://github.com/eguil/Density_bining/issues/18#issuecomment-55430916.

[durack1]

The rsds variable generated by GFDL was duplicated as a downwelling radiation within the ocean itself (so both atmos and ocean variables), and is very weird.. Ok, so no edits required by me then.. If we have all the outputs being generated..

If you have suggestions for what should be on the CMIP6 ocean output list let me know, I'm contributing to the document which should be in a draft form by the end of the month

[eguil]

well, we have the output on the atmos grid if I understand correctly. Which creates 2 problems for us: interpolation on the ocean grid and sea-ice. For CMIP6 we definitively need the net heat flux entering the ocean...

[durack1]

Ok so it does appear that water and heat fluxes on the ocean grid were provided from CMIP5, I just haven't been indexing them:

Flux	long_name	CMOR name	XML status
Water	rainfall_flux	pr	missing
Water	water_evaporation_flux	evs	Indexed
Water	water_flux_into_sea_water	wfo	Indexed
Heat	surface_net_downward_longwave_flux	rlds	missing
Heat	surface_downward_latent_heat_flux	hfls	missing
Heat	surface_downward_sensible_heat_flux	hfss	missing
Heat	surface_downward_heat_flux_in_sea_water	hfds	missing

I'll add these all to the xml scour and we should see what we have in a week or so..

[durack1]

Ok should be indexing with commit https://github.com/durack1/cmip5/commit/d636eab348bf772c9e8692dbfe1f62ab5f8e5f3d - These files should be on the machine mid-next week.. It'd be good to check and then see what we have

[eguil]

Commit 2324093cee9c140d506fbbdcd84b136ef1e7687d ok for global density flux and transformation - tested on MPI-ESM-LR historical CPU for 150 yrs = 600 sec

[eguil]

Will need tauvo (183 xmls) and tauuo (183 xmls) for Ekman transport contribution - compared to hfds (150 xmls)

[durack1]

@eguil ok so I have drive_surface.py working, and have tested this with 1 input: cmip5.ACCESS1-0.historical.r1i1p1.mo.ocn.Omon.surfTrans.ver-1.nc

I've made edits to drive_surface.py and some cleanup to surface_transf.py (mostly cleaning up unused statements and replacing your path with mine) - the commit https://github.com/durack1/Density_bining/commit/123ac2a0a07f0eaf9b962832237ecbaaeaa48d2c seems to work for me..

Give it a go and see how far you get.. There are a couple of to-dos but this should get us 86 files or so to start, obviously pending hitting problems with models..

Over to you..

[durack1]

@eguil we'll likely need to add in some tests to make sure that these fields all line up temporally, as we're using 4 different variable inputs I'd be surprised if this doesn't trigger an issue at some point.. Oh and I didn't add in tauvo and tauuo but will do that once you've had a go and see how far you get..

[eguil]

Back to surface transformation with the arrival of Victor Estella-Perez (LOCEAN Post-doc). See related slack channel

[durack1]

@eguil great!