vopikamm / dino Goto Github PK
View Code? Open in Web Editor NEWA DIabatic NeverwOrld2 in NEMO 4.2.0.
License: MIT License
A DIabatic NeverwOrld2 in NEMO 4.2.0.
License: MIT License
Needed for experiments vopikamm/dino-experiments#1 . When choosing a less steep Scottia-arc, differences between the western and eastern channel bathymetry become bigger, violating a smooth periodicity.
Idea: Use the central gauss-ring profile (western boundary) to taper the entire channel --> periodic bathymetry
Right now, with every restart, the domain_cfg is rebuild from scratch (~ 8GB of data being computed)
Backed up by Wolfe & Cessi (2011) and crude estimates from the IPSL climate model, the SST should always be warmer in the northern hemisphere.
The solar forcing was taken from the Gyre configuration. There, the domain extends from 15°N to 50°N such that the solar forcing results in a downward heatflux everywhere. In DINO this results in a negative heatflux north of ~60°N: See Caneill.
Suggestion: Remove solar radiation, but include its effects in the temperature restoring T*.
All usrdef
files need some cleaning up with old and unused functions.
Diagnosing the monthly MLD of a DINO test-run suggests a bug in the forcing. The MLD appears to have an semi-annual cycle, e.g. January = July.
The retroaction term T* from Romain (low meridional gradient at boundary) demands for negative meridional S* gradients to have a meridional surface density gradient supporting the ACC. Choosing T* from Munday with seasonal cycle and S* with adapted and asymmetric boundary values. Choose salt minimum on the equator.
Adding a detailed description of the DINO configuration: Summary of bathymetry, forcings, default namelist, etc.
From meeting with Feiyu Lu. Having temperature restoring to zonal profile will decrease variability at the surface for high resolution runs (But should reflect a coupling to the atmosphere sst).
Downside: qns does not represent lwr, latent, sensible heatflux depending on the sst fields, but is simply imposed (check with Julie how this is done in climate models)
Right now spin-up takes forever, because all deepwater needs to be formed. Create an initial case which accounts for this.
Romain already has implementation init_case=4, but needs to be tested:
--> start short experiment with init_case=4 and check densest water ~ 1027
Need to implement Zanna & Bolton in NEMO.
Currently NEMO does not accept analytical forcings from usrdef_sbc.F90
AND forcing fields read in fldread.F90
. The idea is to have the functionality of fldread.F90
implemented in the usrdef_sbc.F90
.
Starting the same experiment with a different value for rn_e1_deg
changes the total volume of the domain:
rn_e1_deg = 1.
--> V = 2.70686566e+15
rn_e1_deg = 0.25
--> V = 2.77492743e+15
This (at least most of the difference) comes from the fact that the volume of the total domain stays (nearly) the same, but the masked boundary is wider the coarser the grid. Thus the ration land/ocean decreases with horizontal resolution.
Ideas to solve:
L_y
, L_x
, rn_e1_deg
If doing a true conservative interpolation of the inner points (also conserving volume) and carefully compute first and last row in i-direction? Or conservative interpolation from LR --> HR with small error already sufficient? (slight heating of the domain with higher resolution...)
--> Option 2
While
xgcm.grid.average(EmP, ['X', 'Y']) ~ 1e-21
but
xgcm.grid.integrate(EmP, ['X', 'Y']) ~ 1
because of the very large extend of the domain. This leads to ~ 3g/s mass-flux.
--> 1e8 kg in 1000 years
If the subdomain of mpi is smaller than the sill, the eastern boundary of that subdomain. This issue first appeared in nemo_4.2.1, but might be coincidence.
If the domain_cfg_out
and mesh_mask
are generated on one processor (or few) and read from file (issue #17 ) this issue is patched.
As a result from the experiment in vopikamm/dino-experiments/issues/4 : surface freshwater flux in the form of a SSS-restoring does not allow for a sufficient meridional density gradient and outcropping isopycnals in the channel. Go to a E-P fresh-water flux as in Wolfe & Cessi 2014. But adapt to values from IPSL climate model E-P output.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.