Scheduled Downtime
On Tuesday 24 October 2023 @ 5pm MT the forums will be in read only mode in preparation for the downtime. On Wednesday 25 October 2023 @ 5am MT, this website will be down for maintenance and expected to return online later in the morning.
Normal Operations
The forums are back online with normal operations. If you notice any issues or errors related to the forums, please reach out to help@ucar.edu

Model blows up near boundaries in the regional setup

milicak

Mehmet Ilicak
New Member
Hi,

I am trying to setup a high resolution Pan-Arctic domain. The average resolution is around 3 km.
I computed initial conditions for T,S, zeta and sea ice variables from 1/12 degree NEMO model.
The initial conditions look all right to me.
I computed the open boundary conditions using the PyCNAL_regridding where first I computed the rotated velocities
of the NEMO model, then the Pycnal_regridding computes the open boundary velocities (angle rotated back to model grid) by that code.

The model is forced by CORE-II, however I tried CORE-I as well, it only pushes the truncations errors couple of days later.

I attached my final parameter choices which it ran 15 days with out any truncatione error, then the model blows up, and I attached final day u-velocities.

I tried every trick in the book including changing horizontal Smag coefficients (both laplacian and biharmonic), changing time step down to 200 seconds.
Almost all the Barotropic/Coriolis computation options, even turning on the GM. I wanted no-GM at the beginning, but later to kill the instabilities/eddies etc, I turned it on.

The velocities looks like a Kelvin wave to me, and most of the time truncation errors are around bottom right corner (East Pacific side).

The things I didn't try so far are;
1) compute and balance total barotropic transport across the boundaries, but normally Flather should be good with that. And it is a hassle to do that.
2) nudging. I guess for this I need to generate 4D u,v variables across the domain and file size can be very big!

I am wondering if anybody has a suggestions.

Thanks in advance,

Best,
Mehmet
 

Attachments

  • Arctic_Uvel_day19.png
    Arctic_Uvel_day19.png
    160.6 KB · Views: 21
  • MOM_parameter_doc.all.txt
    162.1 KB · Views: 6

milicak

Mehmet Ilicak
New Member
In the log file af the first truncations;

MOM Date 1958/01/15 00:00:00 336: En 1.114021E-03, MaxCFL 0.52074, Mass 8.607662078862E+19, Salt 34.60174330824, Temp 2.10094083611
MOM Date 1958/01/15 00:00:00 Energy/Mass: 1.11402E-03 Truncations 45
Total Energy: 43754AC7CC7F51BA 9.5891144012274592E+16
Total Mass: 8.6076620788617511E+19, Change: -1.0354125049574693E+13 Error: 3.57163E+13 ( 4.1E-07)
Total Salt: 2.9784011373682970E+18, Change: 1.7292099627978608E+12 Error: 1.89087E+12 ( 6.3E-07)
Total Heat: 7.2192081549688832E+23, Change: -8.0230534841677824E+19 Error: 2.17917E+19 ( 3.0E-05)
MOM Date 1958/01/15 06:00:00 342: En 1.128140E-03, MaxCFL 0.42103, Mass 8.607661023689E+19, Salt 34.60176873247, Temp 2.10069297395
MOM Date 1958/01/15 06:00:00 Energy/Mass: 1.12814E-03 Truncations 31
Total Energy: 43758FDD7301CE62 9.7106492655461920E+16
Total Mass: 8.6076610236887810E+19, Change: -1.0551729686090432E+13 Error: 3.56995E+13 ( 4.1E-07)
Total Salt: 2.9784029606920658E+18, Change: 1.8233237687890693E+12 Error: 1.98291E+12 ( 6.7E-07)
Total Heat: 7.2183555714218033E+23, Change: -8.5258354708051296E+19 Error: 2.11339E+19 ( 2.9E-05)
MOM Date 1958/01/15 12:00:00 348: En 1.133138E-03, MaxCFL 0.49433, Mass 8.607663085287E+19, Salt 34.60179355703, Temp 2.10043684721
MOM Date 1958/01/15 12:00:00 Energy/Mass: 1.13314E-03 Truncations 52
Total Energy: 4375A851F6115A9E 9.7536711761177056E+16
Total Mass: 8.6076630852869046E+19, Change: 2.0615981234041871E+13 Error: 6.72750E+13 ( 7.8E-07)
Total Salt: 2.9784058108558019E+18, Change: 2.8501637363047407E+12 Error: 3.00913E+12 ( 1.0E-06)
Total Heat: 7.2174772029094995E+23, Change: -8.7836851230300897E+19 Error: 2.10834E+19 ( 2.9E-05)
MOM Date 1958/01/15 18:00:00 354: En 1.144607E-03, MaxCFL 0.43028, Mass 8.607672154985E+19, Salt 34.60181852840, Temp 2.10016971807
MOM Date 1958/01/15 18:00:00 Energy/Mass: 1.14461E-03 Truncations 93
Total Energy: 4375E071A38ECD12 9.8524051054317856E+16
 

milicak

Mehmet Ilicak
New Member
Before the truncations it is around 0.1 until the last day and it jumps to 0.3. I attached below couple of days before the first truncation;

MOM Date 1958/01/13 18:00:00 306: En 1.089934E-03, MaxCFL 0.10859, Mass 8.607667125698E+19, Salt 34.60162061730, Temp 2.10199602408
Total Energy: 4374D4EDF186C211 9.3817887339061520E+16
Total Mass: 8.6076671256976638E+19, Change: 1.7959655558548727E+13 Error: 6.47292E+13 ( 7.5E-07)
Total Salt: 2.9783923228343281E+18, Change: 2.8275029661620342E+12 Error: 2.99872E+12 ( 1.0E-06)
Total Heat: 7.2228382042677522E+23, Change: -6.3870526942917968E+19 Error: 2.16671E+19 ( 3.0E-05)
SIS Date 1958/01/14 00:00:00 312: Area 1.322467641960E+13 0.000000000000E+00, Mass 5.70631721714E+15 0.00000000000E+00
Total Ice Mass: 5.7063172171354110E+15, Change: 2.27074E+14 Error: 7.82009E-01 ( 1.4E-16)
Total Ice Salt: 1.7382302504307484E+10, Change: 6.82783E+08 Error: 2.53116E-06 ( 1.5E-16)
Total Ice Heat: -1.9952636619112590E+21, Change: -7.99997E+19 Error: -2.10662E+05 ( 1.1E-16)
MOM Date 1958/01/14 00:00:00 312: En 1.104036E-03, MaxCFL 0.11397, Mass 8.607667591717E+19, Salt 34.60164443602, Temp 2.10182543367
Total Energy: 437519ED9F21E767 9.5031726058665584E+16
Total Mass: 8.6076675917171507E+19, Change: 4.6601948513382461E+12 Error: 5.14310E+13 ( 6.0E-07)
Total Salt: 2.9783945343208832E+18, Change: 2.2114865546398711E+12 Error: 2.38163E+12 ( 8.0E-07)
Total Heat: 7.2222524158097289E+23, Change: -5.8578845802293060E+19 Error: 2.90418E+19 ( 4.0E-05)
MOM Date 1958/01/14 06:00:00 318: En 1.101871E-03, MaxCFL 0.12274, Mass 8.607667219487E+19, Salt 34.60166782222, Temp 2.10162440930
Total Energy: 43750F5594EA2F92 9.4845356094978336E+16
Total Mass: 8.6076672194871640E+19, Change: -3.7222998643510405E+12 Error: 4.33887E+13 ( 5.0E-07)
Total Salt: 2.9783964185289518E+18, Change: 1.8842080688007471E+12 Error: 2.05324E+12 ( 6.9E-07)
Total Heat: 7.2215613473971403E+23, Change: -6.9106841258886038E+19 Error: 2.38378E+19 ( 3.3E-05)
MOM Date 1958/01/14 12:00:00 324: En 1.107503E-03, MaxCFL 0.23948, Mass 8.607664833286E+19, Salt 34.60169410386, Temp 2.10139559853
Total Energy: 43752AE4944C2DBA 9.5330171837668256E+16
Total Mass: 8.6076648332859834E+19, Change: -2.3862011836729105E+13 Error: 2.29613E+13 ( 2.7E-07)
Total Salt: 2.9783978550994924E+18, Change: 1.4365705405015967E+12 Error: 1.60376E+12 ( 5.4E-07)
Total Heat: 7.2207731105092171E+23, Change: -7.8823688792258707E+19 Error: 1.69537E+19 ( 2.3E-05)
MOM Date 1958/01/14 18:00:00 330: En 1.113670E-03, MaxCFL 0.30131, Mass 8.607663114274E+19, Salt 34.60171905682, Temp 2.10117407103
Total Energy: 4375491038F2DEBE 9.5860936554703840E+16
Total Mass: 8.6076631142742557E+19, Change: -1.7190117261270078E+13 Error: 2.92485E+13 ( 3.4E-07)
Total Salt: 2.9783994081583345E+18, Change: 1.5530588416992095E+12 Error: 1.71759E+12 ( 5.8E-07)
Total Heat: 7.2200104603172997E+23, Change: -7.6265019191745167E+19 Error: 2.27581E+19 ( 3.2E-05)

This is with DT=250 seconds and I am trying with 100 seconds. However that is very small for 3 km resolution.
I want to run a CORE-2 cycle (60 years).
 
I was going to get help tracking down why I need a 300 second timestep at 6 km resolution in my Arctic, but everyone is busy.

Alistair and Enrique talked about the NWA options and found a way to pare them down from the OM4 options. It has helped my Arctic, with less mixing and a faster simulation. See the recent changes in my Arctic6 repo.

As for the OBCs, it seems the angle in the grid file for my Arctic is all wrong, zero for half the domain. This angle is not used in the model, but was used to create the OBC file, so my OBC velocities are wrong. Check your angle.
 

milicak

Mehmet Ilicak
New Member
Hi Kate and Alistair,

Because Alistair kept asking about the CFL, therefore I decided to change the grid and when I was exploring, I found a better representation.
I attached a figure of the old grid (left) and new grid (right) dx values. It is a little bit coarser in the Arctic; instead of 3.5 km to 5 km, but there are no small
values on the boundary, and now I can run with 720 seconds!
Although I have to mention that my PBL choices are different than the default values, I guess I carried over from one your Arctic setups Kate (probably an old one). I checked the Arctic6 setup, now you are using default EPBL similar to JRA values.
Also Kv=1e-4 instead of 1e-6, but the first value is the in the JRA input files.
I want to try default ePBL from JRA and smaller Kv, and hope for the best.

Thanks a lot for your help.

P.S. Yes I made sure the angles are correct in the hgrid file.
 

Attachments

  • old_grid_vs_new_grid.png
    old_grid_vs_new_grid.png
    158.1 KB · Views: 20
Top