-
Notifications
You must be signed in to change notification settings - Fork 0
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Icepack #1
Icepack #1
Conversation
MJHarrison-GFDL
commented
Mar 13, 2021
Lemieux landfast ice
Corrected two runtime issues with missing openMP directives by combining expressions to eliminate the recently added variables that were missing from openMP directive lists. All answers are bitwise identical in all cases that compiled.
- These modifications enable ridging code for SIS2 C-grid configurations (not yet implemented for B-grid) using APIs provided by the Icepack module from the CICE consortium (https://github.com/CICE-Consortium/Icepack) - These APIs are reproduced in the config_src/external/Icepack directory in order to remove SIS2 dependency on Icepack by default. In order to activate Icepack, remove the config_src/external/Icepack diretory from your compile list, compile a shared Icepack library and link to SIS2. - Sea-ice ridging rate diagnostics (rdg_rate) report the fractional rate of sea-ice coverage decrease (s-1) due to the ridging scheme. - WARNING:Sea-ice enthalpy diagnostics indicate a decrease in numerical closure from roundoff (10-17) without ridging enabled to values greater than roundoff (10-7) in initial testing. Mass and salinity, however, are consistent to roundoff. The issue should be addressed before using this option to generate scientific results. - The default Icepack ridging options are enabled. Additional options can be enabled easily using the SIS_parameter file and calls to icepack_init_parameters
Matt, are you getting ice concentrations above 1.0? I am. It's going above 2.5 for me in the Bering domain. |
Kate, I only have enough OBC/river data to do a single day test. No errors reported within day 1. At what point in the run does this appear for you? seaice.stats (LEMIEUX_LANDFAST = True)
|
I am starting from a restart in late December, once some ice has formed. I'm getting the errors pretty much right away. Raf and Bob have an OBC file that goes through all of January. Can you ask them for it? |
I am also starting with initial conditions from 12/31/2011. Adjusted the date to 1/1/2012 and now able to run a 10 day test. I get 3 warnings from icepack starting around day 7 (ridge_shift) Ridging error: aice0 < 0 |
- This appears to be needed to avoid propagation of negative ice free categories. Residual ice excess values of order 10-5 are being renormalized.
Applied a re-normalization of part sizes after call to ridge_ice necessitated by small residual excess cover.
P.S. - Have you tried switching to higher order thickness advection and
continuity in SIS2? Doing so gave much thicker ice in my tests (5x max).
|
I have not tried other advection schemes - good idea. The ice fraction larger than one is not in the restart file. It is happening in ice_cat_transport, where m_ice started as 72.3 and goes to 73.6 while mh_ice started at 73.44 and goes to 72.94. The part_size array then gets filled in in the call to cell_ave_state_to_ice_state. In single-category ice models, you advect everything, then bring ice concentration back to 1.0 or less by brute force. @Hallberg-NOAA would know how SIS2 does it. |
OK, so the ice concentration greater than one is a cue to the ridging scheme to ridge. I'm still getting some small amount of ice concentrations over 1.0, but just barely so. |
Yes. I was seeing values O(10^-5) after calling ridging. I re-normalized before applying back to the ice state and that fixes the issue in my tests with no loss in precision. https://github.com/MJHarrison-GFDL/SIS2/blob/Icepack/src/ice_ridge.F90#L447 |
Hi,
The "old Russian scheme" that SIS2 was using for ridging just compresses
(and thickens) the ice starting with the thinnest categories as a crude
form of ridging, but it is very effective at resetting the concentrations
back to 1. If the IcePack ridging scheme is close to doing the job but does
not always give values less than 1, maybe we could consider applying the
"old Russian scheme" after the physically based IcePack ridging to clean up
these small overshoots.
- Bob Hallberg
Dr. Robert Hallberg
Oceanographer
NOAA Geophysical Fluid Dynamics Laboratory
NOAA GFDL || Phone: (609) 452-6508
Princeton University Forrestal Campus || Cell: (732) 599-0459
201 Forrestal Road || Fax: (609) 987-5063
Princeton, New Jersey 08540-6649 || Email: ***@***.***
…On Wed, Mar 17, 2021 at 4:48 PM Matthew Harrison ***@***.***> wrote:
Yes. I was seeing values O(10^-5) after calling ridging. I re-normalized
before applying back to the ice state and that fixes the issue in my tests
with no loss in precision.
https://github.com/MJHarrison-GFDL/SIS2/blob/Icepack/src/ice_ridge.F90#L447
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#1 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABR2BFP2U32RHI3QH4KGUMTTEEITFANCNFSM4ZD4GWQQ>
.
|