Informal "chat" thread about current coupled runs #455
Replies: 15 comments 19 replies
-
|
Based on today discussion, we are starting 2 branches: Cold branch / Warm branch
We will also continue the 64 perturbations
|
Beta Was this translation helpful? Give feedback.
-
|
From @adamrher : "It looks like the warm branch is too warm and the cold branch is way too cold (it already froze)." -> we need to retune ... |
Beta Was this translation helpful? Give feedback.
-
|
orange is cold branch, red the warm branch.
|
Beta Was this translation helpful? Give feedback.
-
|
SWCF and LWCF included:
|
Beta Was this translation helpful? Give feedback.
-
|
SWCF maps (first 20 years)
SST bias relative to HadISST PI (first 20 years)
|
Beta Was this translation helpful? Give feedback.
-
|
A few questions/discussion:
|
Beta Was this translation helpful? Give feedback.
-
|
Updates: Tuning of cold/warm branch:
Following the CESM project meeting:
|
Beta Was this translation helpful? Give feedback.
-
|
Mini-update. 75 cold leg run looking ok. RESTOM averages to zero. 12 years done. Note the lime green line (75) is the only one w/o 5-year running avg; it's the year-to-year value of RESTOM.
|
Beta Was this translation helpful? Give feedback.
-
|
run 75 is out to year 22, but I think it stopped. @cecilehannay can we make sure to resume that run?
|
Beta Was this translation helpful? Give feedback.
-
|
Hi @adamrher: I just resubmitted 75. As a rule of the thumb: By default, I stopped the run at 22 years (except when we specify to make it shorter) so we can look at the first 20 years and decide if the run is ok and we want to make it longer. |
Beta Was this translation helpful? Give feedback.
-
|
@adamrher: could you run your spaghetti plot for 64, 64e,f,g,h and 64intel? |
Beta Was this translation helpful? Give feedback.
-
|
Here's a current look. The only branch of 64 that froze over is 64e. Let's carry on with these experiments.
|
Beta Was this translation helpful? Give feedback.
-
|
I was looking at these in bit more detail this morning. I even added the relevant part of 26g where it freezes over as well:
(note I've changed up the colors). I think this plot emphasizes what you see in the annual plots, the global SST's are trending down for most of the simulations, but then around year 60-70 they start to reverse and trend up again. It looks like if a run can avoid freezing over during the coldest parts, year 40 to 70, then it's out of the wood work, at least until the next inter-decadal SST cooling. It is interesting how 64f (yellow) barely avoids freezing over (see LabSea plots), whereas 64e (purple) couldn't make it. 64g and 64h, having been initialized ten years earlier, cruise right by without much of a threat of freezing. |
Beta Was this translation helpful? Give feedback.
-
|
Here's the current status of the cold (75) and warm branch (76):
|
Beta Was this translation helpful? Give feedback.
-
|
Here's the latest spaghetti plot with Ben's run overlain. I am referring to it as run 77. I like how it doesn't have that big RESTOM bump in the beginning like we see in 76.
@cecilehannay should we open an issue for Ben's run so it's documented officially as run 77? |
Beta Was this translation helpful? Give feedback.
-
|
We have an issue for 77 alias b1850.076base.n2th1 at #484. |
Beta Was this translation helpful? Give feedback.
-
|
To take a closer look at the big RESTOM incursion around year 50 in run 77, here's a spaghetti plot w/o the 5 year running average for RESTOM. It doesn't look so abnormal on a year by year basis.
Here's the 5 year avg RESTOM plot, as a reminder:
|
Beta Was this translation helpful? Give feedback.
-
|
Urgent: I filled my quota (200T) and I needed to do a major clean up before I can run anything again. I can move some case to For Fcase, @PeterHjortLauritzen, @JulioTBacmeister, @swrneale, @adamrher, @bstephens82, @gustavo-marques, @cmcclus @dlawrenncar @megandevlan @islasimpson |
Beta Was this translation helpful? Give feedback.
-
|
can I make a request to keep let me know if these are too large to keep... |
Beta Was this translation helpful? Give feedback.
-
|
I'd request to keep at least That's at least where my initial focus has been; the full 64 series would probably be good to keep too though, space permitting. |
Beta Was this translation helpful? Give feedback.
-
|
Sorry to add something that's not on your list, but I am wondering if we could keep my "077" run (which has an alias), since that's the only long taus run we have? If so, the history files are here: /glade/derecho/scratch/stepheba/archive/b1850.076base.n2th1. If you do move it over to campaign storage, I assume you could rename the folder to the "official" 077 name. I looked at the 080 run and it looks like it has stopped at 22 years...are we going to continue that one? I don't know at this point if it's worth the resources or not considering the taus' disappointing performance at the Friday CAM7 meeting. It looks like we have not found the right area of parameter space yet, to me. |
Beta Was this translation helpful? Give feedback.
-
|
At the CAM7 meeting, Peter said it was up to you to decide if we want to continue 80 or not. I agree that it might be not worth the resources. as these a cross-working group activities. But if you have not reached your quota, you should not be worried about moving it on campaign today. The issue are the ones that are in my archive. |
Beta Was this translation helpful? Give feedback.
-
|
Ok I will make a note to check back in a week or two and possibly move it over then, thanks. I may comment on 80 on the other thread dedicated to that run. |
Beta Was this translation helpful? Give feedback.
-
|
Is the data from all these simulations roughly the same, and which of the h1/h2/h3 file sizes are dominating? |
Beta Was this translation helpful? Give feedback.
-
|
Roughly,
For atmosphere output:
Obviously, it would help to reduce ocean output but I don't have much control on that. |
Beta Was this translation helpful? Give feedback.
-
|
The questions might be whether the oceanographers want much of their data - probably not right? |
Beta Was this translation helpful? Give feedback.
-
|
I think we can delete every coupled simulation prior to 54. @gustavo-marques is currently storing run's 46, 47 (default vs hycom1 vertical coordinate) which I'd like to keep around since they are 100+ years long. He is also storing 45 and 55 (1/4deg MOM, and hycom1 version of 54). I can't think of a reason to keep any other pre-54 simulation around. I'd prefer to keep the F cases around, at least until they're scrubbed by Derecho. |
Beta Was this translation helpful? Give feedback.
-
|
@adamrher: your suggestion sounds good. |
Beta Was this translation helpful? Give feedback.
-
|
My plan is to save the runs we want to keep in the cross-working group space:: |
Beta Was this translation helpful? Give feedback.
-
|
I will delete It will allow me to restart the coupled simulation. |
Beta Was this translation helpful? Give feedback.
-
|
Are you still keeping the climo files for all the runs we have done. I think these are very useful as quite often we end circling back around to revisit tuning/changes we have already attempted and it is good to have these files for guidance. |
Beta Was this translation helpful? Give feedback.
-
|
Yes, I am keeping climos each time I ran the ADF in I believe @justin-richling is doing that too. |
Beta Was this translation helpful? Give feedback.
-
|
Frank wants to keep the 64 series to help us better understand the freezing of the Lab Sea. @cecilehannay, please let me know if you want me to store these cases in the OMWG storage space. |
Beta Was this translation helpful? Give feedback.
-
|
@gustavo-marques: My thinking is that the best location for these runs is in the cross-working group space: |
Beta Was this translation helpful? Give feedback.
-
This is a continuous discussion thread about current coupled runs (replacing some Google chat threads)
Beta Was this translation helpful? Give feedback.
All reactions