Spin up b #28
Spin up b #28
Conversation
…and Dave Bi mom and cice changes
…l mom5 version and Dave Bi's cice code changes (now updated)
|
The model version in the
|
|
!redeploy |
|
🚀 Deploying access-esm1.6 Details and usage instructionsThis
This Prerelease is accessible on Gadi using: module use /g/data/vk83/prerelease/modules
module load access-esm1p6/pr28-2where the binaries shall be on your 🛠️ Using: spack DetailsIt will be deployed using:
If this is not what was expected, commit changes to |
JhanSrbinovsky
temporarily deployed
to
Gadi Prerelease
GitHub Actions
Inactive
|
In these 2 simulations we use the same code base for the LSM/UM7. The script run uses Leuning, no soil thermal fix etc. An older cable.nml. The payu run updates all of these plus includes wombat-lite, generic tracers, Dave's iceberg fix. There is also a difference in netcdf versions, compiler versions. So on the one hand this comparison is like weather from 2 different models. It's a daily average screen Temp after a year. On the other hand, such a large difference scares me that we need to do a payu/script comparrson. Which I really dont want to do. BUT I recall that NRI have done this an it was all acceptable. is this your recollection @rml599gh?
|
|
@JhanSrbinovsky - maybe let me know where the output is and I can have a look in the morning. Is this a single time? Looking at some of my runs, I am typically seeing +/- 15 degrees in the December monthly mean, so if your plot isn't a time average, then maybe the differences you are seeing are plausible. |
|
daily averages. honestly though its the 1st of the three Tscrn fields in the .pe file.. I'm assuming it goes mean, max, min. Either way its mean-mean, max-max or min-min |
|
On the Payu vs script run question – For the ESM1.5 release with payu we regularly checked whether the output from short PI simulations matched the PI-02 simulation on p73. I just checked my longer simulation of the NRI release, and into year 20 the two match: I hadn't looked this far into the simulation before, but surprisingly it starts diverging, with what looks like noise appearing in May of the 20th year: |
|
That is strange. It doesnt sound like a physical difference, pretending the simulation is physically real. I suppose you would know if the script model had been restarted or perturbed on restart. BUT the results of the tiny perturbation we sometimes applied would've caused even greater divergence than you've shown here. I can vaguely make out coast lines there, suggesting it was a land thing. I dont really want to chase it. Almost pointless as the version of CABLE is different. BUT there was (perhaps still is) a dodgy variable in CABLE canopy. This was 10 years ago but I remember running the same single-site simulation back-to-back 10 times (or more) in order to follow it. Was your simulation running in monthly chunks, annual? It very well might have only been 1 timestep in the 20th year that started the divergence. It just snowballs after that. |
|
🚀 Deploying access-esm1.6 Details and usage instructionsThis
This Prerelease is accessible on Gadi using: module use /g/data/vk83/prerelease/modules
module load access-esm1p6/pr28-3where the binaries shall be on your 🛠️ Using: spack DetailsIt will be deployed using:
If this is not what was expected, commit changes to |
JhanSrbinovsky
had a problem deploying
to
Gadi Prerelease
GitHub Actions
Failure
JhanSrbinovsky
had a problem deploying
to
Gadi Prerelease
GitHub Actions
Failure
|
🚀 Deploying access-esm1.6 Details and usage instructionsThis
This Prerelease is accessible on Gadi using: module use /g/data/vk83/prerelease/modules
module load access-esm1p6/pr28-5where the binaries shall be on your 🛠️ Using: spack DetailsIt will be deployed using:
If this is not what was expected, commit changes to |
JhanSrbinovsky
had a problem deploying
to
Gadi Prerelease
GitHub Actions
Failure
Include my branch including CABLE updates intended for spin up run
🚀 The latest prerelease
access-esm1p6/pr28-5at d9eec7a is here: #28 (comment) 🚀