Thanks. Please could you let me know how to perturb the restart as I haven’t done that before.
I had a similar error message when one of my runs crashed and perturbing the initial conditions worked.
This is what I used:
module load python2/2.7.16
module use ~access/modules
module load pythonlib/umfile_utils
python /g/data/access/projects/access/apps/pythonlib/umfile_utils/perturbIC.py -a 0.05 restart_dump.astart
Hi all - just to follow up on the brief discussion at the nesp project meeting, I have checked the N3.4 variability on the second CN run ( /scratch/p66/rml599/access-esm/archive/test-CN-thin-test-CN-thin-aa92d7b2 ) and it also appears to have much reduced variability during the peak of ENSO.
Meanwhile the CNP run is not much changed compared to ESM1.5.
CN ‘thin test’ run:
CN:
CNP:
Might be interesting to check the variability in SD across the ESM1.5 ensemble.
Here’s the shading showing the spread of the ESM1.5 ensemble, looks like the 2nd CN run is right along the bottom edge then lies outside the spread from Dec-Mar. The first CN run (from the post above) would lie within the spread for DJF but interestingly significantly overestimates the variability in winter:
Here are some maps showing the differences in SD during DJF (using the ESM1.5 ensemble mean)
I haven’t updated the main globally-averaged time series for a while - so here goes …
Key notes:
- towards the start of January we noted that the terrestrial biogeochemistry and global climate was not responding as expected in the idealised experiments (1%/yr simulations) that were started from the December spin up. A parameter bug in the phosphorus cycle was discovered, corrected and brought back into the main spinup.
- in addition @MartinDix discovered and reported to CMIP an issue with the CMIP7 ozone forcing (as introduced in the December spin up).
Current status: two section of spin up (around 225 years in total - grey in the plots with overlap of the red December spin up) have been run with the phosphorus fix but the still-incorrect ozone - and the spin up runs now paused.
In short - there is no real discernible change in behaviour in the climatology of the global physical ocean in the latest section (as expected).
Updated configurations
@tiloz @RachelLaw, the development configurations are now all up to date with the latest control runs. They are available at:
- preindustrial+concentrations: GitHub - ACCESS-NRI/access-esm1.6-configs at dev-preindustrial+concentrations
- preindustrial+concentrations+CN: GitHub - ACCESS-NRI/access-esm1.6-configs at dev-preindustrial+concentrations+CN
- historical+concentrations: GitHub - ACCESS-NRI/access-esm1.6-configs at dev-historical+concentrations
- 4xCO2: GitHub - ACCESS-NRI/access-esm1.6-configs at dev-4xCO2+concentrations
- 1pctCO2: GitHub - ACCESS-NRI/access-esm1.6-configs at dev-1pctCO2
- 1pctCO2-rad: GitHub - ACCESS-NRI/access-esm1.6-configs at dev-1pctCO2-rad
- 1pctCO2-bgc: GitHub - ACCESS-NRI/access-esm1.6-configs at dev-1pctCO2-bgc
- preindustiral+emissions: GitHub - ACCESS-NRI/access-esm1.6-configs at dev-preindustrial+emissions
- flat10: GitHub - ACCESS-NRI/access-esm1.6-configs at dev-flat10
- amip: GitHub - ACCESS-NRI/access-esm1.6-configs at dev-amip
We’ve also increased the jobfs from 1500MB to 15GB, which will hopefully resolve the error in the historical configuration. We’ll work on adding the CN versions of the other configurations now
Thanks @spencerwong - much appreciated. Just confirming - does the historical run still need ozone updated, plus greenhouse gases in the ‘namelists’ file? And is the emissions-driven historical also ready apart from these two inputs?
I did notice in the historical config.yaml, the old Bio_1849_2015_ESM1.anc is listed as well as the cmip7 one. I assume this is redundant and should be removed to avoid confusion?
Historical tests
I’ve run a series of historical tests with different options for secondary harvest/wood thinning. This impacts land carbon and leaf area index. Impacts on climate should be minor so these could be considered as different ensemble members if anyone wants to look at them.
Output is in /scratch/p66/rml599/access-esm/archive.
The directories for the historical tests are:
hist-fixedP-thin-forest-dev-historical+concentrations-5b2d467c
hist-fixedP-thin-half-dev-historical+concentrations-303cb2e4
hist-fixedP-thin-quarter-dev-historical+concentrations-c5f22634
test-hist-fixedP-10-test-hist-fixedP-10-e449d343
test-hist-fixedP-10-test-hist-fixedP-thin-limit-94da3ef2
Note that these still use CMIP6 ozone and CMIP6 greenhouse gases and only run to 2014.
That’s correct, the ozone and greenhouse gasses still need to be updated for the historical configuration. The emissions driven historical isn’t quite ready yet.
Thanks for pointing out the redundant Bio_1849_2015_ESM1.anc file, we’ll remove these from the configurations.
The additional carbon-nitrogen versions of the development configurations are now ready, they are available at:
- 4xCO2+concentrations+CN: GitHub - ACCESS-NRI/access-esm1.6-configs at dev-4xCO2+concentrations+CN
- 1pctCO2+CN: GitHub - ACCESS-NRI/access-esm1.6-configs at dev-1pctCO2+CN
- 1pctCO2-rad+CN: GitHub - ACCESS-NRI/access-esm1.6-configs at dev-1pctCO2-rad+CN
- 1pctCO2-bgc+CN: GitHub - ACCESS-NRI/access-esm1.6-configs at dev-1pctCO2-bgc+CN
- historical+concentrations+CN: GitHub - ACCESS-NRI/access-esm1.6-configs at dev-historical+concentrations+CN
- amip+CN: GitHub - ACCESS-NRI/access-esm1.6-configs at dev-amip+CN
- preindustrial+emissions+CN: GitHub - ACCESS-NRI/access-esm1.6-configs at dev-preindustrial+emissions+CN
- flat10+CN: GitHub - ACCESS-NRI/access-esm1.6-configs at dev-flat10+CN
All configurations have also been updated to use the MOM cf_compliant=.true. flag, which will affect some of the attributes in the output netCDF files from the ocean, and will help with the CMORisation.
Thanks @spencerwong - on the emissions-driven historical, I’ve just put a note into the relevant issue in the configs repository on my (mostly successful?) attempt to fudge an emissions-driven historical for running over the last weekend. One thing I noticed is the ‘ihist’ file - is it being used, because it appears to refer to CMIP6 era ancillary files even when some of them have been updated to CMIP7 ones.
4/3/2026 updated full ocean metrics following the formats above - but with the (hopefully finalised) CNP (grey) and CN (violet) concentration driven runs added at the end. Time has been shifted so that the start of the Feb26-CNP run is at year0. mean/trend number provide values for the last 100years of the 2 control runs.
@RachelLaw, the historical configurations now have the CMIP7 GHG concentrations/emissions and ozone and should be ready to use. They are available at:
- historical+concentrations: GitHub - ACCESS-NRI/access-esm1.6-configs at dev-historical+concentrations · GitHub
- historical+concentrations+CN: GitHub - ACCESS-NRI/access-esm1.6-configs at dev-historical+concentrations+CN · GitHub
- historical+emissions: GitHub - ACCESS-NRI/access-esm1.6-configs at dev-historical+emissions · GitHub
We’ll try to get a historical+emissions+CN configuration ready shortly too. Note that these don’t have the extra Bio_1849_2015_ESM1.anc file removed yet, but it won’t impact the runs.
All configurations have also been updated to point to the std_plus_subdaily STASHC files by default, and the latest executable which give a ~3.5% speed boost.
Historical+concentrations runs are underway with the new (hopefully final) cmip7 configuration. No obvious difference from our previous test (red) for global mean surface air temperature. Black and green are started 20 years apart from the CNP picontrol (from restart100 and restart120).
Output is in
/scratch/p66/rml599/access-esm/archive/hist-mar26-hist-mar26-e88229ea/
/scratch/p66/rml599/access-esm/archive/hist-mar26-02-hist-mar26-02-1c87d12c
I will try and get a 3rd ensemble member going over the weekend along with another emissions-driven historical test.
Hi everyone, just a heads up about an update to the configurations. The output filenames for MOM have just been updated to conform with the data specification, which means that freshly cloned configurations will produce ocean outputs with slightly different filenames. Specifically the frequency sections in the filenames have been changed:
yearly→yrmonthly→mondaily→dayhourly→hrfxis now appended to fixed fields
The date in the MOM filenames has also been reduced from YYYY_MM to YYYY
E.g changes to ocean-2d-surface_temp-1monthly-mean-ym_1850_01.nc changes to ocean-2d-surface_temp-1mon-mean-y_1850.nc
Also note that there is currently work to update the atmosphere filenames to follow the same convention for the frequency information.
check in on the pre-industrial control concentration driven spin-ups below - we are up to 688 years (after resetting the clock) in the CNP-configuration (grey) and 446 years in the CN-configuration (purple). Overarching message - no major changes picked up in the global ocean averages (physical or BGC metrics) though there has clearly been some notable interannual variability going in [which could be interesting to evaluate].
Hello everyone. The configuration updates we discussed this morning to include missing core CMIP ocean variables has now been merged into all dev- configurations. If you are running an experiment, please ensure you update your configuration to include these diagnostics.
I also followed up on the different time variables in the atmosphere that we discussed this morning. I’ve reopened STASH settings for CASA fields · Issue #63 · ACCESS-NRI/access-esm1.6-configs and added a comment there. It is not clear to me that we are currently outputing monthly mean CNP pools even though we discussed doing this, only end of month values (unless I am misreading the STASHC file).