Hi,
I have a query about the spin-up history of the ACCESS-ESM1.5 pre-industrial release.
How many years has this configuration been spun-up for? (I.e. is there some equilibration built into it already?)
Is there some way we can link the years of spin-up to the Pre-industrial diagnostics that are saved in the CMIP6 archive? I.e. this one:
/g/data/fs38/publications/CMIP6/CMIP/CSIRO/ACCESS-ESM1-5/piControl/r1i1p1f1
The ‘latest’ version of that run, i.e. v20210316, appears to start from year 101. Are these published results using the same initial restart inputs as in the ACCESS-NRI supported release?
There is some useful information on the ESM1.5 spinup in the The Australian Earth System Model: ACCESS-ESM1.5 paper. From the paper, the physical state was spun up for 3000 years total starting from an ESM1 restart, while the BGC was spun up for 1000 years. Model developments and parameter tuning were done iteratively during the spinup, though the model and configuration were frozen for the last 600 years.
There have been two seperate preindustrial runs done with ESM1.5: PI-01 (with outputs at /g/data/p73/archive/CMIP6/ACCESS-ESM1-5/PI-01), and PI-02 (at /g/data/p73/archive/CMIP6/ACCESS-ESM1-5/PI-02). PI-01 was run on Raijin and PI-02 was run on Gadi. I’ll try to confirm which labels match the different runs in the fs38 archive.
The NRI release for ESM1.5 uses the first available restarts from the PI-02 run (calendar year 101).
Due to technical limitations I believe the beginning date for the PI runs couldn’t be set at year 1, and were instead set at year 101. The initial restarts for PI-01 and PI-02 don’t match, and so I’m not sure if they began from different points. I’m just tagging @tiloz, who knows more of the details.
Thanks for this great information @spencerwong. Good to know it’s had a long spin-up. (I guess I need to read the foundation paper more closely in future )
Thanks @spencerwong for the details on the spinup and control run. Just to add that the first 1000 years (PI-01) were published on the ESGF. PI-02 is a continuation of PI-01 for another 1000 years (although the simulation year was reset to 101), but as Spencer pointed out it might have been run on a different machine. I don’t think we published PI-02 on the ESGF. Overall we have 2000 (continuous) years of the picontrol run for ESM1.5
)