Modifying a UM restart file for ACCESS-ESM1.5

Hi @ben , quick look at the file seems good. It looks like I’ve finally got test runs going so I’ll hopefully start to have some answers early next week.
Thanks for your ongoing help with this.

Hi @ben , test runs have given me lots of output to play with. Initial look makes me think I probably didn’t provide the best instructions for averaging the carbon pools in the .astart file. I suspect we need to account for tile fraction, which I’m guessing hasn’t been included. Rachel

Hi @ben , I’ve tested weighting by vegetation fraction on the wood carbon pool (field 853). There is an example netcdf file here: /g/data/p66/rml599/LandUse-tests/ and the code I used to generate it: /g/data/p66/rml599/LandUse-tests/avTreePFTs.f90 . Please could you make me some new astart files with weighted averages - or show me how to make them. Thanks.

Sure thing @RachelLaw, I’ll make the new weighted file and post my code to GitHub.


Hi @RachelLaw , here is the code that generates the files as is.

(Best to go directly to GitHub to see what it does)

I’ll work on adding your changes for weighting…

Hi @RachelLaw ,

As per our discussions yesterday, please find a sample of the fields weighted by vegetation fraction (Met Office Codes Registry : um/stash/_m01s00i216) at this location. This is the field you used in your example.


If you could take a look and let me know if this satisfies your requirements that would be great.

If you have any issues with the file, please feel free to email me directly.

Have a great day,


Hi @ben ,
Looks like we are getting close. The only thing I noticed was that areas with no trees now look like they are missing - possibly filled with NaNs e.g. Antarctica. Pools should just be zero in this case.
Thanks, Rachel

Hi Rachel,

I’ve added an explicit step to set the expected 0 values and ensure the fill value is being applied correctly.

New file here: /g/data/p66/bjs581/mean5_VFW_851-860_20240117_064315.astart

Kind Regards,

Hi @ben
Thanks - that looks good. I’ll go ahead and set up a model run with it. Please can you now also make me the ‘second step’ file where the vegfrac fields are replaced with
Regards, Rachel

hi @ben ,
The model run with the latest .astart file looks to have performed as expected. It will be interesting to now test the dominant tree case when you are able to make the appropriate .astart (or step me through how to do it).
Regards, Rachel

Will do, I’ll set up new restarts for you.

@RachelLaw , here is a restart dropping in your dominant case.

Please let me know how you go or if there are any issues.

Cheers, Ben

Hi @ben,
I had a look at the new .astart file with xconv and it isn’t looking as I expected so maybe my latest request wasn’t very clear.
I was expecting the 851-860 fields to be the same as in the mean5_VFW_ file and the only change in fields would be to the vegetation fraction (fields 216 and 835) to the version with one dominant tree type per grid-cell.
When I checked the file, the 851-860 fields looked slightly different so I’m wondering whether these are now weighted with the dominant tree type. Possibly another interesting sensitivity test but one I hadn’t considered. Fields 216 and 835 still look like the vegetation fraction for multiple tree types per grid-cell.
Sorry if I’ve confused things.

Hi Rachel,

Can you please check the following file:


It should have the two different options applied…

Cheers, Ben

Hi @ben,
Thanks for the update. The 851-860 fields are now what I’d expected (i.e. the same as in mean5_VFW_851-860_20240117_064315.astart) but the vegetation fraction field isn’t. It should be the same as the vegetation fraction field in mean5_VF_8551-860_20231213_120638.astart.

Hi Rachel,

Please try:

Cheers, Ben

hi @ben,
Looks good so I’ve started the model run with it.
Thanks, Rachel

Hi Rachel,

Good to hear. Here is the latest code that generates your files.

I’m going to mark this ticket as closed, but please feel free to raise another if you need further assistance.

Cheers, Ben