Setting up ACCESS

Hi ACCESS, I am new to the ACCESS model. I plan to use high-resolution ACCESS-CM2 for AMIP-style experiments. Could you please help me set up an account and other related permission on accessdev so I can try to run the model?

Hi Sramana,

Welcome to the ACCESS-Hive forum!

This is a relatively up-to-date “Getting Started with ACCESS-CM2” guide

As a new user your feedback on how easy that is to follow, and any amendments that might be needed would be very valuable. Please feel free to reply to this topic so it can be incorporated in future documentation.

If this answers your question please feel to mark this as the solution (see the best practice guide for details).

(I hope you don’t mind, but I have moved this topic to the Climate category, as I think that is where it is most likely to be seen by others who may have the same query)

This is some more up to date documentation:


Thanks, the documentation highlighted here was a good starting point. However, when I am trying to run the ACCESS-CM2 AMIP suite (u-bn157), there is an error in the fcm_make2_um file. Could I please have some help with this?


You would need to provide some detail about the nature of the error.

You can paste the error into a reply here, and if you use the pre-formatted text button above it will put it in a special box and format it so it is easier to read.

Hi Aidan,

Thank you for the reply. The error was resolved by checking out a copy of the suite and rerunning it.

I was able to run the model for a year and was trying to analyse the model output. According to the guide “Understanding CM2 model outputs” the outputs would be located in the /scratch/$PROJECT/$USER/archive/ directory. However, no such directory gets created even after the AMIP run succeeds. Upon investigating the /work/atmos-main/ directory, there are a number of files created with .arch extension.

ATMOSCNTL              cu803a.pd1975apr.arch  history_archive
cable.nml              cu803a.pd1975feb.arch  IDEALISE
cu803a.p71975apr.arch  cu803a.pd1975jan.arch  IOSCNTL
cu803a.p71975feb.arch  cu803a.pd1975mar.arch  pe_output
cu803a.p71975jan.arch  cu803a.pd1975may.arch  RECONA
cu803a.p71975mar.arch  cu803a.pm1975apr.arch  rose-app-run.conf
cu803a.p71975may.arch  cu803a.pm1975feb.arch  SHARED
cu803a.p81975apr.arch  cu803a.pm1975jan.arch  SIZES
cu803a.p81975feb.arch  cu803a.pm1975jun.arch  STASHC
cu803a.p81975jan.arch  cu803a.pm1975mar.arch  STASHmaster_A
cu803a.p81975mar.arch  cu803a.pm1975may.arch  windmax.dat
cu803a.p81975may.arch  errflag

So I was wondering if you could help me with locating the .nc output files that the model is supposed to output or figure out why the archive folder in the scratch directory doesn’t exist?

Thank you!


Hi hi Sramana,

Somehow it came back to me!! Would you please show me where is your run and I will take a look.

"/scratch/$PROJECT/$USER/archive/ " this one is for coupled model, “CM2”.



1 Like

Hi Arnold,

The output of the AMIP runs is being saved in the work directory under “cycl-run/$suite-name/work/atmos-main” on my home directory in gadi. I can copy it somewhere for you to have a look.


1 Like

Hi Sramana,

Just note that the ~/cylc-run/<suite-name> directory is a simbolic link to the /scratch/$PROJECT/$USER/cylc-run/<suite-name> directory that Arnold was referring to.

In case Arnold does not have access to your $PROJECT, you might wanna copy your suite folder in your scratch public folder → /scratch/public/$USER/<suite-name>



Thanks Davide!

I’ve now put the suite folder at /scratch/public/sn2918.

I hope this helps.


I can see some output in the u-cu803/share/data/History_Data/ folder.
You have the cu803a.p7<YEAR><MON> files which contain the 6-hourly output and cu803a.p8<YEAR><MON> which contain the 3-hourly output.


Hi Sramana,

A few suggestions and wrap up:
% The AMIP outputs pp files and netcdf files should be under something like:

% the setting in rose/cylc should modify
% 1st turn on netcdf convert
% 2nd reduce unwantted fields say daily and subdaily (only convert monthly data), say
% From
% To

After that will encourage to run Archive and CMORise (we can talk about this later).