Main menu


Reproducing: f40.1979_amip.track1.2deg.001

4 posts / 0 new
Last post
Reproducing: f40.1979_amip.track1.2deg.001

From user:

I send you this email in order for your help for a successful reproduction of the original simulation results of the simulation case named f40.1979_amip.track1.2deg.001 that is described on the website 

For our project, we want to reproduce the original simulation results successfully and have tried four attempts of reproduction while all of them are unsuccessful because there are obvious differences or even systematic differences between the original results and the new results from each attempt of reproduction (please refer to the attached doc for more details for the attempts of reproduction). It is very difficult for us to detect the correct reasons why there are systematic differences. So we really hope you'd like to help us for a successful reproduction. Could you please give us some details of the original simulation setting, such as the detail code version, the compiler version, the compiler flag, the input data files and the namelist file? Thanks a lot in any case.



Which tag are you using?

This run was using a beta tag: ccsm4_0_beta46 and the compset F_AMIP_1DEG_CMIP5

create_newcase -mach bluefire -compset F_AMIP_1DEG_CMIP5 -res 1.9x2.5_1.9x2.5 -case /fs/cgd/csm/runs/ccsm4_0/f40.1979_amip.track1.2deg.001

If you don't have acees to beta tags, I think your best bet would be to use the ccsm4 release




From user:

As shown in the attached doc, we have already used the CCSM4.0 code (please refer to the first and second attempt in the attached doc). As the results from the first and second attempt are similar and these two attempts are only different in compiler versions, we guess that the obvious systematic differences between the original results (from ESG of NCAR) and the new results from our attempts are mainly introduced by the differences between the original code and the code versions we used. 

As we failed to reproduce the original results when using the CCSM4.0 code, could you please help us to access the code version ccsm4_0_beta46, for example, could you please send this version of code to me? Thanks a lot. Moreover, according to this command "create_newcase -mach bluefire -compset F_AMIP_1DEG_CMIP5 -res 1.9x2.5_1.9x2.5 -case /fs/cgd/csm/runs/ccsm4_0/f40.1979_amip.track1.2deg.001" you mentioned, I'd like to further ask: are the original results from bluefire that is an IBM PowerPC machine? Thanks a lot.

BTW, we find slight differences of SST data between the original simulation and our four attempts, where the SST data of our four attempts is obtained automatically when configuring a simulation. 




I don't have the authority to give you access to ccsm4_0_beta46. This is development and it requires development access.
You would need to apply to:
but there are some requirements to access development code.

However, I doubt there is much difference between the beta06 version and the release code. 

So my first question: where are you running this ? If it is not a suppotred machine, did you port the code on your machine ?

Another possiblity is that there are differences between your namelists and f40.1979_amip.track1.2deg.001 namelists.  
I am putting the namelists into the ftp server:
in the directory
cd pub/hannay/f40.1979_amip.track1.2deg.001/
so you can compare with your namelists

In teh same directory, I am also providing the SST that was used for: f40.1979_amip.track1.2deg.001


To answer your question about bluefire:
f40.1979_amip.track1.2deg.001 was run on bluefire. Bluefire was an IBM ( and it was retired several years ago. 



Log in or register to post comments

Who's new

  • ezhu@...
  • 1260680980@...
  • 675881908@...
  • liruichao@...
  • krc139@...