Hello,

I would like to know how many fields are needed to restart OpenIFS / what's the number of prognostic variables? Since the restart fields are stored in binary files it's not easy to inspect. I tried to calculate it from the size of the restart files and the byte size of double but arrived at a non integer number. I guess because some fields have 91 layers and others e.g. soil variables have 4? 

Best regards,

Jan Streffing

3 Comments

  1. Unknown User (nagc)

    Hi Jan,

    The model will need all its prognostic variables in order to do an exact restart, as well as surface fields and some extra fields from the physics (e.g. to maintain accumulated variables). So it's the same prognostic dynamics fields you find in the initial files.  Note that the model writes the dynamical fields as spectral fields (might be why you didn't get the number you were expecting?).

    The model code that writes/reads the restart files is probably the best documentation for what's being written ('ifs/utility/wrresf.F90, reresf.F90').  I've not checked the IFS manuals, might be more description in there but it could be out of date.

    Hope that helps,   Glenn

  2. Unknown User (jstreffi)

    Hello Glenn,

    I kind of knew that working my way though wrresf.F90 would yield the result I am looking for. I thought I'd ask here in case anyone of ECMWF just knows that number of the top of their head. If that's no the case I'll have a closer look.

    Cheers, Jan

  3. Unknown User (nagc)

    To be honest, for question like that I'd always refer to the code and not the IFS manuals to be sure. Though the manuals can sometimes be a useful guide they might be out of date.