namelist parameter order

[climatemodelling@gmx_com]

Do the parameters in the namelists have to be in a certain order for the model to read them correctly?The &cam_inparm namelist did not include the pertlim parameter, so I just added it to the end of the namelist:pertlim = 0.1I hope this is correct? Thanks!

 

[hannay]

The order doesn't matter.Make sure you edit the namelist at the right place using for instance in user_cam_nl
You cannot directly edit the file "atm_in" See how to modify namelist at: http://www.cgd.ucar.edu/cms/hannay/publications/CESMTutorial2011.pdf

 

[climatemodelling@gmx_com]

Thanks hannay! From the tutorial, I thought it was possible to just add the statement in the &cam_inparm section of the cam.buildnml.csh script?

 

[hannay]

Yes you can make it there too. 

 

[climatemodelling@gmx_com]

I have a follow up question: I started several ensebmle members with the namelist parameter pertlim = 0.1 included.However, they produce the same output. Does the random number generator always start at the same point?In this case, I guess using different pertlim parameters for each ensemble would be the way to go:pertlim = 0.1pertlim = 0.11pertlim = 0.12 and so on?

 

[eaton]

The seed for the random number generator is based on the global column index, so it's  the same from one run to the next.  The usual technique to generate an ensemble is to slightly vary the pertlim variable.  Note however that pertlim = 0.1 will result in some very large perturbations (10%) to the initial temperature field values.  We more commonly use pertlim to apply unit roundoff size perturbations, and so set it to values around 1.e-14.