[MITgcm-devel] ice/snow pdf

[Martin Losch]

Hi Jean-Michel and others,

I am trying to recover an implementation that allows prescribing a ice thickness distribution instead of the simple 1/seaice_multDim one (effects are described here: http://onlinelibrary.wiley.com/doi/10.1002/2013JC009342/abstract). Unfortunately, results change from 16 to 14-11 digits of aggreement for basically all experiments that use SEAICE_multDim=7. The reason is, that Fortran doesn’t know that 7/7=1, but 0.9999999999978 or so. Do we care? Or do we care enough about prescribing an observed ice thickness distribution that we don’t care about these small difference? If not, I’ll just scrap this effort.

what do you think?

Martin