nancy collins

if the dart file had a global attribute to indicate it was "dart format" that would be good -- make it more self-describing. then maybe a namelist item wouldn't be...

there are 2 issues here. one is scientific and one is implementation related. scientific: when we discussed this before i don't remember any decision on what to do with fields...

p.s. this affects any models that allow missing values in the state. these are rare, fortunately.

i did get them mixed up. for this one: this does affect any models that have NO_COPY_BACK as an option.

> I guess I remember the IO code dealing with inflation files differently. there are slight differences higher in the calling stack, e.g. inflation copy values aren't clamped and the...

something we've talked about for a long time is having a different mechanism for fields which are only needed to compute a forward operator but aren't going to be changed...

there is the additional issue that just setting them NO_COPY_BACK doesn't automatically prevent those fields from getting updated by the assimilation. the user would have to take the additional step...

see my answers/opinions to your (very good) questions below: > At the moment the non-state is lumped in with the state in DART. > > * Are we doing more...

it think it was added when we made the fill_inflation_restart utility. we needed the i/o routines to write everything that was in the state regardless of what the copy back...

i wonder if some of the complexity is related to: 1. if you put 'input' as an output stage, it computes and outputs the ensemble mean and sd, since the...