They have no experience evaluating the forecasts from physical models and even less in running the models themselves as they attempted to do.
they did not run the models themselves. what makes you suggest that they did?
In an off discipline journal the editor may not even know who is an appropriate reviewer.
that is certainly possible. are you suggesting that it was the case for this paper, this journal, this editor?
no offense intended, but have you looked at this paper or issue of the journal?
You do realize that for models of this type every run produces a different realization?
yes, and no. running an identical run on the same machine usually yields an identical result (digital computers merely map one big integer into one big integer). cosmic rays and hardware faults can introduce divergent results, but in my experience this is more of a problem if you run climate models on PCs than it is if you run them on supercomputers.
as i recall reproducibility was an early requirement for IPCC runs.
of course runs with the same code but different inputs (or on different hardware, or different compilers) yield different results. is that is what you intended? if so, how is it relevant here?
your opening claim that british statisticians obtained and ran DePreSys casts some doubt on your familiarity either with the computational power of a British "management sciences" department or with the habits of the UK Met Office. i am happy to stand corrected, given evidence.
You can't evaluate output you don't understand.
evidence please? i can contrast and evaluate three different forecast systems for sunspot number without understanding the dynamics of the sun. no?
but what is your point in this case? as it happens these authors do "understand" global mean temperature as well as the average meteorologist understands that rather unphysical entity....
wasn't it the Met Office which selected the target, not the statisticians?
There is SOME work ongoing to try test them against over 1-4 year range, but this is bleeding edge science that is not even remotely "well established"
i suggest you are a bit out of date on this. decadal forecasts are included in (and predate) CMIP5. you could look at outputs from the ENSEMBLES project, which ended a few years ago
http://www.ensembles-eu.org/