Mergetime problem: resulting in a shift in time dimension by 1 month, unphysical values
Hello,
I'm using cdo mergetime to string multiple months before taking a time mean:
cdo timmean -mergetime *.nc out.nc
However, some weirdness appears (hemispheric asymmetries) that I suspect is partly due to the times not merging properly. When I look at a merged output I can see that it moves the time dimension by 1 month, starting one month late (mo 2 of year 1) and ends even beyond the end of the simulation (it ends year 20 mo 12, but after merging it ends year 21 mo 1). That it makes its own new time series is odd, and I'm stumped as to why.
When I look at the output of cdo ymonmean, it gets even weirder – the first and seven months have a very low (physically not possible) value; see the attached plot for global mean insolation. This prompts me to think that timmean is not working as expected either, making hemispheric asymmetries due to the months not being in phase.
I read that using -b 32 can help mergetime, but it doesn't change these results.
Does anybody know what might be causing this? Many thanks in advance,
Aiden
