How to convert “time” in model HadGEM2-ES in real data from 360 days calendar to Gregorian calendar using serial time (I read and select data by MatLab)
The short answer: there is no reason why you should do that. Comparing the projected data from HadGEM2 for, say, 30 February 2083 to 1 March 2083 of some other model makes no sense (it is not a forecast, after all). If you have daily HADGEM2 data then you should process that as needed on a daily basis (e.g. compare experiments within HadGEM2-ES) and then aggregate to a lower resolution (like monthly data) before you compute anomalies and build a multi-model ensemble. As soon as you aggregate to a resolution lower than a day, the different calendars no longer matter.