Convert daily-to-monthly xarray using customized calendar

Technical
,
1

I have a 3-D Xarray with a 365-day calendar, having dimensions [3650 x 73 x 144] : (time x lat x lon). I want to convert this 10-year daily dataset into a monthly format.
This can be done using groupby(‘time.month’) for a standard calendar.

However, my target calendar has an unusual month length = [32,30,32,31,31,29,30,29,29,30,30,32].

How can I apply this new month length information to convert the daily dataset into monthly, where each month’s length is based on the above array?

2

One way to do this is to manually categorize the axis and then group by that. Something like


month_ids = numpy.concat([[0]*32, [1]*30, [2]*32, ...)

years = [1,2,3,4,...]

year_months = numpy.array([])
for Y in years:
    # Year 2, month 12 has a label of '212'
    year_months.append(Y * 100 + month_ids)

ds.coords['year_month'] = ('time', year_months)

ds.groupby('year_month').mean()

I might have some of the syntax wrong, just going by memory here

3 Likes
3

It works with minor alterations. Thanks @Scott.

1 Like