+1 to this issue. I'm struggling big time with an 1800-year climate model dataset that I need to resample in order to make different annual means (June-May).

@lvankampenhout I agree that it would be nice if xarray had better support for PeriodIndexes.

Do you happen to be using a PeriodIndex because of pandas Timestamp-limitations? Despite the fact that generalized resample has not been implemented yet, I recommend you try using the new CFTimeIndex. As it turns out, for some one-off cases (like this one) resample is not too difficult to mimic using groupby. See the following example for your case. I'm assuming you're looking for resampling with the 'AS-JUN' anchored offset? ```python from itertools import product from cftime import DatetimeProlepticGregorian as datetime import numpy as np import xarray as xr

xr.set_options(enable_cftimeindex=True)

Set up some example data indexed by cftime.DatetimeProlepticGregorian objects

dates = [datetime(year, month, 1) for year, month in product(range(2, 5), range(1, 13))] da = xr.DataArray(np.arange(len(dates)), coords=[dates], dims=['time'])

Mimic resampling with the AS-JUN anchored offset

years = da.time.dt.year - (da.time.dt.month < 6) da['AS-JUN'] = xr.DataArray([datetime(year, 6, 1) for year in years], coords=da.time.coords) resampled = da.groupby('AS-JUN').mean('time').rename({'AS-JUN': 'time'}) This gives the following for `resampled`: <xarray.DataArray (time: 4)> array([ 2. , 10.5, 22.5, 32. ]) Coordinates: * time (time) object 0001-06-01 00:00:00 0002-06-01 00:00:00 ... This is analogous to using `resample(time='AS-JUN')` with a DataArray indexed by a DatetimeIndex:python import pandas as pd dates = pd.date_range('2002-01-01', freq='M', periods=36) da = xr.DataArray(np.arange(len(dates)), coords=[dates], dims='time') resampled = da.resample(time='AS-JUN').mean('time') which gives: <xarray.DataArray (time: 4)> array([ 2. , 10.5, 22.5, 32. ]) Coordinates: * time (time) datetime64[ns] 2001-06-01 2002-06-01 2003-06-01 2004-06-01 ```

Maybe related? Selection with slices also doesn't work:

python da = xr.DataArray(pd.Series(1, pd.period_range('1990-1', '2000-12', freq='M'))) da.sel(dim_0='1991-07') # works fine da.sel(dim_0='1992-02') # works fine da.sel(dim_0=slice('1991-07', '1992-02'))

Throws:

```

AssertionError Traceback (most recent call last) <ipython-input-35-2e962ee8a791> in <module>() 2 da.sel(dim_0='1991-07') # works fine 3 da.sel(dim_0='1992-02') # works fine ----> 4 da.sel(dim_0=slice('1991-07', '1992-02'))

/home/mowglie/.pyvirtualenvs/py3/lib/python3.5/site-packages/xarray/core/dataarray.py in sel(self, method, tolerance, drop, indexers) 668 """ 669 pos_indexers, new_indexes = indexing.remap_label_indexers( --> 670 self, indexers, method=method, tolerance=tolerance 671 ) 672 result = self.isel(drop=drop, pos_indexers)

/home/mowglie/.pyvirtualenvs/py3/lib/python3.5/site-packages/xarray/core/indexing.py in remap_label_indexers(data_obj, indexers, method, tolerance) 286 else: 287 idxr, new_idx = convert_label_indexer(index, label, --> 288 dim, method, tolerance) 289 pos_indexers[dim] = idxr 290 if new_idx is not None:

/home/mowglie/.pyvirtualenvs/py3/lib/python3.5/site-packages/xarray/core/indexing.py in convert_label_indexer(index, label, index_name, method, tolerance) 183 indexer = index.slice_indexer(_try_get_item(label.start), 184 _try_get_item(label.stop), --> 185 _try_get_item(label.step)) 186 if not isinstance(indexer, slice): 187 # unlike pandas, in xarray we never want to silently convert a slice

/home/mowglie/.pyvirtualenvs/py3/lib/python3.5/site-packages/pandas/indexes/base.py in slice_indexer(self, start, end, step, kind) 2995 """ 2996 start_slice, end_slice = self.slice_locs(start, end, step=step, -> 2997 kind=kind) 2998 2999 # return a slice

/home/mowglie/.pyvirtualenvs/py3/lib/python3.5/site-packages/pandas/indexes/base.py in slice_locs(self, start, end, step, kind) 3174 start_slice = None 3175 if start is not None: -> 3176 start_slice = self.get_slice_bound(start, 'left', kind) 3177 if start_slice is None: 3178 start_slice = 0

/home/mowglie/.pyvirtualenvs/py3/lib/python3.5/site-packages/pandas/indexes/base.py in get_slice_bound(self, label, side, kind) 3113 # For datetime indices label may be a string that has to be converted 3114 # to datetime boundary according to its resolution. -> 3115 label = self._maybe_cast_slice_bound(label, side, kind) 3116 3117 # we need to look up the label

/home/mowglie/.pyvirtualenvs/py3/lib/python3.5/site-packages/pandas/tseries/period.py in _maybe_cast_slice_bound(self, label, side, kind) 838 839 """ --> 840 assert kind in ['ix', 'loc', 'getitem'] 841 842 if isinstance(label, datetime):

AssertionError: ```

@fmaussion just to clarify, #1252 is meant as an analogue to pandas' DatetimeIndex for non-standard calendars, and does not address resample (it would be nice to have at some point though). It is not intended to be used in place of (or provide similar functionality to) a PeriodIndex.

@MaximilianR perhaps it's also worth noting (as I understand it) xarray does not yet support upsampling with filling (see #563, and the docs). That being said, independent of that, there's definitely something odd going on, since attempting to downsample via a mean produces the same error: ``` In [21]: da = xr.DataArray(pd.Series(1, pd.period_range('2000-1', '2000-12', freq='W')).rename_axis('date'))

In [22]: da.resample('M', 'date', how='mean')

TypeError Traceback (most recent call last) <ipython-input-22-9c464b8e736c> in <module>() ----> 1 da.resample('M', 'date', how='mean')

/Users/spencerclark/xarray-dev/xarray/xarray/core/common.pyc in resample(self, freq, dim, how, skipna, closed, label, base, keep_attrs) 577 time_grouper = pd.TimeGrouper(freq=freq, how=how, closed=closed, 578 label=label, base=base) --> 579 gb = self.groupby_cls(self, group, grouper=time_grouper) 580 if isinstance(how, basestring): 581 f = getattr(gb, how)

/Users/spencerclark/xarray-dev/xarray/xarray/core/groupby.pyc in init(self, obj, group, squeeze, grouper, bins, cut_kwargs) 242 raise ValueError('index must be monotonic for resampling') 243 s = pd.Series(np.arange(index.size), index) --> 244 first_items = s.groupby(grouper).first() 245 if first_items.isnull().any(): 246 full_index = first_items.index

//anaconda/envs/xarray-dev/lib/python2.7/site-packages/pandas/core/generic.pyc in groupby(self, by, axis, level, as_index, sort, group_keys, squeeze, kwargs) 3989 return groupby(self, by=by, axis=axis, level=level, as_index=as_index, 3990 sort=sort, group_keys=group_keys, squeeze=squeeze, -> 3991 kwargs) 3992 3993 def asfreq(self, freq, method=None, how=None, normalize=False):

//anaconda/envs/xarray-dev/lib/python2.7/site-packages/pandas/core/groupby.pyc in groupby(obj, by, kwds) 1509 raise TypeError('invalid type: %s' % type(obj)) 1510 -> 1511 return klass(obj, by, kwds) 1512 1513

//anaconda/envs/xarray-dev/lib/python2.7/site-packages/pandas/core/groupby.pyc in init(self, obj, keys, axis, level, grouper, exclusions, selection, as_index, sort, group_keys, squeeze, **kwargs) 368 level=level, 369 sort=sort, --> 370 mutated=self.mutated) 371 372 self.obj = obj

//anaconda/envs/xarray-dev/lib/python2.7/site-packages/pandas/core/groupby.pyc in _get_grouper(obj, key, axis, level, sort, mutated) 2390 # a passed-in Grouper, directly convert 2391 if isinstance(key, Grouper): -> 2392 binner, grouper, obj = key._get_grouper(obj) 2393 if key.key is None: 2394 return grouper, [], obj

//anaconda/envs/xarray-dev/lib/python2.7/site-packages/pandas/tseries/resample.pyc in _get_grouper(self, obj) 1059 def _get_grouper(self, obj): 1060 # create the resampler and return our binner -> 1061 r = self._get_resampler(obj) 1062 r._set_binner() 1063 return r.binner, r.grouper, r.obj

//anaconda/envs/xarray-dev/lib/python2.7/site-packages/pandas/tseries/resample.pyc in _get_resampler(self, obj, kind) 1055 raise TypeError("Only valid with DatetimeIndex, " 1056 "TimedeltaIndex or PeriodIndex, " -> 1057 "but got an instance of %r" % type(ax).name) 1058 1059 def _get_grouper(self, obj):

TypeError: Only valid with DatetimeIndex, TimedeltaIndex or PeriodIndex, but got an instance of 'Index' ```

I thought this was one of the motivations behind NetCDFTimeIndex : #1252

Why?

I see no reason why this shouldn't work, so my guess is that this could be fixed pretty easily. We certainly don't have any tests for this right now, though.

My first step would be to drop into a debugger and figure out why the PeriodIndex is getting converted into as base Index when put into a Series in xarray's GroupBy.__init__.

I thought this was one of the motivations behind NetCDFTimeIndex : https://github.com/pydata/xarray/pull/1252