How can I apply Groupby function and still keep the chunks ?

[babameme]

1. I have an 2-dims DataArray which be chunked along 1-dim
   I apply groupby operation on it, and it breaks out the chunks
   I expect it keeps the chunk, how can i process it correctly ?

2. Code:

da = xr.DataArray(np.arange(120).reshape(24, 5), coords = {"x": range(24), "y":range(5)}, dims = ("x", "y"))
da = da.chunk(chunks = {"x" : 5})

da show:

> da
xarray.DataArray x: 24y: 5
dask.array<chunksize=(5, 5), meta=np.ndarray>
Coordinates:
x
(x)
int64
0 1 2 3 4 5 6 ... 18 19 20 21 22 23
y
(y)
int64
0 1 2 3 4
Attributes: (0)

da.groupby("x").quantile(0.02, dim = "y")

Result:

xarray.DataArrayx: 24
dask.array<chunksize=(1,), meta=np.ndarray>
Coordinates:
quantile
()
float64
0.02
x
(x)
int64
0 1 2 3 4 5 6 ... 18 19 20 21 22 23
Attributes: (0)

3. Version of xr:

Details

INSTALLED VERSIONS

commit: None
python: 3.7.8 | packaged by conda-forge | (default, Jul 31 2020, 02:25:08)
[GCC 7.5.0]
python-bits: 64
OS: Linux
OS-release: 5.8.9-1.el7.elrepo.x86_64
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8
LOCALE: en_US.UTF-8
libhdf5: 1.12.0
libnetcdf: 4.7.4

xarray: 0.17.0
pandas: 1.1.4
numpy: 1.19.1
scipy: 1.5.2
netCDF4: 1.5.6
pydap: None
h5netcdf: None
h5py: 3.2.1
Nio: None
zarr: None
cftime: 1.4.1
nc_time_axis: None
PseudoNetCDF: None
rasterio: None
cfgrib: None
iris: None
bottleneck: None
dask: 2.30.0
distributed: 2.30.1
matplotlib: 3.3.2
cartopy: None
seaborn: 0.11.0
numbagg: None
pint: None
setuptools: 49.6.0.post20210108
pip: 21.0.1
conda: None
pytest: 6.0.2
IPython: 7.22.0
sphinx: 3.2.1

[dcherian]

Hi @babameme This is a side effect of how groupby works. You can call .chunk after .quantile to bring it back to what you want.