使用where屏蔽xarray dataarray

Question

我正在尝试使用1,0 xarray数据数组来掩盖另一个datarray。两者具有相同的尺寸。我有一个名为odr的xarray数据数组：

odr
Out[283]: 
<xarray.DataArray 'pr' (time: 366, lat: 99, lon: 111)>
array([[[         nan,          nan, ..., 0.000000e+00, 0.000000e+00],
        [         nan,          nan, ..., 0.000000e+00, 0.000000e+00],
        ...,
        [         nan,          nan, ..., 0.000000e+00, 0.000000e+00],
        [         nan,          nan, ..., 0.000000e+00, 0.000000e+00]],

       [[         nan,          nan, ..., 0.000000e+00, 0.000000e+00],
        [         nan,          nan, ..., 0.000000e+00, 0.000000e+00],
        ...,
        [         nan,          nan, ..., 0.000000e+00, 0.000000e+00],
        [         nan,          nan, ..., 0.000000e+00, 0.000000e+00]],

       ...,

       [[         nan,          nan, ..., 2.467176e-10, 1.868555e-10],
        [         nan,          nan, ..., 0.000000e+00, 0.000000e+00],
        ...,
        [         nan,          nan, ..., 0.000000e+00, 0.000000e+00],
        [         nan,          nan, ..., 0.000000e+00, 0.000000e+00]],

       [[         nan,          nan, ..., 0.000000e+00, 0.000000e+00],
        [         nan,          nan, ..., 0.000000e+00, 0.000000e+00],
        ...,
        [         nan,          nan, ..., 0.000000e+00, 0.000000e+00],
        [         nan,          nan, ..., 0.000000e+00, 0.000000e+00]]])
Coordinates:
  * time     (time) datetime64[ns] 2016-01-01T12:00:00 ... 2016-12-31T12:00:00
  * lon      (lon) float64 147.5 147.6 147.6 147.7 ... 152.9 152.9 153.0 153.0
  * lat      (lat) float64 -36.4 -36.35 -36.3 -36.25 ... -31.6 -31.55 -31.5
Attributes:
    regrid_method:  conservative 

我想使用另一个称为mask3d的数据数组来屏蔽：

mask3d

Out[289]: 
<xarray.DataArray 'mask' (time: 366, lat: 99, lon: 111)>
array([[[0., 0., ..., 0., 0.],
        [0., 0., ..., 0., 0.],
        ...,
        [0., 0., ..., 0., 0.],
        [0., 0., ..., 0., 0.]],

       [[0., 0., ..., 0., 0.],
        [0., 0., ..., 0., 0.],
        ...,
        [0., 0., ..., 0., 0.],
        [0., 0., ..., 0., 0.]],

       ...,

       [[0., 0., ..., 0., 0.],
        [0., 0., ..., 0., 0.],
        ...,
        [0., 0., ..., 0., 0.],
        [0., 0., ..., 0., 0.]],

       [[0., 0., ..., 0., 0.],
        [0., 0., ..., 0., 0.],
        ...,
        [0., 0., ..., 0., 0.],
        [0., 0., ..., 0., 0.]]], dtype=float32)
Coordinates:
  * lat      (lat) float64 -36.4 -36.35 -36.3 -36.25 ... -31.6 -31.55 -31.5
  * lon      (lon) float64 147.5 147.6 147.6 147.7 ... 152.8 152.9 152.9 153.0
  * time     (time) datetime64[ns] 2016-01-01T12:00:00 ... 2016-12-31T12:00:00
Attributes:
    units:          none
    stagger:        M
    standard_name:  MASK
    long_name:      Statistics mask for NARCliM2 evaluation
    description:    1=USE, 0=DO NOT USE

当我尝试odr.where（mask3d> 0.5）时得到

odr.where(mask3d > 0.5)
Out[290]: 
<xarray.DataArray 'pr' (time: 366, lat: 2, lon: 2)>
array([[[nan, nan],
        [nan, nan]],

       [[nan, nan],
        [nan, nan]],

       ...,

       [[nan, nan],
        [nan, nan]],

       [[nan, nan],
        [nan, nan]]])
Coordinates:
  * lat      (lat) float64 -36.4 -36.35
  * lon      (lon) float64 147.5 147.6
  * time     (time) datetime64[ns] 2016-01-01T12:00:00 ... 2016-12-31T12:00:00
Attributes:
    regrid_method:  conservative

我希望数据集的尺寸与odr相同。谁能告诉我发生了什么事？