我目前有一个netCDF文件，该文件的小时读数为1996年7月18日至2006年12月31日，并希望计算数据变量的JJA季节性平均值。我尝试遵循http://xarray.pydata.org/en/latest/examples/monthly-means.html上显示的示例：
ds = xr.open_dataset('BE_Vie.H.nc')
ds
>> <xarray.Dataset>
>> Dimensions: (nt: 2, time: 91632, x: 1, y: 1)
>> Coordinates:
>> * time (time) datetime64[ns] 1996-07-18T01:00:00 ... 2006-12-31
>> latitude (y, x) float32 ...
>> longitude (y, x) float32 ...
>> Dimensions without coordinates: nt, x, y
>> Data variables:
>> time_bounds (time, nt) datetime64[ns] ...
>> gpp_gb (time, y, x) float32 ...
>> resp_p_gb (time, y, x) float32 ...
>> resp_s_gb (time, y, x) float32 ...
>> ftl_gb (time, y, x) float32 ...
>> latent_heat (time, y, x) float32 ...
>> rad_net (time, y, x) float32 ...
>> sw_down (time, y, x) float32 ...
>> precip (time, y, x) float32 ...
>> t1p5m_gb (time, y, x) float32 ...
>> q1p5m_gb (time, y, x) float32 ...
month_length = ds.time.dt.days_in_month
month_length
>> <xarray.DataArray 'days_in_month' (time: 91632)>
>> array([31, 31, 31, ..., 31, 31, 31])
>> Coordinates:
>> * time (time) datetime64[ns] 1996-07-18T01:00:00 ... 2006-12-31
# Calculate the weights by grouping by 'time.season'.
weights = month_length.groupby('time.season') / month_length.groupby('time.season').sum()
# Test that the sum of the weights for each season is 1.0
np.testing.assert_allclose(weights.groupby('time.season').sum().values, np.ones(4))
# Calculate the weighted average
ds_weighted = (ds * weights).groupby('time.season').sum(dim='time')
但是，我遇到此错误：
TypeError Traceback (most recent call last)
<ipython-input-45-51dd727eba52> in <module>
6
7 # Calculate the weighted average
----> 8 ds_weighted = (ds * weights).groupby('time.season').sum(dim='time')
/anaconda3/lib/python3.7/site-packages/xarray/core/dataset.py in func(self, other)
4774 self, other = align(self, other, join=align_type, copy=False)
4775 g = f if not reflexive else lambda x, y: f(y, x)
-> 4776 ds = self._calculate_binary_op(g, other, join=align_type)
4777 return ds
4778
/anaconda3/lib/python3.7/site-packages/xarray/core/dataset.py in _calculate_binary_op(self, f, other, join, inplace)
4845 else:
4846 other_variable = getattr(other, "variable", other)
-> 4847 new_vars = {k: f(self.variables[k], other_variable) for k in self.data_vars}
4848 ds._variables.update(new_vars)
4849 ds._dims = calculate_dimensions(ds._variables)
/anaconda3/lib/python3.7/site-packages/xarray/core/dataset.py in <dictcomp>(.0)
4845 else:
4846 other_variable = getattr(other, "variable", other)
-> 4847 new_vars = {k: f(self.variables[k], other_variable) for k in self.data_vars}
4848 ds._variables.update(new_vars)
4849 ds._dims = calculate_dimensions(ds._variables)
/anaconda3/lib/python3.7/site-packages/xarray/core/variable.py in func(self, other)
1987 new_data = (
1988 f(self_data, other_data)
-> 1989 if not reflexive
1990 else f(other_data, self_data)
1991 )
TypeError: ufunc multiply cannot use operands with types dtype('<M8[ns]') and dtype('float64')
我应该如何转换我的时间变量以使其工作？在这个问题上的任何帮助将不胜感激-谢谢！！
有关时间变量的更多详细信息：
ds.time
>> <xarray.DataArray 'time' (time: 91632)>
>> array(['1996-07-18T01:00:00.000000000', '1996-07-18T02:00:00.000000000',
>> '1996-07-18T03:00:00.000000000', ..., '2006-12-30T22:00:00.000000000',
>> '2006-12-30T23:00:16.000000000', '2006-12-31T00:00:00.000000000'],
>> dtype='datetime64[ns]')
>> Coordinates:
>> * time (time) datetime64[ns] 1996-07-18T01:00:00 ... 2006-12-31
>> Attributes:
>> standard_name: time
>> long_name: Time of data
>> bounds: time_bounds
答案 0 :(得分：1)
问题似乎出在日期时间变量乘以浮点数。调用(ds * weights)时，权重将乘以ds中的每个变量。我猜想ds.time_bounds * weights真的没有道理，而且这个TypeError也同意。
我建议将time_bounds提升为坐标。非索引坐标类似于数据变量，但数学运算不会影响它们。请参阅Coordinates上的xarray文档。
在加权操作之前尝试ds = ds.set_coords('time_bounds')。