xarray - 合并多个具有重叠或缺失时间坐标的 DataArray 对象答案

【问题标题】：xarray - Merge multiple DataArray objects with overlapping or missing time coordinatesxarray - 合并多个具有重叠或缺失时间坐标的 DataArray 对象
【发布时间】：2021-12-21 02:47:46
【问题描述】：

我有多个 netCDF 文件，它们使用具有三个维度 [time、latitude、longitude] 的 xarray 加载到 DataArray 对象中。

latitude 和 longitude 坐标相同，但有些坐标在 time 维度中有重叠或缺失的坐标。

**da1**：

time    (time)    datetime64[ns]    2017-10-03T18:00:00 ... 2017-10-...

array(['2017-10-03T18:00:00.000000000', '2017-10-03T19:00:00.000000000',
       '2017-10-03T20:00:00.000000000', '2017-10-03T21:00:00.000000000',
       '2017-10-03T22:00:00.000000000', '2017-10-03T23:00:00.000000000',
       '2017-10-04T00:00:00.000000000', '2017-10-04T01:00:00.000000000',
       '2017-10-04T02:00:00.000000000', '2017-10-04T03:00:00.000000000',
       '2017-10-04T04:00:00.000000000', '2017-10-04T05:00:00.000000000',
       '2017-10-04T06:00:00.000000000', '2017-10-04T07:00:00.000000000',
       '2017-10-04T08:00:00.000000000', '2017-10-04T09:00:00.000000000',
       '2017-10-04T10:00:00.000000000', '2017-10-04T11:00:00.000000000',
       '2017-10-04T12:00:00.000000000', '2017-10-04T13:00:00.000000000',
       '2017-10-04T14:00:00.000000000', '2017-10-04T15:00:00.000000000',
       '2017-10-04T16:00:00.000000000', '2017-10-04T17:00:00.000000000',
       '2017-10-04T18:00:00.000000000', '2017-10-04T19:00:00.000000000'],
      dtype='datetime64[ns]')

**d2**:
time
(time)
datetime64[ns]
2017-10-04T13:00:00 ... 2017-10-...

array(['2017-10-04T13:00:00.000000000', '2017-10-04T14:00:00.000000000',
       '2017-10-04T15:00:00.000000000', '2017-10-04T16:00:00.000000000',
       '2017-10-04T17:00:00.000000000', '2017-10-04T18:00:00.000000000',
       '2017-10-04T19:00:00.000000000', '2017-10-04T20:00:00.000000000',
       '2017-10-04T21:00:00.000000000', '2017-10-04T22:00:00.000000000',
       '2017-10-04T23:00:00.000000000', '2017-10-05T00:00:00.000000000',
       '2017-10-05T01:00:00.000000000', '2017-10-05T02:00:00.000000000',
       '2017-10-05T03:00:00.000000000', '2017-10-05T04:00:00.000000000',
       '2017-10-05T05:00:00.000000000', '2017-10-05T06:00:00.000000000',
       '2017-10-05T07:00:00.000000000', '2017-10-05T08:00:00.000000000',
       '2017-10-05T09:00:00.000000000', '2017-10-05T10:00:00.000000000',
       '2017-10-05T11:00:00.000000000', '2017-10-05T12:00:00.000000000',
       '2017-10-05T13:00:00.000000000', '2017-10-05T14:00:00.000000000'],
      dtype='datetime64[ns]')

我想通过time 坐标运行“concat”或“merge”或“combine”它们。时间坐标将从“2017-10-03T18:00:00.000000000”到“2017-10-05T14:00:00.000000000”。使用发生重叠坐标的较新值。例如，使用来自 da2 的值，其中 time = '2017-10-04T13:00:00.000000000'。如果输入 DataArray 对象中缺少 time 坐标，请使用 nan。

我试过了：

da_merged = xr.merge([da2, da1], compat="override")

但我最终得到了一大堆nan 值，用于da1 中的时间坐标。我可以告诉我应该使用哪些方法/参数吗

【问题讨论】：

标签： python pandas numpy python-xarray

【解决方案1】：

我希望你解决了它，但对我来说它只是做：

da_merged=xr.Dataset()
file_paths_list =[da1,da2]

for file in file_paths_list:

        da_merged = xr.merge([da_merged,xr.open_mfdataset(file)],compat='override')

【讨论】：