帶有Python 3的NetCDF時間序列切片

Question

我試圖從NetCDF文件中繪制一個星期的時間序列數據，並遇到一些問題。

我正在使用以下軟件包：

import netCDF4
from matplotlib import pyplot as plt
import numpy as np
import xarray as xr
import dask

首先，我導入兩個.nc文件：

ds1 = xr.open_dataset('ERA5_forecast_100V_247_2008.nc')
ds2 = xr.open_dataset('ERA5_analysis_100V_247_2008.nc')

然后我使用xarray選擇時間和網格位置：

dsloc1 = ds1.sel(time='2008-02-10',longitude=2.2,latitude=48.7,method='nearest')
dsloc2 = ds2.sel(time='2008-02-10',longitude=2.2,latitude=48.7,method='nearest')

然后繪制兩個時間序列：

dsloc1['v100'].plot.line('b-',figsize=(15,10))
dsloc2['v100'].plot.line('y-')

哪個產生我的期望：

但是，當我嘗試選擇日期范圍時，出現一些錯誤...

dsloc1 = ds1.sel(time=slice('2008-03-01','2008-03-07'),longitude=2.2,latitude=48.7,method='nearest')
dsloc2 = ds2.sel(time=slice('2008-03-01','2008-03-07'),longitude=2.2,latitude=48.7,method='nearest')

我敢肯定這可能是語法問題，但是我花了比我想花更長的時間來解決。

[編輯]這是回溯：

---------------------------------------------------------------------------
NotImplementedError                       Traceback (most recent call last)
<ipython-input-308-c3385fa732ab> in <module>()
      1 # select time and grid location (Feb 10th, 2008, near Paris)
----> 2 dsloc1 = ds1.sel(time=slice('2008-03-01','2008-03-07'),longitude=2.2,latitude=48.7,method='nearest')
      3 dsloc2 = ds2.sel(time=slice('2008-03-01','2008-03-07'),longitude=2.2,latitude=48.7,method='nearest')

/usr/local/lib/python3.6/site-packages/xarray/core/dataset.py in sel(self, indexers, method, tolerance, drop, **indexers_kwargs)
   1507         indexers = either_dict_or_kwargs(indexers, indexers_kwargs, 'sel')
   1508         pos_indexers, new_indexes = remap_label_indexers(
-> 1509             self, indexers=indexers, method=method, tolerance=tolerance)
   1510         result = self.isel(indexers=pos_indexers, drop=drop)
   1511         return result._replace_indexes(new_indexes)

/usr/local/lib/python3.6/site-packages/xarray/core/coordinates.py in remap_label_indexers(obj, indexers, method, tolerance, **indexers_kwargs)
    353 
    354     pos_indexers, new_indexes = indexing.remap_label_indexers(
--> 355         obj, v_indexers, method=method, tolerance=tolerance
    356     )
    357     # attach indexer's coordinate to pos_indexers

/usr/local/lib/python3.6/site-packages/xarray/core/indexing.py in remap_label_indexers(data_obj, indexers, method, tolerance)
    248         else:
    249             idxr, new_idx = convert_label_indexer(index, label,
--> 250                                                   dim, method, tolerance)
    251             pos_indexers[dim] = idxr
    252             if new_idx is not None:

/usr/local/lib/python3.6/site-packages/xarray/core/indexing.py in convert_label_indexer(index, label, index_name, method, tolerance)
    132         if method is not None or tolerance is not None:
    133             raise NotImplementedError(
--> 134                 'cannot use ``method`` argument if any indexers are '
    135                 'slice objects')
    136         indexer = index.slice_indexer(_sanitize_slice_element(label.start),

NotImplementedError: cannot use ``method`` argument if any indexers are slice objects

Answer 1

似乎不支持使用時間片結合method='nearest'的sel ：

如果任何索引器是切片對象，則無法使用method參數

這在某種程度上是有道理的，因為為切片選擇nearest的位置似乎有些奇怪。

您可以通過分兩步進行sel來解決此問題，即首先選擇時間片，然后從該時間片中選擇一個位置（或相反）。 我不確定這是否是最好的解決方案，但至少可以奏效。

具有一些ERA5數據的快速示例：

import xarray as xr

ds1 = xr.open_dataset('20160502_cabauw_model_fc.nc')

# Works:
dsloc1 = ds1.sel(time='2016-05-02 10:00', longitude=4.9, latitude=51.2, method='nearest')

# Doesn't work:
#dsloc2 = ds1.sel(time=slice('2016-05-02 10:00', '2016-05-02 12:00'), longitude=4.9, latitude=51.2, method='nearest')

# Works:
tmp    = ds1.sel(time=slice('2016-05-02 10:00', '2016-05-02 12:00'))
dsloc2 = tmp.sel(longitude=4.9, latitude=51.2, method='nearest')

結果如下：

In [23]: dsloc2
Out[23]: 
<xarray.Dataset>
Dimensions:    (level: 137, time: 3)
Coordinates:
    longitude  float32 4.8
    latitude   float32 51.3
  * level      (level) int32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 ...
  * time       (time) datetime64[ns] 2016-05-02T10:00:00 2016-05-02T11:00:00 ...
Data variables:
    z          (time, level) float32 ...