xarray 读取远程数据集 (NBM) 时出现 MissingDimensionsError

Question

在读取 NBM 远程数据集 (https://vlab.ncep.noaa.gov/web/mdl/nbm) 时，我得到一个

xarray.core.variable.MissingDimensionsError

。我确信我在

open_dataset

.

中缺少一些参数设置

您可以在此处查看数据的结构：https://thredds-jumbo.unidata.ucar.edu/thredds/dodsC/grib/NCEP/NBM/CONUS/TwoD.html。完整的结构如here所示，使用

ncdump -h https://thredds-jumbo.unidata.ucar.edu/thredds/dodsC/grib/NCEP/NBM/CONUS/TwoD

使用

time1

的变量：

• 降水类型_表面概率_介于_1p0_和_2

import xarray as xr
url = "https://thredds-jumbo.unidata.ucar.edu/thredds/dodsC/grib/NCEP/NBM/CONUS/TwoD"
ds = xr.open_dataset(url)

如果你删除这个变量，它就会进入下一个时间变暗

ds = xr.open_dataset(url, drop_variables="time1")
xarray.core.variable.MissingDimensionsError: 'time2' has more than 1-dimension and the same name as one of its dimensions ('reftime4', 'time2'). xarray disallows such variables because they conflict with the coordinates used to label dimensions.

完整追溯

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Users/ray.bell/miniconda/envs/test_env/lib/python3.8/site-packages/xarray/backends/api.py", line 575, in open_dataset
    ds = maybe_decode_store(store, chunks)
  File "/Users/ray.bell/miniconda/envs/test_env/lib/python3.8/site-packages/xarray/backends/api.py", line 471, in maybe_decode_store
    ds = conventions.decode_cf(
  File "/Users/ray.bell/miniconda/envs/test_env/lib/python3.8/site-packages/xarray/conventions.py", line 600, in decode_cf
    ds = Dataset(vars, attrs=attrs)
  File "/Users/ray.bell/miniconda/envs/test_env/lib/python3.8/site-packages/xarray/core/dataset.py", line 630, in __init__
    variables, coord_names, dims, indexes, _ = merge_data_and_coords(
  File "/Users/ray.bell/miniconda/envs/test_env/lib/python3.8/site-packages/xarray/core/merge.py", line 467, in merge_data_and_coords
    return merge_core(
  File "/Users/ray.bell/miniconda/envs/test_env/lib/python3.8/site-packages/xarray/core/merge.py", line 594, in merge_core
    collected = collect_variables_and_indexes(aligned)
  File "/Users/ray.bell/miniconda/envs/test_env/lib/python3.8/site-packages/xarray/core/merge.py", line 278, in collect_variables_and_indexes
    variable = as_variable(variable, name=name)
  File "/Users/ray.bell/miniconda/envs/test_env/lib/python3.8/site-packages/xarray/core/variable.py", line 154, in as_variable
    raise MissingDimensionsError(
xarray.core.variable.MissingDimensionsError: 'time1' has more than 1-dimension and the same name as one of its dimensions ('reftime', 'time1'). xarray disallows such variables because they conflict with the coordinates used to label dimensions.

您可以在本地测试

wget https://ftp.ncep.noaa.gov/data/nccf/com/blend/prod/blend.20210214/00/core/blend.t00z.core.f001.co.grib2

Answer 1

如果您想从 Xarray 中的 THREDDS Forecast Model Run Collection (FRMC) 虚拟数据集访问这些“TwoD”数据集，您可以先使用 NetCDF 库对它们进行切片，然后将切片变量传递给 Xarray。 如果你用 Dask 包装 NetCDF 变量，你就可以保持懒惰。

以下是为 HRRR 的最后 60 个值提取“最佳时间序列”的示例，但使用 1 小时预测数据（而不是使用 FMRC 最佳时间序列的默认“分析”0 小时预测）：

import netCDF4
import xarray as xr
from dask import array as da
import hvplot.xarray

url = 'https://thredds.unidata.ucar.edu/thredds/dodsC/grib/NCEP/HRRR/CONUS_2p5km/TwoD'
nc = netCDF4.Dataset(url)
arr = da.from_array(nc['Temperature_height_above_ground'])
tau = 1
da = xr.DataArray(arr[-60:,tau,0,:,:], dims=['time','y','x'], name='temp')

这是证明其有效的时间序列图：

Answer 2

现在有一个名为 gribstream.com 的服务，可以免费查询 NBM 数据集。您可以在一个 http 请求中一次检索数月前一小时 (horizon=1) 预报的时间序列和数千个坐标，只需几秒钟。这很酷，因为您不需要下载 GB 或 TB 的数据来提取一些坐标。

历史1小时预测

还预测 11 天后。

预报1-264小时预报

这是Python客户端：https://github.com/GribStream/python-client