Archived GOES datasets not loading. AttributeError: 'area' object has no attribute 'crs'

[plecraw]

Hello all,
I am trying to read in high resolution archival nc files all the way back to GOES 8 in 1995. Unfortunately, satpy errors at scn.load while reading native resolution GOES nc files downloaded from NOAA CLASS. (Goes 8, 9, and 12 tested).

Specifically, this fails on:

• native resolution GOES-12 *BAND_01 nc files
• native resolution GOES-08 *BAND_01 nc files

While it works with:

• 16km resolution GOES-09 *BAND_01 nc files
• 16km resolution GOES-12 *BAND_01 nc files
• full resolution GOES-16 *BAND_01_02_03 nc files

My code:

import satpy
from satpy import Scene
from glob import glob
import xarray

filenames = glob('goes12.2005.240.144514.BAND_01*')
scn = Scene(filenames=filenames)

scn.load(['00_7']) #Error occurs during scn.load

new_scn = scn.resample('worldeqc30km',resampler='nearest',fill_value=0)
new_scn.save_dataset('00_7',filename='Goes12testwork'+'.png')

I expect satpy to load the scene and allow me to continue with the resampling procedure.

Here is the resulting error code with debug_on()

[DEBUG: 2022-05-23 15:46:35 : satpy.readers.yaml_reader] Reading ('C:\\Python39\\Lib\\site-packages\\satpy\\etc\\readers\\goes-imager_nc.yaml',)
C:\Python39\lib\site-packages\satpy\readers\seviri_base.py:453: DeprecationWarning: `np.bool` is a deprecated alias for the builtin `bool`. To silence this warning, use `bool` by itself. Doing this will not 
modify any behavior and is safe. If you specifically wanted the numpy scalar type, use `np.bool_` here.
Deprecated in NumPy 1.20; for more details and guidance: https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations
  ('GsicsCalMode', np.bool),
C:\Python39\lib\site-packages\satpy\readers\seviri_base.py:454: DeprecationWarning: `np.bool` is a deprecated alias for the builtin `bool`. To silence this warning, use `bool` by itself. Doing this will not 
modify any behavior and is safe. If you specifically wanted the numpy scalar type, use `np.bool_` here.
Deprecated in NumPy 1.20; for more details and guidance: https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations
  ('GsicsCalValidity', np.bool),
[DEBUG: 2022-05-23 15:46:35 : satpy.readers.yaml_reader] Assigning to goes-imager_nc: ['goes12.2005.240.144514.BAND_01.nc']
[DEBUG: 2022-05-23 15:46:35 : satpy.composites.config_loader] Looking for composites config file goes_imager.yaml
[DEBUG: 2022-05-23 15:46:35 : satpy.composites.config_loader] Looking for composites config file visir.yaml
[DEBUG: 2022-05-23 15:46:35 : satpy.readers.goes_imager_nc] Reading dataset longitude_00_7
[DEBUG: 2022-05-23 15:46:35 : satpy.readers.goes_imager_nc] Computing earth mask
[DEBUG: 2022-05-23 15:46:35 : satpy.readers.goes_imager_nc] Computing nadir pixel
[DEBUG: 2022-05-23 15:46:39 : satpy.readers.goes_imager_nc] Computing yaw flip flag
[DEBUG: 2022-05-23 15:46:40 : satpy.readers.goes_imager_nc] Computing area definition
Traceback (most recent call last):
  File "e:\OneDrive - Personal\OneDrive\Projects\WESS\Python\WalkerResampler.py", line 23, in <module>   
    scn.load(['00_7'])
  File "C:\Python39\lib\site-packages\satpy\scene.py", line 1275, in load
    self._read_datasets_from_storage(**kwargs)
  File "C:\Python39\lib\site-packages\satpy\scene.py", line 1298, in _read_datasets_from_storage
    return self._read_dataset_nodes_from_storage(nodes, **kwargs)
  File "C:\Python39\lib\site-packages\satpy\scene.py", line 1304, in _read_dataset_nodes_from_storage    
    loaded_datasets = self._load_datasets_by_readers(reader_datasets, **kwargs)
  File "C:\Python39\lib\site-packages\satpy\scene.py", line 1329, in _load_datasets_by_readers
    new_datasets = reader_instance.load(ds_ids, **kwargs)
  File "C:\Python39\lib\site-packages\satpy\readers\yaml_reader.py", line 963, in load
    ds = self._load_dataset_with_area(dsid, coords, **kwargs)
  File "C:\Python39\lib\site-packages\satpy\readers\yaml_reader.py", line 851, in _load_dataset_with_area
    ds = self._load_dataset_data(file_handlers, dsid, **kwargs)
  File "C:\Python39\lib\site-packages\satpy\readers\yaml_reader.py", line 723, in _load_dataset_data     
    proj = self._load_dataset(dsid, ds_info, file_handlers, **kwargs)
  File "C:\Python39\lib\site-packages\satpy\readers\yaml_reader.py", line 699, in _load_dataset
    projectable = fh.get_dataset(dsid, ds_info)
  File "C:\Python39\lib\site-packages\satpy\readers\goes_imager_nc.py", line 1020, in get_dataset
    data = data.where(self.meta['earth_mask'])
  File "C:\Python39\lib\site-packages\satpy\readers\goes_imager_nc.py", line 787, in meta
    area_def_uni = self._get_area_def_uniform_sampling(
  File "C:\Python39\lib\site-packages\satpy\readers\goes_imager_nc.py", line 704, in _get_area_def_uniform_sampling
    xmax, ymax = get_geostationary_angle_extent(
  File "C:\Python39\lib\site-packages\satpy\readers\utils.py", line 90, in get_geostationary_angle_extent
    a, b = _get_geostationary_semi_axes(geos_area)
  File "C:\Python39\lib\site-packages\satpy\readers\utils.py", line 84, in _get_geostationary_semi_axes
    return proj4_radius_parameters(geos_area.crs)
AttributeError: 'area' object has no attribute 'crs'

Environment Info:

• OS: Windows 10 & 11
• Satpy Version: 0.36.0
• PyResample Version: 1.22.3

Relevant Readers:

• Readers: goes-imager_nc: ok

Note that neither cartopy nor geoviews are installed.

Additional context
A friend and I may have tracked down a potential bug in line 705 of readers\goes_imager_nc.py:

xmax, ymax = get_geostationary_angle_extent(
                namedtuple('area', ['proj_dict'])(proj_dict))

When geos_area is later created using this namedtuple, it tries to reference a 'crs' attribute which, in the area definition just before, is not initially defined.

proj_dict = {'a': EQUATOR_RADIUS,
             'b': POLE_RADIUS,
             'lon_0': lon0,
             'h': ALTITUDE,
             'proj': 'geos',
             'units': 'm'}

I'm not sure how much help this is, but it may help point you in the right direction.

Thanks in advance for the help.

[djhoese]

So it looks like this is a case not caught by the tests and definitely a bug. Looks like @sfinkens added this reader and this hacky bit of code that creates a nametuple. Pyresample (and therefore satpy) is slowly moving away from using PROJ.4 dictionaries wherever possible. Recently (within the last year) I updated uses of proj_dict in Satpy to use .crs on AreaDefinition objects including this _get_geostationary_semi_axes function, but I must have missed this usage of proj_dict in the reader since it isn't a normal attribute access and isn't covered in the tests.

Maybe @sfinkens has some ideas for how to fix this.

[sfinkens]

Sorry for taking to long to reply. Indeed that looks hacky! I'll assign it to me and look into it as soon as possible.