AttributeError: 'DataTree' object has no attribute 'sweep_group_name'

[aladinor]

• xradar version: 0.4.3
• Python version: 3.12
• Operating System: Ubuntu

Description

Hi every one,

I've tried to reproduce this grinding example using sigmet files available in amazon.

import xradar as xd
import fsspec
from datetime import datetime
import pyart
import matplotlib.pyplot as plt

def create_query(date, radar_site):
    """
    Creates a string for quering the IDEAM radar files stored in AWS bucket
    :param date: date to be queried. e.g datetime(2021, 10, 3, 12). Datetime python object
    :param radar_site: radar site e.g. Guaviare
    :return: string with a IDEAM radar bucket format
    """
    if (date.hour != 0) and (date.minute != 0):
        return f"l2_data/{date:%Y}/{date:%m}/{date:%d}/{radar_site}/{radar_site[:3].upper()}{date:%y%m%d%H%M}"
    elif (date.hour != 0) and (date.minute == 0):
        return f"l2_data/{date:%Y}/{date:%m}/{date:%d}/{radar_site}/{radar_site[:3].upper()}{date:%y%m%d%H}"
    else:
        return f"l2_data/{date:%Y}/{date:%m}/{date:%d}/{radar_site}/{radar_site[:3].upper()}{date:%y%m%d}"

def main():
    date_query = datetime(2022, 10, 6)
    radar_name = "Guaviare"
    query = create_query(date=date_query, radar_site=radar_name)
    str_bucket = "s3://s3-radaresideam/"
    fs = fsspec.filesystem("s3", anon=True)
    radar_files = sorted(fs.glob(f"{str_bucket}{query}*"))
    file = fsspec.open_local(
        f"filecache::s3://{radar_files[0]}",
        s3={"anon": True},
        filecache={"cache_storage": "/tmp/radar"},
    )

    tree1 = xd.io.open_iris_datatree(file)
    radar = pyart.xradar.Xradar(tree1, default_sweep='sweep_0')
    # Grid using 11 vertical levels, and 101 horizontal grid cells at a resolution on 1 km
    grid = pyart.map.grid_from_radars(
        (radar,),
        grid_shape=(1, 501, 501),
        grid_limits=(
            (0.0, 10_000),
            (-300_000.0, 300_000.0),
            (-300_000, 300_000.0),
        ),
    )

    display = pyart.graph.GridMapDisplay(grid)
    display.plot_grid(
        "DBZH", level=0, vmin=-20, vmax=60, cmap="pyart_ChaseSpectral"
    )
    plt.show()
    print(1)

if __name__ == "__main__":
    main()

However, I got the following error,

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
Cell In[6], line 1
----> 1 radar = pyart.xradar.Xradar(tree1, default_sweep='sweep_0')

File [~/miniconda3/envs/raw2zarr/lib/python3.12/site-packages/pyart/xradar/accessor.py:274](http://127.0.0.1:7022/~/miniconda3/envs/raw2zarr/lib/python3.12/site-packages/pyart/xradar/accessor.py#line=273), in Xradar.__init__(self, xradar, default_sweep, scan_type)
    272 self.xradar = xradar
    273 self.scan_type = scan_type or "ppi"
--> 274 self.combined_sweeps = self._combine_sweeps(self.xradar)
    275 self.fields = self._find_fields(self.combined_sweeps)
    276 self.time = dict(
    277     data=(self.combined_sweeps.time - self.combined_sweeps.time.min()).astype(
    278         "int64"
   (...)
    282     calendar="gregorian",
    283 )

File [~/miniconda3/envs/raw2zarr/lib/python3.12/site-packages/pyart/xradar/accessor.py:552](http://127.0.0.1:7022/~/miniconda3/envs/raw2zarr/lib/python3.12/site-packages/pyart/xradar/accessor.py#line=551), in Xradar._combine_sweeps(self, radar)
    549 def _combine_sweeps(self, radar):
    550     # Loop through and extract the different datasets
    551     ds_list = []
--> 552     for sweep in radar.sweep_group_name.values:
    553         ds_list.append(radar[sweep].ds.drop_duplicates("azimuth"))
    555     # Merge based on the sweep number

File [~/miniconda3/envs/raw2zarr/lib/python3.12/site-packages/datatree/common.py:54](http://127.0.0.1:7022/~/miniconda3/envs/raw2zarr/lib/python3.12/site-packages/datatree/common.py#line=53), in TreeAttrAccessMixin.__getattr__(self, name)
     52         with suppress(KeyError):
     53             return source[name]
---> 54 raise AttributeError(
     55     f"{type(self).__name__!r} object has no attribute {name!r}"
     56 )

AttributeError: 'DataTree' object has no attribute 'sweep_group_name'

Digging around, I found out that the xd.io.open_iris_datatree data loader is not adding the sweep_group_name, sweep_fixed_angle, and radar_parameters to the datatree. I worked around and I submitted a push request where this issue is solved following the FM301 standard WMO data model

[kmuehlbauer]

@aladinor Thanks for raising this.

We need to be careful here. AFAIK sweep_group_name and sweep_fixed_angle are not part of FM 301-2022 WMO-CF RADIAL (as written in https://library.wmo.int/viewer/66287, pp. 944-970).

There might be an issue with the CfRadial1 reader which obviously has these available.

I agree that radar_parameters subgroup should be made available.

[kmuehlbauer]

I see now, that xradar actually defines sweep_group_name and sweep_fixed_angle in xradar.model.required_root_vars. This might be due to inconsistency between the CFRadialDoc-v2.1-20190901 document in https://github.com/NCAR/CfRadial/tree/master/docs and the above linked official WMO document.

@mgrover1 How should we move on here? Removing these two variables from required_root_vars seems reasonable. Also re-checking mandatory, conditional and optional variables/attributes seems necessary now.

[mgrover1]

Let's go ahead and remove sweep_group_name and sweep_fixed_angle from required_root_vars!

[aladinor]

Based on this, I will remove sweep_group_name and sweep_fixed_angle from #165 before merging.

[mgrover1]

Thanks for raising this issue @aladinor - I will work to update the Py-ART side to fix this as well.

[mgrover1]

@aladinor - with #167 merged, a new release for xradar, and the following PR for Py-ART https://github.com/ARM-DOE/pyart/pull/1537

This issue should be resolved; I tested it with existing ODIM_H5 files, as well as the code you included above that would have previously errored. Feel free to give that branch a try - hoping to merge + cut a minor release of Py-ART for you to try out. Thanks again for raising this issue.