it seems the climatology file in the evaluation.ipynb quick setup guide seems to be missing

[ShileiCao]

FileNotFoundError: No such file or directory: 'gs://weatherbench2/datasets/era5-hourly-climatology/1990-2019_6h_64x32_equiangular_conservative.zarr'

[ShileiCao]

when i replace it with “with_poles”, the file can be read. But it went wrong when I run the evaluation：

evaluate_in_memory(data_config, eval_configs)

---

ValueError Traceback (most recent call last)
in <cell line: 1>()
----> 1 evaluate_in_memory(data_config, eval_configs) # Takes around 5 minutes

9 frames
/usr/local/lib/python3.10/dist-packages/weatherbench2/evaluation.py in evaluate_in_memory(data_config, eval_configs, skipna)
515 """
516 for eval_name, eval_config in eval_configs.items():
--> 517 _evaluate_all_metrics(eval_name, eval_config, data_config, skipna=skipna)
518
519

/usr/local/lib/python3.10/dist-packages/weatherbench2/evaluation.py in _evaluate_all_metrics(eval_name, eval_config, data_config, skipna)
475 truth = truth.sel(time=forecast.valid_time)
476
--> 477 results = _metric_and_region_loop(forecast, truth, eval_config, skipna=skipna)
478
479 logging.info(f'Logging Evaluation complete:\n{results}')

/usr/local/lib/python3.10/dist-packages/weatherbench2/evaluation.py in _metric_and_region_loop(forecast, truth, eval_config, skipna, compute_chunk)
421 [region_name], coords={'region': [region_name]}
422 )
--> 423 tmp_result = eval_fn(
424 forecast=forecast, truth=truth, region=region, skipna=skipna
425 )

/usr/local/lib/python3.10/dist-packages/weatherbench2/metrics.py in compute(self, forecast, truth, region, skipna)
131 f"Forecast has neither valid_time or init_time dimension {forecast}"
132 )
--> 133 return self.compute_chunk(
134 forecast, truth, region=region, skipna=skipna
135 ).mean(

/usr/local/lib/python3.10/dist-packages/weatherbench2/metrics.py in compute_chunk(self, forecast, truth, region, skipna)
405 forecast_anom = forecast - climatology_chunk
406 truth_anom = truth - climatology_chunk
--> 407 return _spatial_average(
408 forecast_anom * truth_anom,
409 region=region,

/usr/local/lib/python3.10/dist-packages/weatherbench2/metrics.py in _spatial_average(dataset, region, skipna)
154 dataset: Spatially averaged metric.
155 """
--> 156 weights = get_lat_weights(dataset)
157 if region is not None:
158 dataset, weights = region.apply(dataset, weights)

/usr/local/lib/python3.10/dist-packages/weatherbench2/metrics.py in get_lat_weights(ds)
57 weights = _cell_area_from_latitude(np.deg2rad(ds.latitude.data))
58 weights /= np.mean(weights)
---> 59 weights = ds.latitude.copy(data=weights)
60 return weights
61

/usr/local/lib/python3.10/dist-packages/xarray/core/dataarray.py in copy(self, deep, data)
1300 pandas.DataFrame.copy
1301 """
-> 1302 return self._copy(deep=deep, data=data)
1303
1304 def _copy(

/usr/local/lib/python3.10/dist-packages/xarray/core/dataarray.py in _copy(self, deep, data, memo)
1308 memo: dict[int, Any] | None = None,
1309 ) -> Self:
-> 1310 variable = self.variable._copy(deep=deep, data=data, memo=memo)
1311 indexes, index_vars = self.xindexes.copy_indexes(deep=deep)
1312

/usr/local/lib/python3.10/dist-packages/xarray/core/variable.py in _copy(self, deep, data, memo)
955 ndata = as_compatible_data(data)
956 if self.shape != ndata.shape: # type: ignore[attr-defined]
--> 957 raise ValueError(
958 f"Data shape {ndata.shape} must match shape of object {self.shape}" # type: ignore[attr-defined]
959 )

ValueError: Data shape (1,) must match shape of object (0,)

[raspstephan]

Thank you for noticing. Yes, the 64x32 files in the bucket are wrong (they should be without poles). I just started the upload for the correct files but it might take a few hours. I will then also delete the 64x32 files with poles. Please let me know if it works now.

[ShileiCao]

Thanks for fixing that.
But there is a small error in the beginning when import the apache_beam. It seems to be due to a problem with the pandas version.

---

TypeError Traceback (most recent call last)
in <cell line: 1>()
----> 1 import apache_beam

11 frames
/usr/local/lib/python3.10/dist-packages/apache_beam/init.py in
85 import apache_beam.internal.pickler
86
---> 87 from apache_beam import coders
88 from apache_beam import io
89 from apache_beam import metrics

/usr/local/lib/python3.10/dist-packages/apache_beam/coders/init.py in
15 # limitations under the License.
16 #
---> 17 from apache_beam.coders.coders import *
18 from apache_beam.coders.row_coder import *
19 from apache_beam.coders.typecoders import registry

/usr/local/lib/python3.10/dist-packages/apache_beam/coders/coders.py in
57 import proto
58
---> 59 from apache_beam.coders import coder_impl
60 from apache_beam.coders.avro_record import AvroRecord
61 from apache_beam.portability import common_urns

/usr/local/lib/python3.10/dist-packages/apache_beam/coders/coder_impl.cpython-310-x86_64-linux-gnu.so in init apache_beam.coders.coder_impl()

/usr/local/lib/python3.10/dist-packages/apache_beam/typehints/init.py in
29 pass
30 else:
---> 31 from apache_beam.typehints.pandas_type_compatibility import *
32
33 try:

/usr/local/lib/python3.10/dist-packages/apache_beam/typehints/pandas_type_compatibility.py in
90 ]
91
---> 92 PANDAS_TO_BEAM = {
93 pd.Series([], dtype=dtype).dtype: fieldtype
94 for dtype,

/usr/local/lib/python3.10/dist-packages/apache_beam/typehints/pandas_type_compatibility.py in (.0)
91
92 PANDAS_TO_BEAM = {
---> 93 pd.Series([], dtype=dtype).dtype: fieldtype
94 for dtype,
95 fieldtype in _BIDIRECTIONAL

/usr/local/lib/python3.10/dist-packages/pandas/core/series.py in init(self, data, index, dtype, name, copy, fastpath)
582 data = data.copy()
583 else:
--> 584 data = sanitize_array(data, index, dtype, copy)
585
586 manager = _get_option("mode.data_manager", silent=True)

/usr/local/lib/python3.10/dist-packages/pandas/core/construction.py in sanitize_array(data, index, dtype, copy, allow_2d)
594 _sanitize_non_ordered(data)
595 cls = dtype.construct_array_type()
--> 596 subarr = cls._from_sequence(data, dtype=dtype, copy=copy)
597
598 # GH#846

/usr/local/lib/python3.10/dist-packages/pandas/core/arrays/masked.py in _from_sequence(cls, scalars, dtype, copy)
150 @classmethod
151 def _from_sequence(cls, scalars, *, dtype=None, copy: bool = False) -> Self:
--> 152 values, mask = cls._coerce_to_array(scalars, dtype=dtype, copy=copy)
153 return cls(values, mask)
154

/usr/local/lib/python3.10/dist-packages/pandas/core/arrays/boolean.py in _coerce_to_array(cls, value, dtype, copy)
354 if dtype:
355 assert dtype == "boolean"
--> 356 return coerce_to_array(value, copy=copy)
357
358 def _logical_method(self, other, op):

/usr/local/lib/python3.10/dist-packages/pandas/core/arrays/boolean.py in coerce_to_array(values, mask, copy)
198 integer_like = ("floating", "integer", "mixed-integer-float")
199 if inferred_dtype not in ("boolean", "empty") + integer_like:
--> 200 raise TypeError("Need to pass bool-like values")
201
202 # mypy does not narrow the type of mask_values to npt.NDArray[np.bool_]

TypeError: Need to pass bool-like values