Add WW3 recipe

.github/workflows/deploy.yaml

@@ -56,7 +56,7 @@ jobs:
  f.write(f"max_num_workers={max_num_workers}")
 
  - name: "Deploy recipes"
- uses: "pangeo-forge/deploy-recipe-action@v0.1"
+ uses: "pangeo-forge/deploy-recipe-action@v0.2"
  with:
  select_recipe_by_label: true
  pangeo_forge_runner_config: >
@@ -70,7 +70,8 @@ jobs:
  "service_account_email": "[email protected]",
  "project_id": "leap-pangeo",
  "temp_gcs_location": "gs://leap-scratch/data-library/temp",
- "max_num_workers": ${{ env.max_num_workers }}
+ "max_num_workers": ${{ env.max_num_workers }},
+ "machine_type": "n2-highmem-4"
  },
  "TargetStorage": {
  "fsspec_class": "gcsfs.GCSFileSystem",

feedstock/meta.yaml

@@ -84,6 +84,17 @@ recipes:
  - licensor
  url: https://arxiv.org/abs/2306.08754
  license: "Apache-2.0"
+ - id: WW3
+ object: "ww3:WW3"
+ description: "20 year global ocean surface wave hindcast provided by Ifremer (French Research Institute for Exploitation of the Sea)"
+ provenance:
+ providers:
+ - name: "Ifremer"
+ description: "Ifremer"
+ roles:
+ - host
+ url: https://www.ifremer.fr/en
+ license: "TBW"
 maintainers:
  - name: "Julius Busecke"
  orcid: 0000-0001-8571-865X

feedstock/requirements.txt

@@ -4,3 +4,4 @@
 # once a `0.10.1` release of `pangeo-forge-recipes` is available that includes this fix, we should
 # install from that release instead.
 git+https://github.com/pangeo-forge/pangeo-forge-recipes.git@2739f2264ec385eadd3c73226fb85f5cdbf32a1a
+gcsfs

feedstock/ww3.py

@@ -0,0 +1,59 @@
+"""
+Wave Watch 3
+"""
+import apache_beam as beam
+from pangeo_forge_recipes.patterns import pattern_from_file_sequence
+from pangeo_forge_recipes.transforms import (
+ Indexed,
+ OpenURLWithFSSpec,
+ OpenWithXarray,
+ StoreToZarr,
+ T,
+)
+
+years = range(1993, 2023)
+months = range(1, 13)
+dates = []
+for y in years:
+ for m in months:
+ dates.append((y, m))
+
+
+def make_full_path(date: tuple[int, int]):
+ year, month = date
+ return f'https://data-dataref.ifremer.fr/ww3/GLOBMULTI_ERA5_GLOBCUR_01/GLOB-30M/{year}/FIELD_NC/LOPS_WW3-GLOB-30M_{year}{month:02d}.nc'
+
+
+input_urls = [make_full_path(date) for date in dates]
+pattern = pattern_from_file_sequence(input_urls, concat_dim='time')
+
+
+# does this succeed with all coords stripped?
+class StripCoords(beam.PTransform):
+ @staticmethod
+ def _strip_all_coords(item: Indexed[T]) -> Indexed[T]:
+ """
+ Many netcdfs contain variables other than the one specified in the `variable_id` facet.
+ Set them all to coords
+ """
+ index, ds = item
+ print(f'Preprocessing before {ds =}')
+ ds = ds.reset_coords(drop=True)
+ print(f'Preprocessing after {ds =}')
+ return index, ds
+
+ def expand(self, pcoll: beam.PCollection) -> beam.PCollection:
+ return pcoll | 'Debug: Remove coordinates' >> beam.Map(self._strip_all_coords)
+
+
+WW3 = (
+ beam.Create(pattern.items())
+ | OpenURLWithFSSpec()
+ | OpenWithXarray(xarray_open_kwargs={'preprocess':lambda ds: ds.set_coords('MAPSTA')})
+ | StripCoords()
+ | StoreToZarr(
+ store_name='WW3.zarr',
+ combine_dims=pattern.combine_dim_keys,
+ target_chunks={'time': 200},
+ )
+)