C-iTRACE tracer pointing to NetCDF4 files

recipes/citrace_tracer/meta.yaml

@@ -0,0 +1,37 @@
+title: 'C-iTRACE-tracers'
+description: 'Coarse resolution, ocean-only simulation from 22ka to present using the isotope-enabled ocean component of CESM1'
+pangeo_forge_version: '0.9.0'
+pangeo_notebook_version: '2022.06.02'
+recipes:
+  - id: citrace_tracer
+    object: 'recipe:recipe'
+provenance:
+  providers:
+    - name: 'University of Colorado Boulder'
+      description: 'University of Colorado Boulder'
+      roles:
+        - producer
+      url: https://sites.google.com/colorado.edu/citrace
+    - name: 'University of Wisconsin-Madison'
+      description: 'University of Wisconsin-Madison'
+      roles:
+        - producer
+      url: https://sites.google.com/colorado.edu/citrace
+    - name: 'Ohio State University'
+      description: 'Ohio State University'
+      roles:
+        - producer
+      url: https://sites.google.com/colorado.edu/citrace
+    - name: 'UCAR/NCAR'
+      description: 'University Corporation for Atmospheric Research/National Center for Atmospheric Research'
+      roles:
+        - curator
+        - licensor
+      url: https://gdex.ucar.edu/dataset/204_ajahn.html
+  license: 'Creative Commons Attribution 4.0 International'
+maintainers:
+  - name: 'Jordan Landers'
+    orcid: '0000-0001-9772-7617'
+    github: jordanplanders
+bakery:
+  id: 'pangeo-ldeo-nsf-earthcube'

recipes/citrace_tracer/recipe.py

@@ -0,0 +1,92 @@
+from pangeo_forge_recipes.patterns import ConcatDim, FilePattern, MergeDim
+from pangeo_forge_recipes.recipes.xarray_zarr import XarrayZarrRecipe
+
+# C-iTRACE covers a number of tracer quantities
+variables = [
+    'd18O',
+    'ABIO_D14Cocn',
+    'ABIO_D14Catm',
+    'CISO_DIC_d13C',
+    'ND143',
+    'ND144',
+    'PD',
+    'SALT',
+    'TEMP',
+    'IAGE',
+]
+
+
+def make_url(time, variable):
+    url_d = {
+        'IAGE': 'https://figshare.com/ndownloader/files/38534591',
+        'd18O': 'https://figshare.com/ndownloader/files/38530430',
+        'ABIO_D14Cocn': 'https://figshare.com/ndownloader/files/38231892',
+        'ABIO_D14Catm': 'https://figshare.com/ndownloader/files/38231991',
+        'CISO_DIC_d13C': 'https://figshare.com/ndownloader/files/38526806',
+        'ND143': 'https://figshare.com/ndownloader/files/38232651',
+        'ND144': 'https://figshare.com/ndownloader/files/38232060',
+        'SALT': 'https://figshare.com/ndownloader/files/38541851',
+        'TEMP': 'https://figshare.com/ndownloader/files/38543534',
+        'PD': 'https://figshare.com/ndownloader/files/38543969',
+    }
+
+    url = url_d[variable]
+    return url
+
+
+# these are single ~6 Gb files, each covering the full timeseries
+time_concat_dim = ConcatDim('time', [0])
+pattern = FilePattern(make_url, time_concat_dim, MergeDim(name='variable', keys=variables))
+
+
+# clean up the dataset so that lat and lon are included as dimension coordinates
+def postproc(ds):
+
+    import xarray as xr
+
+    variable = [var for var in ds.data_vars.keys() if 'bound' not in var][0]
+
+    coords = [key for key in ds.coords.keys()]
+    coord_d = {}
+    for coord_var in ['lat', 'lon']:
+        coord_d[coord_var] = [coord for coord in coords if coord_var in coord.lower()][0]
+
+    times = ds.coords['time'].values
+    lats = [
+        ds.coords[coord_d['lat']].values[ik][0]
+        for ik in range(len(ds.coords[coord_d['lat']].values))
+    ]
+    lons = ds.coords[coord_d['lon']][0].values % 360
+
+    _ds = xr.Dataset()
+    _ds.coords['lat'] = (('lat'), lats)
+    _ds.coords['lon'] = (('lon'), lons)
+    _ds.coords['time'] = (('time'), times)
+
+    if 'z_t' in ds.coords:
+        z_ts = ds.coords['z_t'].values
+        _ds.coords['z_t'] = (('z_t'), z_ts)
+        coord_tuple = ('time', 'z_t', 'lat', 'lon')
+    else:
+        coord_tuple = ('time', 'lat', 'lon')
+
+    _ds[variable] = (coord_tuple, ds[variable].values)
+
+    _ds = _ds.sortby(['lon', 'lat'])
+    return _ds
+
+
+# Create recipe object
+# use subset_inputs to make the processing more tractable.
+# using `target_chunks` instead of specifying `nitems_per_file`
+# in ConcatDim massively decreased run time.
+recipe = XarrayZarrRecipe(
+    pattern,
+    inputs_per_chunk=1,
+    consolidate_zarr=True,
+    subset_inputs={'time': 120},
+    target_chunks={'time': 1},
+    process_chunk=postproc,
+    copy_input_to_local_file=False,
+    xarray_open_kwargs={'decode_coords': True, 'use_cftime': True, 'decode_times': True},
+)