Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

use cf-xarray heuristic to identify bound variables. Add test #54

Merged
merged 4 commits into from
Feb 29, 2024
Merged

use cf-xarray heuristic to identify bound variables. Add test #54

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
a585d9e
use cf-xarray heuristic to identify bound variables. Add test
huard Feb 27, 2024
2bc0b21
added checks in test_auxiliary_variables
huard Feb 28, 2024
4df99fb
merge
huard Feb 29, 2024
c99e29f
added bounds description
huard Feb 29, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 2 additions & 1 deletion CHANGES.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,8 @@

## [Unreleased](https://github.com/crim-ca/stac-populator) (latest)

<!-- insert list items of new changes here -->
* Make sure *bounds* variables are given the auxiliary type attribute.
* Fix for variables that have no attributes.

## [0.6.0](https://github.com/crim-ca/stac-populator/tree/0.6.0) (2024-02-22)

Expand Down
54 changes: 29 additions & 25 deletions STACpopulator/extensions/datacube.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -184,6 +184,7 @@ def dimensions(self) -> dict[str, Dimension]:
def variables(self) -> dict[str, Variable]:
"""Return Variable objects required for Datacube extension."""
variables = {}
bounds = self.bounds()

for name, meta in self.attrs["variables"].items():
if name in self.attrs["dimensions"]:
Expand All @@ -192,44 +193,47 @@ def variables(self) -> dict[str, Variable]:
# Some variables like "time_bnds" in some model files do not have any attributes.
attrs = meta.get("attributes", {})

self._infer_variable_units_description(name, attrs)
if name in bounds:
# Bounds are auxiliary variables
dtype = VariableType.AUXILIARY.value

# We can safely assume that the bounds variable has the same units as the variable it bounds.
if "units" not in attrs:
if (u := self.attrs["variables"][bounds[name]].get("attributes", {}).get("units")) is not None:
attrs["units"] = u

elif self.is_coordinate(attrs):
# Using the CF-xarray heuristics to determine if variable is a coordinate.
dchandan marked this conversation as resolved.
Show resolved Hide resolved
dtype = VariableType.AUXILIARY.value
else:
dtype = VariableType.DATA.value

variables[name] = Variable(
properties=dict(
dimensions=meta["shape"],
type=VariableType.AUXILIARY.value if self.is_coordinate(attrs) else VariableType.DATA.value,
type=dtype,
description=attrs.get("description", attrs.get("long_name", "")),
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think we should add a description to the bounds variable too. Something like "bounds for the time coordinate" as in #51. We can construct an appropriate description string in the if-elif-else block before this and use that here.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

So I wanted to hear your thoughts on this.

  1. What's the rationale for "augmenting" the metadata attributes when creating the STAC items?
  2. Is the datacube extension logic the appropriate place to do this?
  3. Will it scale well to other missing infos?
  4. Are we worried about a "surprise" factor coming from having information in the STAC item that does not reflect what people will find in the actual dataset?

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Good questions, here's what I think:

  1. I think the rationale would be that the STAC item should be as informative as reasonably possible. If, at the STAC item generation phase adding information that doesn't exist in the file metadata achieves that goal, then I don't see why not. In general, there is no principle/constraint requiring one-to-one correspondence between metadata in a STAC representation and that in the underlying asset. Here, we just use metadata from the CMIP6 files to guide the generation of STAC items, but even if the CMIP6 files were poorly constructed with relevant metadata, we could still generate the STAC items by programmatically adding the relevant metadata at item generation time.
  2. I really don't know what is the purpose of the datacube extension. I have never used it, and I don't envision using it. But from looking at the data it represents, it seems it just described the dimensions and variable available in an asset, and if that is the case, then we definitely can use the extension to add a description to a variable that is listed by the extension.
  3. I don't see why not. Going back to point 1, I think it's all about including whatever one feels is necessary for describing the data variables and dimensions.
  4. No, I am not. Going back to point 1, there is no constraint about a one-to-one correspondence between metadata described in the STAC representation and that in an underlying file. Consider the case of some obscure dataset where the underlying .nc files do not have any metadata whatsoever (on variable level or global). There is nothing preventing me from appropriately describing the metadata in the STAC layer and make the dataset available via the catalog. In this case, none of the metadata in the STAC representation would be available in the file itself, and that is perfectly OK.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ok, I'm happy adopting your perspective on 1 and 4. I'm less convinced about 2 and 3 and I believe we should spend some time thinking about the "metadata augmentation architecture" before writing too much code. I'm happy to add the description you mentioned, but I don't think this should serve as a template to add other kinds of information.

My discomfort comes from mixing process logic (NcML -> STAC) with thematic knowledge (metadata fixes) in the same code base. I think we can get the process logic fairly robust and maintenance should be minimal. The thematic knowledge will always grow with new datasets. Mixing both has the potential to generate maintenance work in the future.

unit=attrs.get("units", ""),
)
)
return variables

def _infer_variable_units_description(self, name, attrs):
"""Try to infer the units and description of some simple coordinate variables."""
if name == "time_bnds":
related_variable = "time"
attrs["description"] = "bounds for the time coordinate"
elif name == "lat_bnds":
related_variable = "lat"
attrs["description"] = "bounds for the latitude coordinate"
elif name == "lon_bnds":
related_variable = "lon"
attrs["description"] = "bounds for the longitude coordinate"
else:
return

try:
attrs["units"] = self.attrs["variables"][related_variable]["attributes"]["units"]
except KeyError:
pass
def bounds(self):
"""Return a list of variables that are bounds for other variables."""
out = {}
for name, meta in self.attrs["variables"].items():
attrs = meta.get("attributes", {})
if "bounds" in attrs:
out[attrs["bounds"]] = name
return out

def is_coordinate(self, attrs: MutableMapping[str, Any]) -> bool:
"""Return whether variable is a coordinate."""

if (desc := attrs.get("description", None)) is not None:
if "bounds for" in desc:
return True
def is_coordinate(self, attrs: MutableMapping[str, Any]) -> bool:
"""Return whether variable is a coordinate.

- data: a variable indicating some measured value, for example "precipitation", "temperature", etc.
- auxiliary: a variable that contains coordinate data, but isn't a dimension in cube:dimensions.
"""
for key, criteria in self.coordinate_criteria.items():
for criterion, expected in criteria.items():
if attrs.get(criterion, None) in expected:
Expand Down
133 changes: 133 additions & 0 deletions tests/data/clt_Amon_EC-Earth3_historical_r2i1p1f1_gr_185001-201412.xml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,133 @@
<ncml:netcdf xmlns:ncml="http://www.unidata.ucar.edu/namespaces/netcdf/ncml-2.2" location="https://redoak.cs.toronto.edu/twitcher/ows/proxy/thredds/dodsC/datasets/CMIP6/CMIP/EC-Earth-Consortium/EC-Earth3/historical/r2i1p1f1/Amon/clt/gr/v20201215/clt_Amon_EC-Earth3_historical_r2i1p1f1_gr_185001-201412.nc">
<ncml:attribute name="Conventions" value="CF-1.7 CMIP-6.2"/>
<ncml:attribute name="activity_id" value="CMIP"/>
<ncml:attribute name="branch_time" type="double" value="0.0"/>
<ncml:attribute name="contact" value="[email protected]"/>
<ncml:attribute name="creation_date" value="2019-08-13T19:08:49Z"/>
<ncml:attribute name="data_specs_version" value="01.00.27"/>
<ncml:attribute name="experiment" value="all-forcing simulation of the recent past"/>
<ncml:attribute name="experiment_id" value="historical"/>
<ncml:attribute name="external_variables" value="areacella"/>
<ncml:attribute name="forcing_index" type="int" value="1"/>
<ncml:attribute name="frequency" value="mon"/>
<ncml:attribute name="further_info_url" value="https://furtherinfo.es-doc.org/CMIP6.EC-Earth-Consortium.EC-Earth3.historical.none.r2i1p1f1"/>
<ncml:attribute name="grid" value="ORCA1T255"/>
<ncml:attribute name="grid_label" value="gr"/>
<ncml:attribute name="initialization_index" type="int" value="1"/>
<ncml:attribute name="institution" value="AEMET, Spain; BSC, Spain; CNR-ISAC, Italy; DMI, Denmark; ENEA, Italy; FMI, Finland; Geomar, Germany; ICHEC, Ireland; ICTP, Italy; IDL, Portugal; IMAU, The Netherlands; IPMA, Portugal; KIT, Karlsruhe, Germany; KNMI, The Netherlands; Lund University, Sweden; Met Eireann, Ireland; NLeSC, The Netherlands; NTNU, Norway; Oxford University, UK; surfSARA, The Netherlands; SMHI, Sweden; Stockholm University, Sweden; Unite ASTR, Belgium; University College Dublin, Ireland; University of Bergen, Norway; University of Copenhagen, Denmark; University of Helsinki, Finland; University of Santiago de Compostela, Spain; Uppsala University, Sweden; Utrecht University, The Netherlands; Vrije Universiteit Amsterdam, the Netherlands; Wageningen University, The Netherlands. Mailing address: EC-Earth consortium, Rossby Center, Swedish Meteorological and Hydrological Institute/SMHI, SE-601 76 Norrkoping, Sweden"/>
<ncml:attribute name="institution_id" value="EC-Earth-Consortium"/>
<ncml:attribute name="mip_era" value="CMIP6"/>
<ncml:attribute name="parent_activity_id" value="CMIP"/>
<ncml:attribute name="parent_experiment_id" value="piControl"/>
<ncml:attribute name="parent_mip_era" value="CMIP6"/>
<ncml:attribute name="parent_source_id" value="EC-Earth3"/>
<ncml:attribute name="parent_sub_experiment_id" value="no parent"/>
<ncml:attribute name="parent_time_units" value="days since 1850-01-01"/>
<ncml:attribute name="physics_index" type="int" value="1"/>
<ncml:attribute name="product" value="model-output"/>
<ncml:attribute name="realization_index" type="int" value="2"/>
<ncml:attribute name="realm" value="atmos"/>
<ncml:attribute name="source" value="EC-Earth3 (2019): aerosol: none atmos: IFS cy36r4 (TL255, linearly reduced Gaussian grid equivalent to 512 x 256 longitude/latitude; 91 levels; top level 0.01 hPa) atmosChem: none land: HTESSEL (land surface scheme built in IFS) landIce: none ocean: NEMO3.6 (ORCA1 tripolar primarily 1 deg with meridional refinement down to 1/3 degree in the tropics; 362 x 292 longitude/latitude; 75 levels; top grid cell 0-1 m) ocnBgchem: none seaIce: LIM3"/>
<ncml:attribute name="source_id" value="EC-Earth3"/>
<ncml:attribute name="source_type" value="AOGCM"/>
<ncml:attribute name="sub_experiment" value="none"/>
<ncml:attribute name="sub_experiment_id" value="none"/>
<ncml:attribute name="table_id" value="Amon"/>
<ncml:attribute name="table_info" value="Creation Date:(20 July 2018) MD5:b2dc68c38656de0e72d665d82e3c038c"/>
<ncml:attribute name="title" value="EC-Earth3 output prepared for CMIP6"/>
<ncml:attribute name="variable_id" value="clt"/>
<ncml:attribute name="variant_info" value="forcing: Nat.Ant. Member generated from autosubmit member fc0"/>
<ncml:attribute name="variant_label" value="r2i1p1f1"/>
<ncml:attribute name="license" value="CMIP6 model data produced by EC-Earth-Consortium is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License (https://creativecommons.org/licenses). Consult https://pcmdi.llnl.gov/CMIP6/TermsOfUse for terms of use governing CMIP6 output, including citation requirements and proper acknowledgment. Further information about this data, including some limitations, can be found via the further_info_url (recorded as a global attribute in this file) . The data producers and data providers make no warranty, either express or implied, including, but not limited to, warranties of merchantability and fitness for a particular purpose. All liabilities arising from the supply of the information (including any liability arising in negligence) are excluded to the fullest extent permitted by law."/>
<ncml:attribute name="cmor_version" value="3.4.0"/>
<ncml:attribute name="nominal_resolution" value="100 km"/>
<ncml:attribute name="branch_method" value="standard"/>
<ncml:attribute name="branch_time_in_parent" type="double" value="164358.0"/>
<ncml:attribute name="branch_time_in_child" type="double" value="0.0"/>
<ncml:attribute name="parent_variant_label" value="r1i1p1f1"/>
<ncml:attribute name="history" value="This file was generated on Fri Nov 3 20:42:43 2023 by combining two or more individual files for this dataset."/>
<ncml:attribute name="_CoordSysBuilder" value="ucar.nc2.dataset.conv.CF1Convention"/>
<ncml:dimension name="time" length="1980" isUnlimited="true"/>
<ncml:dimension name="lat" length="256"/>
<ncml:dimension name="lon" length="512"/>
<ncml:dimension name="bnds" length="2"/>
<group xmlns="http://www.unidata.ucar.edu/namespaces/netcdf/ncml-2.2" name="CFMetadata">
<attribute name="geospatial_lon_min" value=".0" type="float"/>
<attribute name="geospatial_lat_min" value="-89.46282" type="float"/>
<attribute name="geospatial_lon_max" value="359.29688" type="float"/>
<attribute name="geospatial_lat_max" value="89.46282" type="float"/>
<attribute name="geospatial_lon_units" value="degrees_east"/>
<attribute name="geospatial_lat_units" value="degrees_north"/>
<attribute name="geospatial_lon_resolution" value="0.703125"/>
<attribute name="geospatial_lat_resolution" value="0.7016691764705881"/>
<attribute name="time_coverage_start" value="1850-01-16T12:00:00Z"/>
<attribute name="time_coverage_end" value="2014-12-16T12:00:00Z"/>
<attribute name="time_coverage_units" value="seconds"/>
<attribute name="time_coverage_resolution" value="2629720.0"/>
<attribute name="time_coverage_duration" value="P0Y0M60234DT0H0M0.000S"/>
</group>
<group xmlns="http://www.unidata.ucar.edu/namespaces/netcdf/ncml-2.2" name="NCISOMetadata">
<attribute name="metadata_creation" value="2024-02-27"/>
<attribute name="nciso_version" value="2.2.3"/>
</group>
<group xmlns="http://www.unidata.ucar.edu/namespaces/netcdf/ncml-2.2" name="THREDDSMetadata">
<attribute name="id" value="datasets/CMIP6/CMIP/EC-Earth-Consortium/EC-Earth3/historical/r2i1p1f1/Amon/clt/gr/v20201215/clt_Amon_EC-Earth3_historical_r2i1p1f1_gr_185001-201412.nc"/>
<attribute name="full_name" value="v20201215/clt_Amon_EC-Earth3_historical_r2i1p1f1_gr_185001-201412.nc"/>
<group name="services">
<attribute name="httpserver_service" value="https://redoak.cs.toronto.edu/twitcher/ows/proxy/thredds/fileServer/datasets/CMIP6/CMIP/EC-Earth-Consortium/EC-Earth3/historical/r2i1p1f1/Amon/clt/gr/v20201215/clt_Amon_EC-Earth3_historical_r2i1p1f1_gr_185001-201412.nc"/>
<attribute name="opendap_service" value="https://redoak.cs.toronto.edu/twitcher/ows/proxy/thredds/dodsC/datasets/CMIP6/CMIP/EC-Earth-Consortium/EC-Earth3/historical/r2i1p1f1/Amon/clt/gr/v20201215/clt_Amon_EC-Earth3_historical_r2i1p1f1_gr_185001-201412.nc"/>
<attribute name="nccs_service" value="https://redoak.cs.toronto.edu/twitcher/ows/proxy/thredds/ncss/datasets/CMIP6/CMIP/EC-Earth-Consortium/EC-Earth3/historical/r2i1p1f1/Amon/clt/gr/v20201215/clt_Amon_EC-Earth3_historical_r2i1p1f1_gr_185001-201412.nc/dataset.html"/>
</group>
<group name="dates">
<attribute name="date" value="2023-11-03T20:43:10.555Z" type="modified"/>
</group>
</group>
<ncml:variable name="time_bnds" shape="time bnds" type="double">
<ncml:attribute name="units" value="days since 1850-01-01"/>
<ncml:attribute name="_ChunkSizes" type="int" value="1 2"/>
</ncml:variable>
<ncml:variable name="lat_bnds" shape="lat bnds" type="double">
<ncml:attribute name="_ChunkSizes" type="int" value="256 2"/>
</ncml:variable>
<ncml:variable name="lon_bnds" shape="lon bnds" type="double">
<ncml:attribute name="_ChunkSizes" type="int" value="512 2"/>
</ncml:variable>
<ncml:variable name="clt" shape="time lat lon" type="float">
<ncml:attribute name="_FillValue" type="float" value="1.0E20"/>
<ncml:attribute name="standard_name" value="cloud_area_fraction"/>
<ncml:attribute name="long_name" value="Total Cloud Fraction"/>
<ncml:attribute name="comment" value="Total cloud area fraction for the whole atmospheric column, as seen from the surface or the top of the atmosphere. Includes both large-scale and convective cloud."/>
<ncml:attribute name="units" value="%"/>
<ncml:attribute name="cell_methods" value="area: time: mean"/>
<ncml:attribute name="cell_measures" value="area: areacella"/>
<ncml:attribute name="history" value="2019-08-13T19:08:49Z altered by CMOR: Reordered dimensions, original order: lat lon time."/>
<ncml:attribute name="missing_value" type="float" value="1.0E20"/>
<ncml:attribute name="_ChunkSizes" type="int" value="1 256 512"/>
</ncml:variable>
<ncml:variable name="time" shape="time" type="double">
<ncml:attribute name="bounds" value="time_bnds"/>
<ncml:attribute name="axis" value="T"/>
<ncml:attribute name="long_name" value="time"/>
<ncml:attribute name="standard_name" value="time"/>
<ncml:attribute name="units" value="days since 1850-01-01"/>
<ncml:attribute name="calendar" value="gregorian"/>
<ncml:attribute name="_ChunkSizes" type="int" value="1"/>
<ncml:attribute name="_CoordinateAxisType" value="Time"/>
</ncml:variable>
<ncml:variable name="lat" shape="lat" type="double">
<ncml:attribute name="bounds" value="lat_bnds"/>
<ncml:attribute name="units" value="degrees_north"/>
<ncml:attribute name="axis" value="Y"/>
<ncml:attribute name="long_name" value="latitude"/>
<ncml:attribute name="standard_name" value="latitude"/>
<ncml:attribute name="_CoordinateAxisType" value="Lat"/>
</ncml:variable>
<ncml:variable name="lon" shape="lon" type="double">
<ncml:attribute name="bounds" value="lon_bnds"/>
<ncml:attribute name="units" value="degrees_east"/>
<ncml:attribute name="axis" value="X"/>
<ncml:attribute name="long_name" value="Longitude"/>
<ncml:attribute name="standard_name" value="longitude"/>
<ncml:attribute name="_CoordinateAxisType" value="Lon"/>
</ncml:variable>
</ncml:netcdf>
19 changes: 19 additions & 0 deletions tests/test_cmip6_datacube.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -35,3 +35,22 @@ def test_datacube_helper():
assert len(schemas) >= 2
assert "item.json" in schemas[0]
assert "datacube" in schemas[1]


def test_auxiliary_variables():
# https://github.com/crim-ca/stac-populator/issues/52

file_path = DIR / "data" / "clt_Amon_EC-Earth3_historical_r2i1p1f1_gr_185001-201412.xml"

ds = xncml.Dataset(filepath=str(file_path))
attrs = ds.to_cf_dict()
attrs["access_urls"] = {"HTTPServer": "http://example.com"}
item = CMIP6Helper(attrs, GeoJSONPolygon).stac_item()

dc = DataCubeHelper(attrs)
dc_ext = DatacubeExtension.ext(item, add_if_missing=True)
dc_ext.apply(dimensions=dc.dimensions, variables=dc.variables)

p = dc_ext.properties
assert set(['time', 'lat', 'lon']) == set(p['cube:dimensions'].keys())
assert p["cube:variables"]["lon_bnds"]["unit"] == "degrees_east"
huard marked this conversation as resolved.
Show resolved Hide resolved
Loading