Merge pull request #342 from roocs/regrid_masks_and_fixes

Regridding masks, robustness of dataset_utils functions and bug fixes

.github/workflows/main.yml

@@ -13,6 +13,7 @@ concurrency:
 
 env:
   ESGF_TEST_DATA_VERSION: v1
+  XCLIM_TEST_DATA_VERSION: v2024.8.23
 
 permissions:
   contents: read
@@ -50,7 +51,7 @@ jobs:
     name: Python${{ matrix.python-version }} (PyPI + Tox)
     needs: lint
     runs-on: ubuntu-latest
-    timeout-minutes: 10
+    timeout-minutes: 30
     strategy:
       matrix:
         include:
@@ -94,7 +95,7 @@ jobs:
     name: Python${{ matrix.python-version }} (Anaconda, upstream=${{ matrix.upstream }})
     needs: lint
     runs-on: ubuntu-latest
-    timeout-minutes: 15
+    timeout-minutes: 60
     strategy:
       fail-fast: false
       matrix:
@@ -140,9 +141,10 @@ jobs:
         run: |
           micromamba list
           python -m pip check || true
+          pip install pytest-timeout
       - name: Test with conda
         run: |
-          python -m pytest --numprocesses=logical --durations=10 --cov=clisops --cov-report=term-missing
+          python -m pytest -m "not slow" --timeout=180 --durations=10 --cov=clisops --cov-report=term-missing
 
   finish:
     needs:

HISTORY.rst

@@ -1,6 +1,47 @@
 Version History
 ===============
 
+v0.14.1 (unreleased)
+--------------------
+
+New Features
+^^^^^^^^^^^^
+* Added new methods to `clisops.core.regrid.Grid`
+    * Added possibility to apply land or ocean mask if present in the file
+    * Adapted method from ESMF to detect smashed cells
+    * Masking degenerate (i.e. collapsed and smashed) cells
+    * Dropping lat/lon bounds if an integrity check fails
+    * Added a few attributes to the `Grid` object
+* `clisops.ops.regrid`: added option to request a land/sea mask for the output grid
+* `clisops.utils.dataset_utils`
+    * Added function `determine_lon_lat_range` to determine the min. and max. lat and lon values
+    * Added function `fix_unmasked_missing_values_lon_lat` to identify and mask yet unmasked missing values in lat and lon arrays
+    * Added `force` parameter to `cf_convert_between_lon_frames`
+
+Bug Fixes
+^^^^^^^^^
+* `clisops.utils.dataset_utils`
+    * Fixed issue in `cf_convert_between_lon_frames` causing the longitude frame to not be adjusted in case of NaNs in the longitude array
+    * Addressed issues in `generate_bounds_curvilinear`
+        * latitudes are now clipped above 90 or below -90 degrees north
+        * longitudes are converted to longitude frame -180, 180
+        * longitude bounds are adjusted at the Greenwich meridian or anti meridian to avoid grid cells wrapping once or more times around the globe
+        * Bounds are generated significantly faster due to making use of index slicing and `numpy.vectorize`
+
+Breaking Changes
+^^^^^^^^^^^^^^^^
+* Adapted functions from `roocs_utils.xarray_utils.xarray_utils` into `clisops.utils.dataset_utils`
+    * `get_coord_by_type` now returns the name of the coordinate variable and not the coordinate variable
+    * `get_coord_by_type` optionally returns a list with further matches for the coordinate variable
+    * `get_coord_by_type` does no longer raise an exception when more than one coordinate variable matches the requested type
+    * `get_coord_by_type` raises `ValueError` instead of `Exception` when the coordinate type is unknown
+    * `detect_coordinate` raises `KeyError` instead of `AttributeError` if no coordinate could be detected
+    * `detect_gridtype` raises `ValueError` for unsupported grid types rather than `InvalidParameterValue` and `Exception`
+* `clisops.core.regrid`
+    * `Grid.detect_coordinate`: raises `KeyError` instead of `AttributeError` if no coordinate could be detected
+* `clisops.ops.regrid`
+    * `Regrid._calculate`: issues `UserWarning` instead of letting `clisops.core.Weights.__init__` raise an `Exception` when input and output grid are alike
+
 v0.14.0 (2024-10-03)
 --------------------
 

binder/environment.yml

@@ -0,0 +1 @@
+../environment.yml

clisops/core/average.py

@@ -10,20 +10,20 @@
 import numpy as np
 import xarray as xr
 from roocs_utils.exceptions import InvalidParameterValue
-from roocs_utils.xarray_utils.xarray_utils import (
+
+from clisops.core.regrid import XESMF_MINIMUM_VERSION
+from clisops.core.subset import shape_bbox_indexer
+from clisops.utils.dataset_utils import (
     get_coord_by_type,
     get_coord_type,
     known_coord_types,
 )
-
-from clisops.core.regrid import XESMF_MINIMUM_VERSION
-from clisops.core.subset import shape_bbox_indexer
 from clisops.utils.time_utils import create_time_bounds
 
 __all__ = ["average_over_dims", "average_shape", "average_time"]
 
 # see https://pandas.pydata.org/pandas-docs/stable/user_guide/timeseries.html#dateoffset-objects
-freqs = {"day": "1D", "month": "1MS", "year": "1AS"}
+freqs = {"day": "1D", "month": "1MS", "year": "1YS"}
 
 
 def average_shape(
@@ -281,15 +281,11 @@ def average_time(
         )
 
     # check time coordinate exists and get name
-    # For roocs_utils 0.5.0
     t = get_coord_by_type(ds, "time", ignore_aux_coords=False)
     if t is None:
         raise Exception("Time dimension could not be found")
-    # For latest roocs_utils (master)
-    # try:
-    #    t = get_coord_by_type(ds, "time", ignore_aux_coords=False)
-    # except KeyError:
-    #    raise Exception("Time dimension could not be found")
+    else:
+        t = ds[t]
 
     # resample and average over time
     ds_t_avg = ds.resample(indexer={t.name: freqs[freq]}).mean(