Merge pull request #39 from dgergel/fix_dimnaming_regridding_service

update regridding output dimensions

Author: Diana Gergel

dodola/cli.py

@@ -140,16 +140,14 @@ def rechunk(x, variable, chunk, maxmemory, out):
     required=True,
     help="Regridding method - 'bilinear' or 'conservative'",
 )
-@click.option(
-    "--targetresolution", "-r", default=1.0, help="Global-grid resolution to regrid to"
-)
+@click.option("--domain-file", "-d", help="Domain file to regrid to")
 @click.option(
     "--weightspath",
     "-w",
     default=None,
     help="Local path to existing regrid weights file",
 )
-def regrid(x, out, method, targetresolution, weightspath):
+def regrid(x, out, method, domain_file, weightspath):
     """Regrid a target climate dataset
 
     Note, the weightspath only accepts paths to NetCDF files on the local disk. See
@@ -162,6 +160,6 @@ def regrid(x, out, method, targetresolution, weightspath):
         out=str(out),
         method=str(method),
         storage=_authenticate_storage(),
+        domain_file=domain_file,
         weights_path=weightspath,
-        target_resolution=float(targetresolution),
     )

dodola/core.py

@@ -92,26 +92,27 @@ def build_xesmf_weights_file(x, method, target_resolution, filename=None):
     return str(out.filename)
 
 
-def xesmf_regrid(x, method, target_resolution, weights_path=None):
+def xesmf_regrid(x, domain, method, weights_path=None):
     """
 
     Parameters
     ----------
     x : xr.Dataset
+    domain : xr.Dataset
+        Domain to regrid to.
     method : str
         Method of regridding. Passed to ``xesmf.Regridder``.
-    target_resolution : float
-        Decimal-degree resolution of global latxlon grid to regrid to.
     weights_path : str, optional
         Local path to netCDF file of pre-calculated XESMF regridding weights.
 
     Returns
     -------
     xr.Dataset
     """
+
     regridder = xe.Regridder(
         x,
-        xe.util.grid_global(target_resolution, target_resolution),
+        domain,
         method=method,
         filename=weights_path,
     )

dodola/services.py

@@ -126,7 +126,7 @@ def rechunk(x, target_chunks, out, max_mem, storage):
 
 
 @log_service
-def regrid(x, out, method, storage, weights_path=None, target_resolution=1.0):
+def regrid(x, out, method, storage, domain_file, weights_path=None):
     """Regrid climate data
 
     Parameters
@@ -139,18 +139,22 @@ def regrid(x, out, method, storage, weights_path=None, target_resolution=1.0):
         Method of regridding. Passed to ``xesmf.Regridder``.
     storage : dodola.repository._ZarrRepo
         Storage abstraction for data IO.
+    domain_file : str
+        Storage URL to input xr.Dataset domain file to regrid to.
     weights_path : optional
         Local file path name to write regridding weights file to.
-    target_resolution : float, optional
-        Decimal-degree resolution of global grid to regrid to.
     """
     ds = storage.read(x)
+
+    ds_domain = storage.read(domain_file)
+
     regridded_ds = xesmf_regrid(
         ds,
+        ds_domain,
         method=method,
-        target_resolution=target_resolution,
         weights_path=weights_path,
     )
+
     storage.write(out, regridded_ds)
 
 

dodola/tests/test_services.py

@@ -32,6 +32,18 @@ def _datafactory(x, start_time="1950-01-01"):
     return out
 
 
+@pytest.fixture
+def domain_file(request):
+    """ Creates a fake domain Dataset for testing"""
+    lon_name = "lon"
+    lat_name = "lat"
+    domain = grid_global(request.param, request.param)
+    domain[lat_name] = np.unique(domain[lat_name].values)
+    domain[lon_name] = np.unique(domain[lon_name].values)
+
+    return domain
+
+
 @pytest.mark.parametrize(
     "method, expected_head, expected_tail",
     [
@@ -163,21 +175,24 @@ def test_rechunk():
 
 
 @pytest.mark.parametrize(
-    "regrid_method, expected_shape",
+    "domain_file, regrid_method, expected_shape",
     [
         pytest.param(
+            1.0,
             "bilinear",
             (180, 360),
             id="Bilinear regrid",
         ),
         pytest.param(
+            1.0,
             "conservative",
             (180, 360),
             id="Conservative regrid",
         ),
     ],
+    indirect=["domain_file"],
 )
-def test_regrid_methods(regrid_method, expected_shape):
+def test_regrid_methods(domain_file, regrid_method, expected_shape):
     """Smoke test that services.regrid outputs with different regrid methods
 
     The expected shape is the same, but change in methods should not error.
@@ -189,6 +204,7 @@ def test_regrid_methods(regrid_method, expected_shape):
     fakestorage = memory_repository(
         {
             "an/input/path.zarr": ds_in,
+            "a/domainfile/path.zarr": domain_file,
         }
     )
 
@@ -197,28 +213,30 @@ def test_regrid_methods(regrid_method, expected_shape):
         out="an/output/path.zarr",
         method=regrid_method,
         storage=fakestorage,
+        domain_file="a/domainfile/path.zarr",
     )
     actual_shape = fakestorage.read("an/output/path.zarr")["fakevariable"].shape
     assert actual_shape == expected_shape
 
 
 @pytest.mark.parametrize(
-    "target_resolution, expected_shape",
+    "domain_file, expected_shape",
     [
         pytest.param(
             1.0,
             (180, 360),
-            id="Regrid to global 1.0° x 1.0° grid",
+            id="Regrid to domain file grid",
         ),
         pytest.param(
             2.0,
             (90, 180),
             id="Regrid to global 2.0° x 2.0° grid",
         ),
     ],
+    indirect=["domain_file"],
 )
-def test_regrid_resolution(target_resolution, expected_shape):
-    """Smoke test that services.regrid outputs with different regrid methods
+def test_regrid_resolution(domain_file, expected_shape):
+    """Smoke test that services.regrid outputs with different grid resolutions
 
     The expected shape is the same, but change in methods should not error.
     """
@@ -229,21 +247,25 @@ def test_regrid_resolution(target_resolution, expected_shape):
     fakestorage = memory_repository(
         {
             "an/input/path.zarr": ds_in,
+            "a/domainfile/path.zarr": domain_file,
         }
     )
 
     regrid(
         "an/input/path.zarr",
         out="an/output/path.zarr",
-        target_resolution=target_resolution,
         method="bilinear",
         storage=fakestorage,
+        domain_file="a/domainfile/path.zarr",
     )
     actual_shape = fakestorage.read("an/output/path.zarr")["fakevariable"].shape
     assert actual_shape == expected_shape
 
 
-def test_regrid_weights_integration(tmpdir):
+@pytest.mark.parametrize(
+    "domain_file", [pytest.param(1.0, id="Regrid to domain file grid")], indirect=True
+)
+def test_regrid_weights_integration(domain_file, tmpdir):
     """Test basic integration between service.regrid and service.build_weights"""
     expected_shape = (180, 360)
     # Output to tmp dir so we cleanup & don't clobber existing files...
@@ -256,6 +278,7 @@ def test_regrid_weights_integration(tmpdir):
     fakestorage = memory_repository(
         {
             "an/input/path.zarr": ds_in,
+            "a/domainfile/path.zarr": domain_file,
         }
     )
 
@@ -272,6 +295,7 @@ def test_regrid_weights_integration(tmpdir):
         method="bilinear",
         weights_path=weightsfile,
         storage=fakestorage,
+        domain_file="a/domainfile/path.zarr",
     )
     actual_shape = fakestorage.read("an/output/path.zarr")["fakevariable"].shape
     assert actual_shape == expected_shape