def test_subset_with_lat_lon_single_values(load_esgf_test_data):
"""Creates subset where lat and lon only have one value. Then
subsets that. This tests that the `lat_bnds` and `lon_bnds`
are not being reversed by the `_check_desc_coords` function in
`clisops.core.subset`.
"""
area = (20.0, 10.0, 250.0, 90.0)
outputs = subset(
ds=CMIP6_RLDS,
time=("2001-01-01T00:00:00", "2002-12-30T00:00:00"),
area=area,
output_type="xarray",
)
ds = outputs[0]
outputs2 = subset(
ds=ds,
time=("2001-01-01T00:00:00", "2002-12-30T00:00:00"),
area=area,
output_type="xarray",
)
ds2 = outputs2[0]
assert len(ds2.lat) == 1
assert len(ds2.lon) == 1
def test_subset_invalid_area(tmpdir):
""" Tests subset with invalid area param."""
with pytest.raises(InvalidParameterValue):
subset(
ds=CMIP5_TAS_FILE,
area=("zero", 49.0, 10.0, 65.0),
output_dir=tmpdir,
)
def test_subset_no_ds(tmpdir):
""" Tests subset with no dataset provided."""
with pytest.raises(TypeError):
subset(
time=("2020-01-01T00:00:00", "2020-12-30T00:00:00"),
area=(0, -90.0, 360.0, 90.0),
output_dir=tmpdir,
)
def test_subset_invalid_area(cmip5_tas_file, tmpdir):
"""Tests subset with invalid area param."""
with pytest.raises(InvalidParameterValue):
subset(
ds=cmip5_tas_file,
area=("zero", 49.0, 10.0, 65.0),
output_dir=tmpdir,
)
def test_subset_invalid_time(tmpdir):
""" Tests subset with invalid time param."""
with pytest.raises(InvalidParameterValue):
subset(
dset=CMIP5_TAS_FILE,
time=("yesterday", "2020-12-30T00:00:00"),
output_dir=tmpdir,
)
def test_subset_ds_is_none(tmpdir):
""" Tests subset with ds=None."""
with pytest.raises(MissingParameterValue):
subset(
ds=None,
time=("2020-01-01T00:00:00", "2020-12-30T00:00:00"),
area=(0, -90.0, 360.0, 90.0),
output_dir=tmpdir,
)
def test_time_invariant_subset_with_time(load_esgf_test_data):
with pytest.raises(AttributeError) as exc:
subset(
ds=CMIP6_MRSOFC,
time=("2005-01-01T00:00:00", "2020-12-30T00:00:00"),
area=(5.0, 10.0, 360.0, 90.0),
output_type="xarray",
)
assert str(exc.value) == "'Dataset' object has no attribute 'time'"
def test_subset_invalid_time(tmpdir):
""" Tests subset with invalid time param."""
with pytest.raises(InvalidParameterValue):
subset(
ds=CMIP5_TAS_FILE,
time=("yesterday", "2020-12-30T00:00:00"),
area=(0, -90.0, 360.0, 90.0),
output_dir=tmpdir,
output_type="nc",
file_namer="simple",
)
def test_0_360_cross(tmpdir):
ds = _load_ds(
"/badc/cmip6/data/CMIP6/CMIP/IPSL/IPSL-CM6A-LR/historical/r1i1p1f1/Amon/rlds/gr/v20180803/"
"rlds_Amon_IPSL-CM6A-LR_historical_r1i1p1f1_gr_185001-201412.nc")
with pytest.raises(NotImplementedError):
subset(
ds=ds,
area=(-50.0, -90.0, 100.0, 90.0),
output_dir=tmpdir,
output_type="nc",
file_namer="simple",
)
def test_curvilinear_ds_no_data_in_bbox():
with pytest.raises(ValueError) as exc:
subset(
ds=CMIP6_TOS_ONE_TIME_STEP,
area="1,5,1.2,4",
time="2021-01-01/2050-12-31",
output_type="xarray",
)
assert (
str(exc.value) ==
"There were no valid data points found in the requested subset. Please expand the area covered by the bounding box."
)
def test_reverse_level(self, cmip6_o3):
result = subset(
ds=cmip6_o3,
level="100000/100",
output_type="xarray",
)
result_rev = subset(
ds=cmip6_o3,
level="100/100000",
output_type="xarray",
)
np.testing.assert_array_equal(result[0].o3, result_rev[0].o3)
def test_reverse_lat_and_lon_curvilinear(self, load_esgf_test_data):
result = subset(
ds=CMIP6_TOS_ONE_TIME_STEP,
area=(20, -45, 240, 45),
output_type="xarray",
)
result_rev = subset(
ds=CMIP6_TOS_ONE_TIME_STEP,
area=(20, 45, 240, -45),
output_type="xarray",
)
np.testing.assert_array_equal(result[0].tos, result_rev[0].tos)
def test_reverse_lat_and_lon_regular(self, load_esgf_test_data):
result = subset(
ds=CMIP6_RLDS_ONE_TIME_STEP,
area=(-70, -45, 240, 45),
output_type="xarray",
)
result_rev = subset(
ds=CMIP6_RLDS_ONE_TIME_STEP,
area=(240, 45, -70, -45),
output_type="xarray",
)
np.testing.assert_array_equal(result[0].rlds, result_rev[0].rlds)
def test_no_lon_in_range():
with pytest.raises(Exception) as exc:
subset(
ds=CMIP6_RLDS_ONE_TIME_STEP,
area=(8.37, -90, 8.56, 90),
time=("2006-01-01T00:00:00", "2099-12-30T00:00:00"),
output_type="xarray",
)
assert (
str(exc.value) ==
"There were no valid data points found in the requested subset. Please expand "
"the area covered by the bounding box, the time period or the level range you have selected."
)
def test_curvilinear_ds_no_data_in_bbox_real_data_swap_lat():
ds = _load_ds(
"/badc/cmip6/data/CMIP6/ScenarioMIP/CNRM-CERFACS/CNRM-CM6-1/ssp245/r1i1p1f2/Omon/tos/gn/v20190219/tos_Omon_CNRM-CM6-1_ssp245_r1i1p1f2_gn_201501-210012.nc"
)
with pytest.raises(ValueError) as exc:
subset(
ds=ds,
area="1,4,2,40",
time="2021-01-01/2050-12-31",
output_type="xarray",
)
assert (
str(exc.value) ==
"There were no valid data points found in the requested subset. Please expand the area covered by the bounding box."
)
def test_general_subset_dset(tmpdir):
""" Tests clisops subset function with only a dataset"""
result = subset(
dset=TAS_NC,
output_dir=tmpdir,
)
assert "output.nc" in result
def test_roll_convert_lon_coords(load_esgf_test_data):
ds, lon = setup_test()
# work out how much to roll by
offset = calculate_offset(180)
ds_roll = ds.roll(shifts={f"{lon.name}": offset}, roll_coords=False)
# check roll with roll_coords=False actually does something
np.testing.assert_raises(
AssertionError,
np.testing.assert_array_equal,
ds_roll.rlds.values,
ds.rlds.values,
)
ds_roll.coords[lon.name] = ds_roll.coords[lon.name] - 180
assert isclose(ds_roll.lon.values.min(), -180, abs_tol=10**2)
assert isclose(ds_roll.lon.values.max(), 180, abs_tol=10**2)
result = subset(
ds=ds_roll,
area=(-50.0, -90.0, 100.0, 90.0),
output_type="xarray",
)
assert result
def test_subset_with_time_and_area(cmip5_tas_file, tmpdir):
"""Tests clisops subset function with time and area subsets.
On completion:
- assert all dimensions have been reduced.
"""
start_time, end_time = ("2019-01-16", "2020-12-16")
bbox = (0.0, -80, 170.0, 65.0)
outputs = subset(
ds=cmip5_tas_file,
time=(start_time, end_time),
area=bbox,
output_dir=tmpdir,
output_type="xarray",
)
ds = outputs[0]
assert _format_time(ds.time.values.min()) == start_time
assert _format_time(ds.time.values.max()) == end_time
assert ds.lon.values.tolist() == [0]
assert ds.lat.values.tolist() == [35]
def test_cross_prime_meridian(tmpdir):
ds = _load_ds(
"/badc/cmip6/data/CMIP6/ScenarioMIP/MIROC/MIROC6/ssp119/r1i1p1f1/day/tas/gn/v20191016"
"/tas_day_MIROC6_ssp119_r1i1p1f1_gn_20150101-20241231.nc")
with pytest.raises(NotImplementedError) as exc:
subset(
ds=ds,
area=(-5, 50, 30, 65),
output_dir=tmpdir,
output_type="nc",
file_namer="simple",
)
assert (str(exc.value) ==
"Input longitude bounds ([-5. 30.]) cross the 0 degree meridian "
"but dataset longitudes are all positive.")
def test_time_slices_in_subset_rh():
start_time, end_time = "2001-01-01T00:00:00", "2200-12-30T00:00:00"
time_slices = [
("2001-01-16", "2002-09-16"),
("2002-10-16", "2004-06-16"),
("2004-07-16", "2005-11-16"),
]
config_max_file_size = CONFIG["clisops:write"]["file_size_limit"]
temp_max_file_size = "10KB"
CONFIG["clisops:write"]["file_size_limit"] = temp_max_file_size
outputs = subset(
ds=CMIP5_RH,
time=(start_time, end_time),
area=(0.0, 5.0, 50.0, 90.0),
output_type="xarray",
file_namer="simple",
)
CONFIG["clisops:write"]["file_size_limit"] = config_max_file_size
assert _format_time(outputs[0].time.values.min()) >= start_time
assert _format_time(outputs[-1].time.values.max()) <= end_time
count = 0
for _ in outputs:
assert _format_time(
outputs[count].time.values.min()) >= time_slices[count][0]
assert _format_time(
outputs[count].time.values.max()) >= time_slices[count][1]
count += 1
def test_subset_level(tmpdir):
""" Tests clisops subset function with a level subset."""
# Levels are: 100000, ..., 100
ds = _load_ds(CMIP6_O3)
result1 = subset(ds=CMIP6_O3, level="100000/100", output_type="xarray")
np.testing.assert_array_equal(result1[0].o3.values, ds.o3.values)
result2 = subset(ds=CMIP6_O3, level="100/100", output_type="xarray")
np.testing.assert_array_equal(result2[0].o3.shape, (1200, 1, 2, 3))
result3 = subset(ds=CMIP6_O3, level="101/-23.234", output_type="xarray")
np.testing.assert_array_equal(result3[0].o3.values, result2[0].o3.values)
def test_xarray_roll_lon(tmpdir, load_esgf_test_data):
ds, lon = setup_test()
# work out how much to roll by
offset = calculate_offset(180)
ds_roll = ds.roll(shifts={f"{lon.name}": offset}, roll_coords=True)
# testing after rolling still raises an error
with pytest.raises(NotImplementedError):
subset(
ds=ds_roll,
area=(-50.0, -90.0, 100.0, 90.0),
output_dir=tmpdir,
output_type="nc",
file_namer="simple",
)
def test_general_subset_space(tmpdir):
""" Tests clisops subset function with a space subset."""
result = subset(
dset=TAS_NC,
space=(0.0, 49.0, 10.0, 65.0),
output_dir=tmpdir,
)
assert "output.nc" in result
def test_general_subset_time(tmpdir):
""" Tests clisops subset function with a time subset."""
result = subset(
dset=TAS_NC,
time=("2020-01-01T00:00:00", "2020-12-30T00:00:00"),
output_dir=tmpdir,
)
assert "output.nc" in result
def test_subset_with_multiple_files_rh(tmpdir):
""" Tests with multiple rh files"""
result = subset(
dset=CMIP5_RH,
time=("2020-01-01T00:00:00", "2020-12-30T00:00:00"),
output_dir=tmpdir,
)
assert "output.nc" in result
def test_subset_with_multiple_files_zostoga(tmpdir):
""" Tests with multiple tas files"""
result = subset(
dset=CMIP5_ZOSTOGA,
time=("2020-01-01T00:00:00", "2020-12-30T00:00:00"),
output_dir=tmpdir,
)
assert "output.nc" in result
def test_subset_with_tas_series(tmpdir, tas_series):
""" Test with tas_series fixture"""
result = subset(
dset=tas_series(["20", "22", "25"]),
time=("2020-01-01T00:00:00", "2020-12-30T00:00:00"),
output_dir=tmpdir,
)
assert "output.nc" in result
def test_general_subset_all(tmpdir):
""" Tests clisops subset function with time, space, level subsets."""
result = subset(
dset=TAS_NC,
time=("2020-01-01T00:00:00", "2020-12-30T00:00:00"),
space=(0.0, 49.0, 10.0, 65.0),
output_dir=tmpdir,
)
assert "output.nc" in result
def test_subset_with_multiple_files_tas(tmpdir):
""" Tests with multiple tas files"""
result = subset(
dset=CMIP5_TAS,
time=("2020-01-01T00:00:00", "2020-12-30T00:00:00"),
space=(0.0, 49.0, 10.0, 65.0),
output_dir=tmpdir,
)
assert "output.nc" in result
def test_subset_with_time_and_space(tmpdir):
""" Tests clisops subset function with time, space, level subsets."""
result = subset(
dset=CMIP5_TAS_FILE,
time=("2020-01-01T00:00:00", "2020-12-30T00:00:00"),
space=(0.0, 49.0, 10.0, 65.0),
output_dir=tmpdir,
)
assert "output.nc" in result
