def test_masking_adapter():
for col in (None, "x"):
ds = TestDataset("", n=20)
ds_mask = MaskingAdapter(ds, "<", 10, col)
data_masked = ds_mask.read_ts()
data_masked2 = ds_mask.read()
nptest.assert_almost_equal(
data_masked["x"].values,
np.concatenate(
[np.ones((10), dtype=bool),
np.zeros((10), dtype=bool)]),
)
nptest.assert_almost_equal(
data_masked2["x"].values,
np.concatenate(
[np.ones((10), dtype=bool),
np.zeros((10), dtype=bool)]),
)
if col is None:
nptest.assert_almost_equal(data_masked["y"].values,
np.ones((20), dtype=bool))
nptest.assert_almost_equal(data_masked2["y"].values,
np.ones((20), dtype=bool))
def test_adapters_with_ascat():
ascat_data_folder = os.path.join(
os.path.dirname(__file__),
"..",
"test-data",
"sat",
"ascat",
"netcdf",
"55R22",
)
ascat_grid_folder = os.path.join(
os.path.dirname(__file__),
"..",
"test-data",
"sat",
"ascat",
"netcdf",
"grid",
)
grid_fname = os.path.join(ascat_grid_folder, "TUW_WARP5_grid_info_2_1.nc")
ascat_reader = AscatGriddedNcTs(
ascat_data_folder,
"TUW_METOP_ASCAT_WARP55R22_{:04d}",
grid_filename=grid_fname,
)
ascat_anom = AnomalyAdapter(ascat_reader, window_size=35, columns=["sm"])
data = ascat_anom.read_ts(12.891455, 45.923004)
assert data is not None
assert np.any(data["sm"].values != 0)
data = ascat_anom.read(12.891455, 45.923004)
assert data is not None
assert np.any(data["sm"].values != 0)
ascat_self = SelfMaskingAdapter(ascat_reader, ">", 0, "sm")
data2 = ascat_self.read_ts(12.891455, 45.923004)
assert data2 is not None
assert np.all(data2["sm"].values > 0)
data2 = ascat_self.read(12.891455, 45.923004)
assert data2 is not None
assert np.all(data2["sm"].values > 0)
ascat_mask = MaskingAdapter(ascat_reader, ">", 0, "sm")
data3 = ascat_mask.read_ts(12.891455, 45.923004)
assert data3 is not None
assert np.any(data3["sm"].values)
data3 = ascat_mask.read(12.891455, 45.923004)
assert data3 is not None
assert np.any(data3["sm"].values)
ascat_clim = AnomalyClimAdapter(ascat_reader, columns=["sm"])
data4 = ascat_clim.read_ts(12.891455, 45.923004)
assert data4 is not None
assert np.any(data["sm"].values != 0)
data4 = ascat_clim.read(12.891455, 45.923004)
assert data4 is not None
assert np.any(data["sm"].values != 0)
def test_masking_adapter():
ds = TestDataset('', n=20)
ds_mask = MaskingAdapter(ds, '<', 10)
data_masked = ds_mask.read_ts()
nptest.assert_almost_equal(data_masked['x'].values,
np.concatenate([np.ones((10), dtype=bool),
np.zeros((10), dtype=bool)]))
nptest.assert_almost_equal(
data_masked['y'].values, np.ones((20), dtype=bool))
def test_masking_adapter():
ds = TestDataset('', n=20)
ds_mask = MaskingAdapter(ds, '<', 10)
data_masked = ds_mask.read_ts()
nptest.assert_almost_equal(
data_masked['x'].values,
np.concatenate([np.ones((10), dtype=bool),
np.zeros((10), dtype=bool)]))
nptest.assert_almost_equal(data_masked['y'].values,
np.ones((20), dtype=bool))
def test_timezone_removal():
tz_reader = TestTimezoneReader()
reader_anom = AnomalyAdapter(tz_reader, window_size=35, columns=["data"])
assert reader_anom.read_ts(0) is not None
reader_self = SelfMaskingAdapter(tz_reader, ">", 0, "data")
assert reader_self.read_ts(0) is not None
reader_mask = MaskingAdapter(tz_reader, ">", 0, "data")
assert reader_mask.read_ts(0) is not None
reader_clim = AnomalyClimAdapter(tz_reader, columns=["data"])
assert reader_clim.read_ts(0) is not None
def test_adapters_with_ascat():
ascat_data_folder = os.path.join(os.path.dirname(__file__), '..',
'test-data', 'sat', 'ascat', 'netcdf',
'55R22')
ascat_grid_folder = os.path.join(os.path.dirname(__file__), '..',
'test-data', 'sat', 'ascat', 'netcdf',
'grid')
ascat_reader = AscatSsmCdr(ascat_data_folder,
ascat_grid_folder,
grid_filename='TUW_WARP5_grid_info_2_1.nc')
ascat_anom = AnomalyAdapter(ascat_reader, window_size=35, columns=['sm'])
data = ascat_anom.read_ts(12.891455, 45.923004)
assert data is not None
assert np.any(data['sm'].values != 0)
data = ascat_anom.read(12.891455, 45.923004)
assert data is not None
assert np.any(data['sm'].values != 0)
ascat_self = SelfMaskingAdapter(ascat_reader, '>', 0, 'sm')
data2 = ascat_self.read_ts(12.891455, 45.923004)
assert data2 is not None
assert np.all(data2['sm'].values > 0)
data2 = ascat_self.read(12.891455, 45.923004)
assert data2 is not None
assert np.all(data2['sm'].values > 0)
ascat_mask = MaskingAdapter(ascat_reader, '>', 0, 'sm')
data3 = ascat_mask.read_ts(12.891455, 45.923004)
assert data3 is not None
assert np.any(data3['sm'].values)
data3 = ascat_mask.read(12.891455, 45.923004)
assert data3 is not None
assert np.any(data3['sm'].values)
ascat_clim = AnomalyClimAdapter(ascat_reader, columns=['sm'])
data4 = ascat_clim.read_ts(12.891455, 45.923004)
assert data4 is not None
assert np.any(data['sm'].values != 0)
data4 = ascat_clim.read(12.891455, 45.923004)
assert data4 is not None
assert np.any(data['sm'].values != 0)
# # Masking datasets are datasets that return a pandas DataFrame with boolean values. `True` means that the observation # should be masked, `False` means it should be kept. All masking datasets are temporally matched in pairs to the # temporal reference dataset. Only observations for which all masking datasets have a value of `False` are kept for # further validation. # # The masking datasets have the same format as the dataset dictionary and can be specified in the Validation class # with the `masking_datasets` keyword. # # ### Masking adapter # # To easily transform an existing dataset into a masking dataset `pytesmo` offers a adapter class that calls the # `read_ts` method of an existing dataset and creates a masking dataset based on an operator, a given threshold, and (optionally) a column name. # In[12]: from pytesmo.validation_framework.adapters import MaskingAdapter ds_mask = MaskingAdapter(ismn_reader, '<', 0.2, 'soil moisture') print(ds_mask.read_ts(ids[0]).head()) # ### Self-masking adapter # `pytesmo` also has a class that masks a dataset "on-the-fly", based on one of the columns it contains and an operator and a threshold. In contrast to the masking adapter mentioned above, the output of the self-masking adapter is the masked data, not the the mask. The self-masking adapter wraps a data reader, which must have a `read_ts` or `read` method. Calling its `read_ts`/`read` method will return the masked data - more precisely a DataFrame with only rows where the masking condition is true. # In[13]: from pytesmo.validation_framework.adapters import SelfMaskingAdapter ds_mask = SelfMaskingAdapter(ismn_reader, '<', 0.2, 'soil moisture') print(ds_mask.read_ts(ids[0]).head())
# ```python # from pytesmo.validation_framework import start_validation # # # Note that before starting the validation you must start a controller # # and engines, for example by using: ipcluster start -n 4 # # This command will launch a controller and 4 engines on the local machine. # # Also, do not forget to change the setup_code path to your current setup. # # setup_code = "my_validation.py" # start_validation(setup_code) # ``` # ## Masking datasets # # Masking datasets are datasets that return a pandas DataFrame with boolean values. `True` means that the observation should be masked, `False` means it should be kept. All masking datasets are temporally matched in pairs to the temporal reference dataset. Only observations for which all masking datasets have a value of `False` are kept for further validation. # # The masking datasets have the same format as the dataset dictionary and can be specified in the Validation class with the `masking_datasets` keyword. # # ### Masking adapter # # To easily transform an existing dataset into a masking dataset `pytesmo` offers a adapter class that calls the `read_ts` method of an existing dataset and performs the masking based on an operator and a given threshold. # In[12]: from pytesmo.validation_framework.adapters import MaskingAdapter ds_mask = MaskingAdapter(ismn_reader, '<', 0.2) print ds_mask.read_ts(ids[0])['soil moisture'].head()
# further validation. # # The masking datasets have the same format as the dataset dictionary and can be specified in the Validation class # with the `masking_datasets` keyword. # # ### Masking adapter # # To easily transform an existing dataset into a masking dataset `pytesmo` offers a adapter class that calls the # `read_ts` method of an existing dataset and creates a masking dataset based on an operator, a given threshold, and (optionally) a column name. # In[12]: from pytesmo.validation_framework.adapters import MaskingAdapter ds_mask = MaskingAdapter(ismn_reader, '<', 0.2, 'soil moisture') print(ds_mask.read_ts(ids[0]).head()) # ### Self-masking adapter # `pytesmo` also has a class that masks a dataset "on-the-fly", based on one of the columns it contains and an operator and a threshold. In contrast to the masking adapter mentioned above, the output of the self-masking adapter is the masked data, not the the mask. The self-masking adapter wraps a data reader, which must have a `read_ts` or `read` method. Calling its `read_ts`/`read` method will return the masked data - more precisely a DataFrame with only rows where the masking condition is true. # In[13]: from pytesmo.validation_framework.adapters import SelfMaskingAdapter ds_mask = SelfMaskingAdapter(ismn_reader, '<', 0.2, 'soil moisture') print(ds_mask.read_ts(ids[0]).head())
