def merge(date):
dslist = []
ds = cml.load_dataset("s2s-ai-competition",
date=date,
parameter="2t",
version=version)
dslist.append(ds.to_xarray())
ds = cml.load_dataset("s2s-ai-competition",
date=date,
parameter="tp",
version=version)
dslist.append(ds.to_xarray())
def check(ds):
print(dict(ds.dims), list(ds.keys()))
for ds in dslist:
check(ds)
ds = xr.merge(dslist)
print("-- Merged into --")
check(ds)
dslist[0].step.values, dslist[1].step.values
def test_era5_temperature():
if not os.path.exists(os.path.expanduser("~/.cdsapirc")):
pytest.skip("No ~/.cdsapirc")
cml.load_dataset("era5-temperature",
period=(1979, 1982),
domain="France",
time=12)
def test_grib_index_eumetnet_with_plugin():
ds = cml.load_dataset(
"eumetnet-postprocessing-benchmark-training-data-gridded-forecasts-efi",
date="2017-12-28",
parameter="2ti",
)
xds = ds.to_xarray()
print(xds)
def test_read():
ds = cml.load_dataset(
"{{ cookiecutter.plugin_name }}-{{ cookiecutter.dataset_name }}",
year="2021",
parameter="t2m",
)
xds = ds.to_xarray()
print(xds)
def test_datetime():
if not os.path.exists(os.path.expanduser("~/.cdsapirc")):
pytest.skip("No ~/.cdsapirc")
data = cml.load_dataset("era5-temperature",
domain="france",
period=(1980, ),
time=12)
data["1980-12-09 12:00"]
def test_datetime():
if not os.path.exists(os.path.expanduser("~/.cdsapirc")):
pytest.skip("No ~/.cdsapirc")
data = cml.load_dataset(
"era5-temperature", domain="france", period=(1980,), time=12
)
data["1980-12-09 12:00"]
with pytest.raises(NotImplementedError):
data.sel(date="1980-12-09 12:00")
def __init__(self, dt="time", basins=["atlantic", "pacific"]):
self._dt = dt
self._df = pd.DataFrame()
for basin in basins:
if self._df.empty:
self._df = load_dataset("hurricane-database",
basin).to_pandas()
else:
self._df = pd.concat(
[
self._df,
load_dataset("hurricane-database", basin).to_pandas()
],
ignore_index=True,
)
assert all(
[c in self._df.columns for c in [dt, "lat", "lon", "pressure"]])
self._df.sort_values(by=dt)
def merge_multiple_dates(dates):
dslist = []
for date in dates:
ds = cml.load_dataset("s2s-ai-competition",
date=date,
parameter="2t",
version=version)
dslist.append(ds.to_xarray())
def check(ds):
print(dict(ds.dims), list(ds.keys()))
for ds in dslist:
check(ds)
ds = xr.merge(dslist)
print("-- Merged into --")
check(ds)
def test_tfdataset_2():
from tensorflow.keras.layers import Dense, Flatten, Input
from tensorflow.keras.models import Sequential
ds = cml.load_dataset("high-low")
train, test = ds.to_tfdataset(split=["train", "test"])
shape = train.element_spec[0].shape
model = Sequential()
model.add(Input(shape=(shape[-2], shape[-1])))
model.add(Flatten())
model.add(Dense(64, activation="sigmoid"))
model.add(Dense(4, activation="softmax"))
model.compile(
optimizer="adam",
loss="categorical_crossentropy",
metrics=["accuracy"],
)
print(model.summary())
model.fit(train, epochs=1, verbose=0)
model.evaluate(test)
def test_dataset_1():
load_dataset("sample-bufr-data")
def test_register_dataset():
register_dataset("test-register-dataset", RegisteredDataset)
ds = cml.load_dataset("test-register-dataset", shape=(3, 3))
assert ds.to_numpy().shape == (3, 3)
def wrapped(*args, **kwargs):
return climetlab.load_dataset(name.replace("_", "-"), *args, **kwargs)
import climetlab as cml
# cml.settings.set("plotting-options", {'dump_yaml': True})
dataset = cml.load_dataset("example-dataset")
data = dataset[0]
cml.plot_map(data, foreground=False)
cml.plot_map(data, foreground=True)
cml.plot_map(data, foreground="example-foreground")
cml.plot_map(
data,
foreground=dict(
map_grid=False,
map_label=False,
map_grid_frame=True,
map_grid_frame_thickness=5,
map_boundaries=True,
),
)
# Partial update of the current `foreground`
# How to do is still to be decided
# Option 1
cml.plot_map(
data,
foreground={
def test_read_rt_2dates():
ds = cml.load_dataset("s2s-ai-competition", date=["20200102", "20200102"])
xds = ds.to_xarray()
print(xds)
sst = xds.sel()
def test_read_rt_tp_and_2t():
ds = cml.load_dataset("s2s-ai-competition", date="20200102", parameter=["tp", "2t"])
xds = ds.to_xarray()
print(xds)
def test_samples():
cml.load_dataset("meteonet-samples-radar")
def test_remote_dataset_from_climetlab_catalog():
load_dataset("sample-netcdf-data")
def test_unknown_dataset():
with pytest.raises(NameError):
load_dataset("do-not-exist-lkj45a45qsdf3")
def test_pandas_filter():
data = cml.load_dataset("hurricane-database", bassin="atlantic")
irma = data.to_pandas(name="irma", year=2017)
assert len(irma) == 66
#!/usr/bin/env python
# coding: utf-8
# In[73]:
import climetlab as cml
ds = cml.load_dataset("era5-temperature", period=(1979, 1982), domain="France", time=12)
def test_numpy_1():
ds = cml.load_dataset("weather-bench")
z500 = ds.to_xarray()
z = z500.sel({"time": "1979-01-01"}).z.values
cml.plot_map(z[0], metadata=z500.z)
def test_read_hc():
ds = cml.load_dataset("s2s-ai-competition", date="20200102", hindcast=True)
xds = ds.to_xarray()
print(xds)
def test_read_zarr():
ds = cml.load_dataset("s2s-ai-competition", version="0.1.7", format="zarr")
xds = ds.to_xarray()
print(xds)
def test_load_dataset_meteonet_sample_masks():
cml.load_dataset("meteonet-samples-masks", domain="SE")
import climetlab as cml
data = cml.load_dataset("hurricane-database", "atlantic")
print(data.home_page)
irma = data.to_pandas(name="irma", year=2017)
cml.plot_map(irma)
