pycontrails 0.53.0__cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl

pycontrails/datalib/ecmwf/__init__.py

@@ -0,0 +1,53 @@
+"""ECMWF Data Access."""
+
+from __future__ import annotations
+
+from pycontrails.datalib.ecmwf.arco_era5 import ARCOERA5
+from pycontrails.datalib.ecmwf.era5 import ERA5
+from pycontrails.datalib.ecmwf.era5_model_level import ERA5ModelLevel
+from pycontrails.datalib.ecmwf.hres import HRES
+from pycontrails.datalib.ecmwf.hres_model_level import HRESModelLevel
+from pycontrails.datalib.ecmwf.ifs import IFS
+from pycontrails.datalib.ecmwf.variables import (
+    ECMWF_VARIABLES,
+    MODEL_LEVEL_VARIABLES,
+    PRESSURE_LEVEL_VARIABLES,
+    SURFACE_VARIABLES,
+    CloudAreaFraction,
+    CloudAreaFractionInLayer,
+    Divergence,
+    PotentialVorticity,
+    RelativeHumidity,
+    RelativeVorticity,
+    SpecificCloudIceWaterContent,
+    SpecificCloudLiquidWaterContent,
+    SurfaceSolarDownwardRadiation,
+    TOAIncidentSolarRadiation,
+    TopNetSolarRadiation,
+    TopNetThermalRadiation,
+)
+
+__all__ = [
+    "ARCOERA5",
+    "ERA5",
+    "ERA5ModelLevel",
+    "HRES",
+    "HRESModelLevel",
+    "IFS",
+    "CloudAreaFraction",
+    "CloudAreaFractionInLayer",
+    "Divergence",
+    "PotentialVorticity",
+    "RelativeHumidity",
+    "RelativeVorticity",
+    "SpecificCloudIceWaterContent",
+    "SpecificCloudLiquidWaterContent",
+    "SurfaceSolarDownwardRadiation",
+    "TOAIncidentSolarRadiation",
+    "TopNetSolarRadiation",
+    "TopNetThermalRadiation",
+    "ECMWF_VARIABLES",
+    "PRESSURE_LEVEL_VARIABLES",
+    "SURFACE_VARIABLES",
+    "MODEL_LEVEL_VARIABLES",
+]

pycontrails/datalib/ecmwf/arco_era5.py

@@ -0,0 +1,527 @@
+"""Support for `ARCO ERA5 <https://cloud.google.com/storage/docs/public-datasets/era5>`_.
+
+This module supports:
+
+- Downloading ARCO ERA5 model level data for specific times and pressure level variables.
+- Downloading ARCO ERA5 single level data for specific times and single level variables.
+- Interpolating model level data to a target lat-lon grid and pressure levels.
+- Local caching of the downloaded and interpolated data as netCDF files.
+- Opening cached data as a :class:`pycontrails.MetDataset` object.
+
+This module requires the following additional dependencies:
+
+- `metview (binaries and python bindings) <https://metview.readthedocs.io/en/latest/python.html>`_
+- `gcsfs <https://gcsfs.readthedocs.io/en/latest/>`_
+- `zarr <https://zarr.readthedocs.io/en/stable/>`_
+
+"""
+
+from __future__ import annotations
+
+import contextlib
+import dataclasses
+import datetime
+import hashlib
+import multiprocessing
+import pathlib
+import tempfile
+import warnings
+from collections.abc import Iterable
+from typing import Any
+
+import xarray as xr
+from overrides import overrides
+
+from pycontrails.core import cache, met_var
+from pycontrails.core.met import MetDataset
+from pycontrails.datalib._met_utils import metsource
+from pycontrails.datalib.ecmwf import common as ecmwf_common
+from pycontrails.datalib.ecmwf import variables as ecmwf_variables
+from pycontrails.datalib.ecmwf.model_levels import pressure_levels_at_model_levels
+from pycontrails.utils import dependencies
+
+try:
+    import gcsfs
+except ModuleNotFoundError as e:
+    dependencies.raise_module_not_found_error(
+        "arco_era5 module",
+        package_name="gcsfs",
+        module_not_found_error=e,
+        pycontrails_optional_package="zarr",
+    )
+
+MOISTURE_STORE = "gs://gcp-public-data-arco-era5/co/model-level-moisture.zarr"
+WIND_STORE = "gs://gcp-public-data-arco-era5/co/model-level-wind.zarr"
+SURFACE_STORE = "gs://gcp-public-data-arco-era5/co/single-level-surface.zarr"
+SINGLE_LEVEL_PREFIX = "gs://gcp-public-data-arco-era5/raw/date-variable-single_level"
+
+WIND_STORE_VARIABLES = [
+    met_var.AirTemperature,
+    met_var.VerticalVelocity,
+    met_var.EastwardWind,
+    met_var.NorthwardWind,
+    ecmwf_variables.RelativeVorticity,
+    ecmwf_variables.Divergence,
+]
+
+MOISTURE_STORE_VARIABLES = [
+    met_var.SpecificHumidity,
+    ecmwf_variables.CloudAreaFractionInLayer,
+    ecmwf_variables.SpecificCloudIceWaterContent,
+    ecmwf_variables.SpecificCloudLiquidWaterContent,
+]
+
+PRESSURE_LEVEL_VARIABLES = [*WIND_STORE_VARIABLES, *MOISTURE_STORE_VARIABLES, met_var.Geopotential]
+
+
+def _attribute_fix(ds: xr.Dataset | None) -> None:
+    """Fix GRIB attributes.
+
+    See:
+    https://github.com/google-research/arco-era5/blob/90f4c3dfc31692be73006e0ee841b620ecf81e7c/docs/moisture_dataset.py#L12
+    """
+
+    if ds is None:
+        return
+
+    for da in ds.values():
+        da.attrs.pop("GRIB_cfName", None)
+
+
+@dataclasses.dataclass
+class _ARCOERA5Datasets:
+    wind: xr.Dataset | None
+    moisture: xr.Dataset | None
+    surface: xr.Dataset | None
+
+
+def _required_wind_short_names(variables: list[met_var.MetVariable]) -> list[str]:
+    """Get the required wind variable short names needed to compute the requested variables."""
+    out = set()
+    for var in variables:
+        if var in (met_var.AirTemperature, met_var.Geopotential):
+            out.add("t")
+        elif var in (met_var.EastwardWind, met_var.NorthwardWind):
+            out.add("d")
+            out.add("vo")
+        elif var == met_var.VerticalVelocity:
+            out.add("w")
+        elif var == ecmwf_variables.RelativeVorticity:
+            out.add("vo")
+        elif var == ecmwf_variables.Divergence:
+            out.add("d")
+
+    return sorted(out)
+
+
+def _required_moisture_short_names(variables: list[met_var.MetVariable]) -> list[str]:
+    """Get the required moisture variable short names needed to compute the requested variables."""
+    moisture_vars = set(MOISTURE_STORE_VARIABLES)
+
+    out = set()
+    for var in variables:
+        if var in moisture_vars:
+            out.add(var.short_name)
+        elif var == met_var.Geopotential:
+            out.add("q")
+    return sorted(out)
+
+
+def _required_surface_short_names(variables: list[met_var.MetVariable]) -> list[str]:
+    """Get the required surface variable short names needed to compute the requested variables."""
+    if met_var.Geopotential in variables:
+        return ["lnsp", "z"]
+    return ["lnsp"] if variables else []
+
+
+def _download_data(
+    t: datetime.datetime,
+    variables: list[met_var.MetVariable],
+) -> _ARCOERA5Datasets:
+    """Download slices of the ARCO ERA5 model level Zarr stores."""
+
+    wind_vars = _required_wind_short_names(variables)
+    moisture_vars = _required_moisture_short_names(variables)
+    surface_vars = _required_surface_short_names(variables)
+
+    kw: dict[str, Any] = {"chunks": None, "consolidated": True}
+    wind_ds = xr.open_zarr(WIND_STORE, **kw)[wind_vars].sel(time=t) if wind_vars else None
+    moisture_ds = (
+        xr.open_zarr(MOISTURE_STORE, **kw)[moisture_vars].sel(time=t) if moisture_vars else None
+    )
+    surface_ds = (
+        xr.open_zarr(SURFACE_STORE, **kw)[surface_vars].sel(time=t) if surface_vars else None
+    )
+    return _ARCOERA5Datasets(wind=wind_ds, moisture=moisture_ds, surface=surface_ds)
+
+
+def _handle_metview(
+    data: _ARCOERA5Datasets,
+    variables: list[met_var.MetVariable],
+    pressure_levels: list[int],
+    grid: float,
+) -> xr.Dataset:
+    try:
+        import metview as mv
+    except ModuleNotFoundError as exc:
+        dependencies.raise_module_not_found_error(
+            "arco_era5 module",
+            package_name="metview",
+            module_not_found_error=exc,
+            extra="See https://metview.readthedocs.io/en/latest/install.html for instructions.",
+        )
+    except ImportError as exc:
+        msg = "Failed to import metview"
+        raise ImportError(msg) from exc
+
+    # Extract any moisture data (defined on a Gaussian grid)
+    gg_ml = mv.Fieldset()  # Gaussian grid on model levels
+    if data.moisture:
+        moisture_gg = mv.dataset_to_fieldset(data.moisture, no_warn=True)
+        gg_ml = mv.merge(gg_ml, moisture_gg)
+
+    # Convert any wind data (defined on a spherical harmonic grid) to the Gaussian grid
+    if data.wind:
+        wind_sh = mv.dataset_to_fieldset(data.wind, no_warn=True)
+        if met_var.EastwardWind in variables or met_var.NorthwardWind in variables:
+            uv_wind_sh = mv.uvwind(data=wind_sh, truncation=639)
+            wind_sh = mv.merge(wind_sh, uv_wind_sh)
+        wind_gg = mv.read(data=wind_sh, grid="N320")
+        gg_ml = mv.merge(gg_ml, wind_gg)
+
+    # Convert any surface data (defined on a spherical harmonic grid) to the Gaussian grid
+    surface_sh = mv.dataset_to_fieldset(data.surface, no_warn=True)
+    surface_gg = mv.read(data=surface_sh, grid="N320")
+    lnsp = surface_gg.select(shortName="lnsp")
+
+    # Compute Geopotential if requested
+    if met_var.Geopotential in variables:
+        t = gg_ml.select(shortName="t")
+        q = gg_ml.select(shortName="q")
+        zs = surface_gg.select(shortName="z")
+        zp = mv.mvl_geopotential_on_ml(t, q, lnsp, zs)
+        gg_ml = mv.merge(gg_ml, zp)
+
+    # Convert the Gaussian grid to a lat-lon grid
+    gg_pl = mv.Fieldset()  # Gaussian grid on pressure levels
+    for var in variables:
+        var_gg_ml = gg_ml.select(shortName=var.short_name)
+        var_gg_pl = mv.mvl_ml2hPa(lnsp, var_gg_ml, pressure_levels)
+        gg_pl = mv.merge(gg_pl, var_gg_pl)
+
+    # Regrid the Gaussian grid pressure level data to a lat-lon grid
+    ll_pl = mv.read(data=gg_pl, grid=[grid, grid])
+
+    ds = ll_pl.to_dataset()
+    return MetDataset(ds.rename(isobaricInhPa="level").expand_dims("time")).data
+
+
+def open_arco_era5_model_level_data(
+    t: datetime.datetime,
+    variables: list[met_var.MetVariable],
+    pressure_levels: list[int],
+    grid: float,
+) -> xr.Dataset:
+    r"""Open ARCO ERA5 model level data for a specific time and variables.
+
+    This function downloads moisture, wind, and surface data from the
+    `ARCO ERA5 <https://cloud.google.com/storage/docs/public-datasets/era5>`_
+    Zarr stores and interpolates the data to a target grid and pressure levels.
+
+    This function requires the `metview <https://metview.readthedocs.io/en/latest/python.html>`_
+    package to be installed. It is not available as an optional pycontrails dependency,
+    and instead must be installed manually.
+
+    Parameters
+    ----------
+    t : datetime.datetime
+        Time of the data to open.
+    variables : list[met_var.MetVariable]
+        List of variables to open. Unsupported variables are ignored.
+    pressure_levels : list[int]
+        Target pressure levels, [:math:`hPa`]. For ``metview`` compatibility, this should be
+        a sorted (increasing or decreasing) list of integers. Floating point values
+        are treated as integers in ``metview``.
+    grid : float
+        Target grid resolution, [:math:`\deg`]. A value of 0.25 is recommended.
+
+    Returns
+    -------
+    xr.Dataset
+        Dataset with the requested variables on the target grid and pressure levels.
+        Data is reformatted for :class:`MetDataset` conventions.
+        Data **is not** cached.
+
+    References
+    ----------
+    - :cite:`carverARCOERA5AnalysisReadyCloudOptimized2023`
+    - `ARCO ERA5 moisture workflow <https://github.com/google-research/arco-era5/blob/main/docs/moisture_dataset.py>`_
+    - `Model Level Walkthrough <https://github.com/google-research/arco-era5/blob/main/docs/1-Model-Levels-Walkthrough.ipynb>`_
+    - `Surface Reanalysis Walkthrough <https://github.com/google-research/arco-era5/blob/main/docs/0-Surface-Reanalysis-Walkthrough.ipynb>`_
+    """
+    data = _download_data(t, variables)
+
+    if not data.surface:
+        msg = "No variables provided"
+        raise ValueError(msg)
+
+    _attribute_fix(data.wind)
+    _attribute_fix(data.moisture)
+    _attribute_fix(data.surface)
+
+    # Ignore all the metview warnings from deprecated pandas usage
+    # This could be removed after metview updates their python API
+    with warnings.catch_warnings():
+        warnings.filterwarnings(
+            "ignore",
+            message="A value is trying to be set on a copy of a DataFrame",
+            category=FutureWarning,
+        )
+        return _handle_metview(data, variables, pressure_levels, grid)
+
+
+def open_arco_era5_single_level(
+    t: datetime.date,
+    variables: list[met_var.MetVariable],
+) -> xr.Dataset:
+    """Open ARCO ERA5 single level data for a specific date and variables.
+
+    Parameters
+    ----------
+    t : datetime.date
+        Date of the data to open.
+    variables : list[met_var.MetVariable]
+        List of variables to open.
+
+    Returns
+    -------
+    xr.Dataset
+        Dataset with the requested variables.
+        Data is reformatted for :class:`MetDataset` conventions.
+        Data **is not** cached.
+
+    Raises
+    ------
+    FileNotFoundError
+        If the variable is not found at the requested date. This could
+        indicate that the variable is not available in the ARCO ERA5 dataset,
+        or that the time requested is outside the available range.
+    """
+    gfs = gcsfs.GCSFileSystem()
+
+    prefix = f"{SINGLE_LEVEL_PREFIX}/{t.year}/{t.month:02}/{t.day:02}"
+
+    ds_list = []
+    for var in variables:
+        uri = f"{prefix}/{var.standard_name}/surface.nc"
+
+        try:
+            data = gfs.cat(uri)
+        except FileNotFoundError as exc:
+            msg = f"Variable {var.standard_name} at date {t} not found"
+            raise FileNotFoundError(msg) from exc
+
+        ds = xr.open_dataset(data)
+        ds_list.append(ds)
+
+    ds = xr.merge(ds_list)
+    return MetDataset(ds.expand_dims(level=[-1])).data
+
+
+class ARCOERA5(ecmwf_common.ECMWFAPI):
+    r"""ARCO ERA5 data accessed remotely through Google Cloud Storage.
+
+    This is a high-level interface to access and cache
+    `ARCO ERA5 <https://cloud.google.com/storage/docs/public-datasets/era5>`_
+    for a predefined set of times, variables, and pressure levels.
+
+        .. versionadded:: 0.50.0
+
+    Parameters
+    ----------
+    time : TimeInput
+        Time of the data to open.
+    variables : VariableInput
+        List of variables to open.
+    pressure_levels : PressureLevelInput, optional
+        Target pressure levels, [:math:`hPa`]. For pressure level data, this should be
+        a sorted (increasing or decreasing) list of integers. For single level data,
+        this should be ``-1``. By default, the pressure levels are set to the
+        pressure levels at each model level between 20,000 and 50,000 ft assuming a
+        constant surface pressure.
+    grid : float, optional
+        Target grid resolution, [:math:`\deg`]. Default is 0.25.
+    cachestore : CacheStore, optional
+        Cache store to use. By default, a new disk cache store is used. If None, no caching is done.
+    n_jobs : int, optional
+        EXPERIMENTAL: Number of parallel jobs to use for downloading data. By default, 1.
+    cleanup_metview_tempfiles : bool, optional
+        If True, cleanup all ``TEMP_DIRECTORY/tmp*.grib`` files. Implementation is brittle and may
+        not work on all systems. By default, True.
+
+    References
+    ----------
+    :cite:`carverARCOERA5AnalysisReadyCloudOptimized2023`
+
+    See Also
+    --------
+    :func:`open_arco_era5_model_level_data`
+    :func:`open_arco_era5_single_level`
+    """
+
+    grid: float
+
+    __marker = object()
+
+    def __init__(
+        self,
+        time: metsource.TimeInput,
+        variables: metsource.VariableInput,
+        pressure_levels: metsource.PressureLevelInput | None = None,
+        grid: float = 0.25,
+        cachestore: cache.CacheStore | None = __marker,  # type: ignore[assignment]
+        n_jobs: int = 1,
+        cleanup_metview_tempfiles: bool = True,
+    ) -> None:
+        self.timesteps = metsource.parse_timesteps(time)
+
+        if pressure_levels is None:
+            self.pressure_levels = pressure_levels_at_model_levels(20_000.0, 50_000.0)
+        else:
+            self.pressure_levels = metsource.parse_pressure_levels(pressure_levels)
+
+        self.paths = None
+        self.variables = metsource.parse_variables(variables, self.supported_variables)
+        self.grid = grid
+        self.cachestore = cache.DiskCacheStore() if cachestore is self.__marker else cachestore
+        self.n_jobs = max(1, n_jobs)
+        self.cleanup_metview_tempfiles = cleanup_metview_tempfiles
+
+    @property
+    def pressure_level_variables(self) -> list[met_var.MetVariable]:
+        """Variables available in the ARCO ERA5 model level data.
+
+        Returns
+        -------
+        list[MetVariable] | None
+            List of MetVariable available in datasource
+        """
+        return PRESSURE_LEVEL_VARIABLES
+
+    @property
+    def single_level_variables(self) -> list[met_var.MetVariable]:
+        """Variables available in the ARCO ERA5 single level data.
+
+        Returns
+        -------
+        list[MetVariable] | None
+            List of MetVariable available in datasource
+        """
+        return ecmwf_variables.SURFACE_VARIABLES
+
+    @overrides
+    def download_dataset(self, times: list[datetime.datetime]) -> None:
+        if not times:
+            return
+
+        # Download single level data sequentially
+        if self.is_single_level:
+            unique_dates = sorted({t.date() for t in times})
+            for t in unique_dates:
+                ds = open_arco_era5_single_level(t, self.variables)
+                self.cache_dataset(ds)
+            return
+
+        stack = contextlib.ExitStack()
+        if self.cleanup_metview_tempfiles:
+            stack.enter_context(_MetviewTempfileHandler())
+
+        n_jobs = min(self.n_jobs, len(times))
+
+        # Download sequentially if n_jobs == 1
+        if n_jobs == 1:
+            for t in times:
+                with stack:  # cleanup after each iteration
+                    _download_convert_cache_handler(self, t)
+            return
+
+        # Download in parallel
+        args = [(self, t) for t in times]
+        mp = multiprocessing.get_context("spawn")
+        with mp.Pool(n_jobs) as pool, stack:  # cleanup after pool finishes work
+            pool.starmap(_download_convert_cache_handler, args, chunksize=1)
+
+    @overrides
+    def create_cachepath(self, t: datetime.datetime) -> str:
+        if self.cachestore is None:
+            msg = "Attribute self.cachestore must be defined to create cache path"
+            raise ValueError(msg)
+
+        string = (
+            f"{t:%Y%m%d%H}-"
+            f"{'.'.join(str(p) for p in self.pressure_levels)}-"
+            f"{'.'.join(sorted(self.variable_shortnames))}-"
+            f"{self.grid}"
+        )
+        name = hashlib.md5(string.encode()).hexdigest()
+        cache_path = f"arcoera5-{name}.nc"
+
+        return self.cachestore.path(cache_path)
+
+    @overrides
+    def open_metdataset(
+        self,
+        dataset: xr.Dataset | None = None,
+        xr_kwargs: dict[str, Any] | None = None,
+        **kwargs: Any,
+    ) -> MetDataset:
+
+        if dataset:
+            msg = "Parameter 'dataset' is not supported for ARCO ERA5"
+            raise ValueError(msg)
+
+        if self.cachestore is None:
+            msg = "Cachestore is required to download data"
+            raise ValueError(msg)
+
+        xr_kwargs = xr_kwargs or {}
+        self.download(**xr_kwargs)
+
+        disk_cachepaths = [self.cachestore.get(f) for f in self._cachepaths]
+        ds = self.open_dataset(disk_cachepaths, **xr_kwargs)
+
+        mds = self._process_dataset(ds, **kwargs)
+
+        self.set_metadata(mds)
+        return mds
+
+    @overrides
+    def set_metadata(self, ds: xr.Dataset | MetDataset) -> None:
+        ds.attrs.update(
+            provider="ECMWF",
+            dataset="ERA5",
+            product="reanalysis",
+        )
+
+
+def _download_convert_cache_handler(arco: ARCOERA5, t: datetime.datetime) -> None:
+    """Download, convert, and cache ARCO ERA5 model level data."""
+    ds = open_arco_era5_model_level_data(t, arco.variables, arco.pressure_levels, arco.grid)
+    arco.cache_dataset(ds)
+
+
+def _get_grib_files() -> Iterable[pathlib.Path]:
+    """Get all temporary GRIB files."""
+    tmp = pathlib.Path(tempfile.gettempdir())
+    return tmp.glob("tmp*.grib")
+
+
+class _MetviewTempfileHandler:
+    def __enter__(self) -> None:
+        self.existing_grib_files = set(_get_grib_files())
+
+    def __exit__(self, exc_type, exc_value, traceback) -> None:  # type: ignore[no-untyped-def]
+        new_grib_files = _get_grib_files()
+        for f in new_grib_files:
+            if f not in self.existing_grib_files:
+                f.unlink(missing_ok=True)

pycontrails/datalib/ecmwf/common.py

@@ -0,0 +1,109 @@
+"""Common utilities for ECMWF Data Access."""
+
+from __future__ import annotations
+
+import logging
+import os
+from typing import Any
+
+LOG = logging.getLogger(__name__)
+
+import numpy as np
+import pandas as pd
+import xarray as xr
+from overrides import overrides
+
+from pycontrails.core import met
+from pycontrails.datalib._met_utils import metsource
+
+
+class ECMWFAPI(metsource.MetDataSource):
+    """Abstract class for all ECMWF data accessed remotely through CDS / MARS."""
+
+    @property
+    def variable_ecmwfids(self) -> list[int]:
+        """Return a list of variable ecmwf_ids.
+
+        Returns
+        -------
+        list[int]
+            List of int ECMWF param ids.
+        """
+        return [v.ecmwf_id for v in self.variables if v.ecmwf_id is not None]
+
+    def _process_dataset(self, ds: xr.Dataset, **kwargs: Any) -> met.MetDataset:
+        """Process the :class:`xr.Dataset` opened from cache or local files.
+
+        Parameters
+        ----------
+        ds : xr.Dataset
+            Dataset loaded from netcdf cache files or input paths.
+        **kwargs : Any
+            Keyword arguments passed through directly into :class:`MetDataset` constructor.
+
+        Returns
+        -------
+        MetDataset
+        """
+
+        # downselect variables
+        try:
+            ds = ds[self.variable_shortnames]
+        except KeyError as exc:
+            missing = set(self.variable_shortnames).difference(ds.variables)
+            msg = f"Input dataset is missing variables {missing}"
+            raise KeyError(msg) from exc
+
+        # downselect times
+        if not self.timesteps:
+            self.timesteps = ds["time"].values.astype("datetime64[ns]").tolist()
+        else:
+            try:
+                ds = ds.sel(time=self.timesteps)
+            except KeyError as exc:
+                # this snippet shows the missing times for convenience
+                np_timesteps = {np.datetime64(t, "ns") for t in self.timesteps}
+                missing_times = sorted(np_timesteps.difference(ds["time"].values))
+                msg = f"Input dataset is missing time coordinates {[str(t) for t in missing_times]}"
+                raise KeyError(msg) from exc
+
+        # downselect pressure level
+        # if "level" is not in dims and
+        # length of the requested pressure levels is 1
+        # expand the dims with this level
+        if "level" not in ds.dims and len(self.pressure_levels) == 1:
+            ds = ds.expand_dims(level=self.pressure_levels)
+
+        try:
+            ds = ds.sel(level=self.pressure_levels)
+        except KeyError as exc:
+            # this snippet shows the missing levels for convenience
+            missing_levels = sorted(set(self.pressure_levels) - set(ds["level"].values))
+            msg = f"Input dataset is missing level coordinates {missing_levels}"
+            raise KeyError(msg) from exc
+
+        # harmonize variable names
+        ds = met.standardize_variables(ds, self.variables)
+
+        kwargs.setdefault("cachestore", self.cachestore)
+        return met.MetDataset(ds, **kwargs)
+
+    @overrides
+    def cache_dataset(self, dataset: xr.Dataset) -> None:
+        if self.cachestore is None:
+            LOG.debug("Cache is turned off, skipping")
+            return
+
+        for t, ds_t in dataset.groupby("time", squeeze=False):
+            cache_path = self.create_cachepath(pd.Timestamp(t).to_pydatetime())
+            if os.path.exists(cache_path):
+                LOG.debug(f"Overwriting existing cache file {cache_path}")
+                # This may raise a PermissionError if the file is already open
+                # If this is the case, the user should explicitly close the file and try again
+                os.remove(cache_path)
+
+            ds_t.to_netcdf(cache_path)
+
+
+class CDSCredentialsNotFound(Exception):
+    """Raise when CDS credentials are not found by :class:`cdsapi.Client` instance."""