pycontrails 0.54.0__cp312-cp312-macosx_10_13_x86_64.whl

pycontrails/models/ps_model/static/ps-synonym-list-20240524.csv

@@ -0,0 +1,103 @@
+ICAO Aircraft Code,PS ATYP
+A178,E190
+A19N,A20N
+A20N,A20N
+A21N,A21N
+A306,A306
+A30B,A30B
+A310,A310
+A313,A313
+A318,A318
+A319,A319
+A320,A320
+A321,A321
+A332,A332
+A333,A333
+A338,A338
+A339,A339
+A342,A342
+A343,A343
+A345,A345
+A346,A346
+A359,A359
+A35K,A35K
+A388,A388
+B37M,B37M
+B38M,B38M
+B39M,B39M
+B701,B752
+B712,B712
+B720,B752
+B721,B752
+B722,B722
+B732,B732
+B733,B733
+B734,B734
+B735,B735
+B736,B736
+B737,B737
+B738,B738
+B739,B739
+B741,B743
+B742,B742
+B743,B743
+B744,B744
+B748,B748
+B74D,B743
+B74R,B743
+B74S,B742
+B752,B752
+B753,B753
+B762,B762
+B763,B763
+B764,B764
+B772,B772
+B773,B773
+B77L,B77L
+B77W,B77W
+B788,B788
+B789,B789
+B78X,B78X
+BCS1,BCS1
+BCS3,BCS3
+BLCF,B77W
+C141,A310
+C17,B764
+C5,A345
+C5M,A345
+C919,A20N
+CRJ7,CRJ9
+CRJ9,CRJ9
+DC91,B712
+DC93,DC93
+E135,E135
+E145,E145
+E170,E170
+E190,E190
+E195,E195
+E290,E290
+E295,E295
+E3CF,B762
+E3TF,B762
+E737,B738
+E75L,E75L
+E75S,E75S
+E767,B763
+GLF5,GLF5
+IL62,A30B
+J328,E135
+KC39,A321
+MC23,A20N
+MD81,MD82
+MD82,MD82
+MD83,MD83
+MD87,MD82
+MD88,MD82
+MD90,MD83
+NIM,B738
+P1,B38M
+P8,B738
+R721,B722
+R722,B722
+RJ1H,RJ1H
+SLCH,A388

pycontrails/models/sac.py

@@ -0,0 +1,459 @@
+"""Schmidt-Appleman criteria (SAC)."""
+
+from __future__ import annotations
+
+import dataclasses
+from typing import Any, overload
+
+import numpy as np
+import scipy.optimize
+
+import pycontrails
+from pycontrails.core.flight import Flight
+from pycontrails.core.fuel import Fuel, JetA
+from pycontrails.core.met import MetDataset
+from pycontrails.core.met_var import AirTemperature, SpecificHumidity
+from pycontrails.core.models import Model, ModelParams
+from pycontrails.core.vector import GeoVectorDataset
+from pycontrails.models.humidity_scaling import HumidityScaling
+from pycontrails.physics import constants, thermo
+from pycontrails.utils.types import ArrayLike, ArrayScalarLike, apply_nan_mask_to_arraylike
+
+# -----------------
+# Models as classes
+# -----------------
+
+
+@dataclasses.dataclass
+class SACParams(ModelParams):
+    """Parameters for :class:`SAC`."""
+
+    #: Jet engine efficiency, [:math:`0 - 1`]
+    engine_efficiency: float = 0.3
+
+    #: Fuel type.
+    #: Overridden by Fuel provided on input ``source`` attributes
+    fuel: Fuel = dataclasses.field(default_factory=JetA)
+
+    #: Humidity scaling
+    humidity_scaling: HumidityScaling | None = None
+
+
+class SAC(Model):
+    """Determine points where Schmidt-Appleman Criteria is satisfied.
+
+    Parameters
+    ----------
+    met : MetDataset
+        Dataset containing "air_temperature", "specific_humidity" variables.
+    params : dict[str, Any], optional
+        Override :class:`SACParams` with dictionary.
+    **params_kwargs
+        Override :class:`SACParams` with keyword arguments.
+    """
+
+    name = "sac"
+    long_name = "Schmidt-Appleman contrail formation criteria"
+    met_variables = AirTemperature, SpecificHumidity
+    default_params = SACParams
+
+    @overload
+    def eval(self, source: Flight, **params: Any) -> Flight: ...
+
+    @overload
+    def eval(self, source: GeoVectorDataset, **params: Any) -> GeoVectorDataset: ...
+
+    @overload
+    def eval(self, source: MetDataset | None = ..., **params: Any) -> MetDataset: ...
+
+    def eval(
+        self, source: GeoVectorDataset | Flight | MetDataset | None = None, **params: Any
+    ) -> GeoVectorDataset | Flight | MetDataset:
+        """Evaluate the Schmidt-Appleman criteria along flight trajectory or on meteorology grid.
+
+        .. versionchanged:: 0.27.0
+
+            Humidity scaling now handled automatically. This is controlled by
+            model parameter ``humidity_scaling``.
+
+        .. versionchanged:: 0.48.0
+
+            If the ``source`` is a :class:`MetDataset`, the returned object will
+            also be a :class:`MetDataset`. Previous the "sac" :class:`MetDataArray`
+            was returned.
+
+        Parameters
+        ----------
+        source : GeoVectorDataset | Flight | MetDataset | None, optional
+            Input GeoVectorDataset or Flight.
+            If None, evaluates at the :attr:`met` grid points.
+        **params : Any
+            Overwrite model parameters before eval
+
+        Returns
+        -------
+        GeoVectorDataset | Flight | MetDataset
+            Returns 1 where SAC is satisfied, 0 everywhere else.
+            Returns ``np.nan`` if interpolating outside meteorology grid.
+
+        """
+
+        self.update_params(params)
+        self.set_source(source)
+        self.require_source_type((GeoVectorDataset, MetDataset))
+
+        if isinstance(self.source, GeoVectorDataset):
+            self.downselect_met()
+            self.source.setdefault("air_pressure", self.source.air_pressure)
+
+        humidity_scaling = self.params["humidity_scaling"]
+        scale_humidity = humidity_scaling is not None and "specific_humidity" not in self.source
+        self.set_source_met()
+
+        # apply humidity scaling, warn if no scaling is provided for ECMWF data
+        if scale_humidity:
+            humidity_scaling.eval(self.source, copy_source=False)
+
+        # Extract source data
+        air_temperature = self.source.data["air_temperature"]
+        specific_humidity = self.source.data["specific_humidity"]
+        air_pressure = self.source.data["air_pressure"]
+        engine_efficiency = self.get_source_param("engine_efficiency")
+
+        # Flight class has fuel attribute, use this instead of params
+        if isinstance(self.source, Flight):
+            fuel = self.source.fuel
+        else:
+            # NOTE: Not setting fuel on MetDataset source
+            fuel = self.get_source_param("fuel", set_attr=False)
+        assert isinstance(fuel, Fuel), "The fuel attribute must be of type Fuel"
+
+        ei_h2o = fuel.ei_h2o
+        q_fuel = fuel.q_fuel
+
+        G = slope_mixing_line(specific_humidity, air_pressure, engine_efficiency, ei_h2o, q_fuel)
+        T_sat_liquid_ = T_sat_liquid(G)
+        rh_crit_sac = rh_critical_sac(air_temperature, T_sat_liquid_, G)
+        rh = thermo.rh(specific_humidity, air_temperature, air_pressure)
+        sac_ = sac(rh, rh_crit_sac)
+
+        # Attaching some intermediate artifacts onto the source
+        self.source["G"] = G
+        self.source["T_sat_liquid"] = T_sat_liquid_
+        self.source["rh"] = rh
+        self.source["rh_critical_sac"] = rh_crit_sac
+        self.source["sac"] = sac_
+
+        # Tag output with additional metadata attrs
+        self.transfer_met_source_attrs()
+        self.source.attrs["pycontrails_version"] = pycontrails.__version__
+        if scale_humidity:
+            for k, v in humidity_scaling.description.items():
+                self.source.attrs[f"humidity_scaling_{k}"] = v
+        if isinstance(engine_efficiency, int | float):
+            self.source.attrs["engine_efficiency"] = engine_efficiency
+
+        return self.source
+
+
+# -------------------
+# Models as functions
+# -------------------
+
+
+def slope_mixing_line(
+    specific_humidity: ArrayLike,
+    air_pressure: ArrayLike,
+    engine_efficiency: float | ArrayLike,
+    ei_h2o: float,
+    q_fuel: float,
+) -> ArrayLike:
+    r"""Calculate the slope of the mixing line in a temperature-humidity diagram.
+
+    This quantity is often notated with ``G`` in the literature.
+
+    Parameters
+    ----------
+    specific_humidity : ArrayLike
+        A sequence or array of specific humidity values, [:math:`kg_{H_{2}O} \ kg_{air}`]
+    air_pressure : ArrayLike
+        A sequence or array of atmospheric pressure values, [:math:`Pa`].
+    engine_efficiency: float | ArrayLike
+        Engine efficiency, [:math:`0 - 1`]
+    ei_h2o : float
+        Emission index of water vapor, [:math:`kg \ kg^{-1}`]
+    q_fuel : float
+        Specific combustion heat of fuel combustion, [:math:`J \ kg^{-1} \ K^{-1}`]
+
+    Returns
+    -------
+    ArrayLike
+        Slope of the mixing line in a temperature-humidity diagram, [:math:`Pa \ K^{-1}`]
+    """
+    c_pm = thermo.c_pm(specific_humidity)  # Often taken as 1004 (= constants.c_pd)
+    return (ei_h2o * c_pm * air_pressure) / (constants.epsilon * q_fuel * (1.0 - engine_efficiency))
+
+
+def T_sat_liquid(G: ArrayLike) -> ArrayLike:
+    r"""Calculate temperature at which liquid saturation curve has slope G.
+
+    Parameters
+    ----------
+    G : ArrayLike
+        Slope of the mixing line in a temperature-humidity diagram.
+
+    Returns
+    -------
+    ArrayLike
+        Maximum threshold temperature for 100% relative humidity with respect to liquid,
+        [:math:`K`]. This can also be interpreted as the temperature at which the liquid
+        saturation curve has slope G.
+
+    References
+    ----------
+    - :cite:`schumannConditionsContrailFormation1996`
+
+    See Also
+    --------
+    :func:`T_sat_liquid_high_accuracy`
+
+    Notes
+    -----
+    Defined (using notation T_LM) in :cite:`schumannConditionsContrailFormation1996`
+    in the first full paragraph on page 10 as
+
+      for T = T_LC, the mixing line just touches [is tangent to]
+      the saturation curve. See equation (10).
+
+    The formula used here is taken from equation (31).
+    """
+    # FIXME: Presently, mypy is not aware that numpy ufuncs will return `xr.DataArray``
+    # when xr.DataArray is passed in. This will get fixed at some point in the future
+    # as `numpy` their typing patterns, after which the "type: ignore" comment can
+    # get ripped out.
+    # We could explicitly check for `xr.DataArray` then use `xr.apply_ufunc`, but
+    # this only renders our code more boilerplate and less performant.
+    # This comment is pasted several places in `pycontrails` -- they should all be
+    # addressed at the same time.
+    log_ = np.log(G - 0.053)
+    return -46.46 - constants.absolute_zero + 9.43 * log_ + 0.72 * log_**2  # type: ignore[return-value]
+
+
+def _e_sat_liquid_prime(T: ArrayScalarLike) -> ArrayScalarLike:
+    r"""Calculate derivative of :func:`thermo.e_sat_liquid`.
+
+    Parameters
+    ----------
+    T : ArrayScalarLike
+        Temperature, [:math:`K`].
+
+    Returns
+    -------
+    ArrayScalarLike
+        Derivative of :func:`thermo.e_sat_liquid`, [:math:``Pa \ K^{-1}`].
+    """
+    d_inside = 6096.9385 / (T**2) - 0.02711193 + 1.673952 * 1e-5 * 2 * T + 2.433502 / T
+    return thermo.e_sat_liquid(T) * d_inside
+
+
+def T_sat_liquid_high_accuracy(
+    G: ArrayLike,
+    maxiter: int = 5,
+) -> ArrayLike:
+    """Calculate temperature at which liquid saturation curve has slope G.
+
+    The function :func:`T_sat_liquid` gives a first order approximation to equation (10)
+    of the Schumann paper referenced below. This function uses Newton's method to
+    compute the numeric solution to (10).
+
+    Parameters
+    ----------
+    G : ArrayLike
+        Slope of the mixing line
+    maxiter : int, optional
+        Passed into :func:`scipy.optimize.newton`. Because ``T_sat_liquid`` is already
+        fairly accurate, few iterations are needed for Newton's method to converge.
+        By default, 5.
+
+    Returns
+    -------
+    ArrayLike
+        Maximum threshold temperature for 100% relative humidity with respect to liquid,
+        [:math:`K`].
+
+    References
+    ----------
+    - :cite:`schumannConditionsContrailFormation1996`
+
+    See Also
+    --------
+    :func:`T_sat_liquid_high`
+    """
+    init_guess = T_sat_liquid(G)
+
+    def func(T: ArrayLike) -> ArrayLike:
+        """Equation (10) from Schumann 1996."""
+        return _e_sat_liquid_prime(T) - G
+
+    return scipy.optimize.newton(func, init_guess, maxiter=maxiter)
+
+
+def rh_critical_sac(air_temperature: ArrayLike, T_sat_liquid: ArrayLike, G: ArrayLike) -> ArrayLike:
+    r"""Calculate critical relative humidity threshold of contrail formation.
+
+    Parameters
+    ----------
+    air_temperature : ArrayLike
+        A sequence or array of temperature values, [:math:`K`]
+    T_sat_liquid : ArrayLike
+        Maximum threshold temperature for 100% relative humidity with respect to liquid, [:math:`K`]
+    G : ArrayLike
+        Slope of the mixing line in a temperature-humidity diagram.
+
+    Returns
+    -------
+    ArrayLike
+        Critical relative humidity of contrail formation, [:math:`[0 - 1]`]
+
+    References
+    ----------
+    - :cite:`ponaterContrailsComprehensiveGlobal2002`
+    """
+    e_sat_T_sat_liquid = thermo.e_sat_liquid(T_sat_liquid)  # always positive
+    e_sat_T = thermo.e_sat_liquid(air_temperature)  # always positive
+
+    # Below, `rh_crit` can be negative
+    rh_crit = (G * (air_temperature - T_sat_liquid) + e_sat_T_sat_liquid) / e_sat_T
+
+    # Per Ponater, section 2.3:
+    # >>> The critical relative humidity `r_contr` can range from 0 to 1
+    # The "first order" term `G * (air_temperature - T_sat_liquid)` can be negative.
+    # Consequently, we clip rh_crit at 0 and 1.
+    # After clipping, when rh_crit = 0 the SAC is guaranteed to be satisfied.
+
+    # Per Ponater, beginning of section 2.3:
+    # >>> "No contrails are possible if T > T_contr" <<<
+    # In our notation, this inequality is `air_temperature > T_sat_liquid`
+    # We set the corresponding rh_crit to infinity, indicating no SAC is possible
+
+    # numpy case
+    if isinstance(rh_crit, np.ndarray):
+        rh_crit.clip(0.0, 1.0, out=rh_crit)  # clip in place
+        rh_crit[air_temperature > T_sat_liquid] = np.inf
+        return rh_crit
+
+    # xarray case
+    # FIXME: the two cases handle nans differently. Unfortunately, unit tests break
+    # if I try to consolidate to a single condition
+    rh_crit = rh_crit.clip(0.0, 1.0)
+    return rh_crit.where(air_temperature <= T_sat_liquid, np.inf)
+
+
+def sac(
+    rh: ArrayLike,
+    rh_crit_sac: ArrayLike,
+) -> ArrayLike:
+    r"""Points at which the Schmidt-Appleman Criteria is satisfied.
+
+    Parameters of type :class:`ArrayLike` must have compatible shapes.
+
+    Parameters
+    ----------
+    rh : ArrayLike
+        Relative humidity values
+    rh_crit_sac: ArrayLike
+        Critical relative humidity threshold of contrail formation
+
+    Returns
+    -------
+    ArrayLike
+        SAC state of each point indexed by the :class:`ArrayLike` parameters.
+        Returned array has floating ``dtype`` with values
+
+            - 0.0 signifying SAC fails
+            - 1.0 signifying SAC holds
+
+        NaN entries of parameters propagate into the returned array.
+    """
+    nan_mask = np.isnan(rh) | np.isnan(rh_crit_sac)
+
+    dtype = np.result_type(rh, rh_crit_sac)
+    sac_ = (rh > rh_crit_sac).astype(dtype)
+    return apply_nan_mask_to_arraylike(sac_, nan_mask)
+
+
+def T_critical_sac(
+    T_LM: ArrayLike,
+    relative_humidity: ArrayLike,
+    G: ArrayLike,
+    maxiter: int = 10,
+) -> ArrayLike:
+    r"""Estimate temperature threshold for persistent contrail formation.
+
+    This quantity is defined as ``T_LC`` in Schumann (see reference below). Equation (11)
+    of this paper implicitly defines ``T_LC`` as the solution to the equation
+    ::
+
+        T_LC = T_LM - (e_L(T_LM) - rh * e_L(T_LC)) / G
+
+    For relative humidity above 0.999, the corresponding entry from ``T_LM``
+    is returned (page 10, top of the right-hand column). Otherwise, the solution
+    to the equation above is approximated via Newton's method.
+
+    Parameters
+    ----------
+    T_LM : ArrayLike
+        Output of :func:`T_sat_liquid` calculation.
+    relative_humidity : ArrayLike
+        Relative humidity values
+    G : ArrayLike
+        Slope of the mixing line in a temperature-humidity diagram.
+    maxiter : int, optional
+        Passed into :func:`scipy.optimize.newton`. By default, 10.
+
+    Returns
+    -------
+    ArrayLike
+        Critical temperature threshold values.
+
+    References
+    ----------
+    - :cite:`schumannConditionsContrailFormation1996`
+    """
+    # Near U = 1, Newton's method is slow to converge (I believe this is because
+    # the function `func` has a double root at T_LM when U = 1, so Newton's method
+    # is somewhat degenerate here)
+    # But the answer is known in this case. This is discussed at the top of the
+    # right hand column on page 10 in Schumann 1996.
+    # We only apply Newton's method at points with rh bounded below 1 (scipy will
+    # raise an error if Newton's method is not converging well).
+    filt = (relative_humidity < 0.999) & np.isfinite(T_LM)
+    if not np.any(filt):
+        return T_LM
+
+    U_filt = relative_humidity[filt]
+    T_LM_filt = T_LM[filt]
+    e_L_of_T_LM_filt = thermo.e_sat_liquid(T_LM_filt)
+    G_filt = G[filt]
+
+    def func(T: ArrayLike) -> ArrayLike:
+        """Equation (11) from Schumann."""
+        return T - T_LM_filt + (e_L_of_T_LM_filt - U_filt * thermo.e_sat_liquid(T)) / G_filt
+
+    def fprime(T: ArrayLike) -> ArrayLike:
+        return 1.0 - U_filt * _e_sat_liquid_prime(T) / G_filt
+
+    # This initial guess should be less than T_LM.
+    # For relative_humidity away from 1, Newton's method converges quickly, and so
+    # any initial guess will work.
+    init_guess = T_LM_filt - 1.0
+    newton_approx = scipy.optimize.newton(
+        func, init_guess, fprime=fprime, maxiter=maxiter, disp=False
+    )
+
+    # For relative_humidity > 0.999, we just use T_LM
+    # We copy over the entire array T_LM here instead of using np.empty_like
+    # in order to keep typing compatible with xarray types
+    out = T_LM.copy()
+    out[filt] = newton_approx
+    return out