def a_priori_shape(t):
transformation = Atanh()
transformation.z_max = 1.2
transformation.z_min = 0.0
x = np.maximum(np.minimum(0.7 - (270 - t) / 100.0, 0.7), 0.2)
x = 0.5 * np.ones(t.shape)
return transformation(x)
def __init__(self, covariance, mask=None, mask_value=1e-12):
super().__init__("H2O", covariance)
self.mask = mask
self.mask_value = mask_value
self.transformation = Atanh()
self.transformation.z_max = 1.1
self.transformation.z_min = 0.0
def _setup_retrieval(self):
"""
Setup the parts simulation used to perform the retrieval.
"""
for q in self.hydrometeors:
limit_low_1, limit_low_2 = q.limits_low
if hasattr(q, "limits_high"):
limit_high_1, limit_high_2 = q.limits_high
else:
limit_high_1, limit_high_2 = np.inf, np.inf
md = q.moments[0]
self.simulation.retrieval.add(md)
md.transformation = q.transformations[0]
md.retrieval.limit_low = q.limits_low[0]
md.retrieval.limit_high = limit_high_1
n0 = q.moments[1]
self.simulation.retrieval.add(n0)
n0.transformation = q.transformations[1]
n0.retrieval.limit_low = q.limits_low[1]
n0.retrieval.limit_high = limit_high_2
h2o = self.simulation.atmosphere.absorbers[-1]
h2o_a = [p for p in self.data_provider.subproviders \
if getattr(p, "name", "") == "H2O"]
if len(h2o_a) > 0:
h2o_a = h2o_a[0]
self.simulation.retrieval.add(h2o)
atanh = Atanh(0.0, 1.1)
h2o.transformation = h2o_a.transformation
h2o.retrieval.unit = RelativeHumidity()
self.h2o = h2o
else:
self.h2o = None
if self.include_cloud_water:
cw_a = [p for p in self.data_provider.subproviders \
if getattr(p, "name", "") == "cloud_water"][0]
cw = self.simulation.atmosphere.absorbers[-2]
self.simulation.retrieval.add(cw)
pl = PiecewiseLinear(cw_a)
cw.transformation = Composition(Log10(), pl)
cw.retrieval.limit_high = -3
self.cw = cw
else:
self.cw = None
t_a = [p for p in self.data_provider.subproviders \
if getattr(p, "name", "") == "temperature"]
if len(t_a) > 0:
t = self.simulation.atmosphere.temperature
self.temperature = t
self.simulation.retrieval.add(self.temperature)
else:
self.temperature = None
def a_priori_shape(t):
transformation = Atanh()
transformation.z_max = upper_limit
transformation.z_min = lower_limit
x = np.maximum(np.minimum(0.8 - (270 - t) / 100.0, 0.7), 0.1)
return transformation(x)
rh_mask = AltitudeMask(-1, 20e3)
rh_covariance = Diagonal(1.0, mask=rh_mask)
rh_covariance = SpatialCorrelation(rh_covariance, 2e3, mask=rh_mask)
rh_a_priori = FunctionalAPriori("H2O",
"temperature",
a_priori_shape,
rh_covariance,
mask=rh_mask,
mask_value=-100)
rh_a_priori = MaskedRegularGrid(rh_a_priori,
21,
rh_mask,
quantity="altitude",
provide_retrieval_grid=False)
rh_a_priori.unit = "rh"
rh_a_priori.transformation = Composition(Atanh(lower_limit, upper_limit),
PiecewiseLinear(rh_a_priori))
################################################################################
# Observation error
################################################################################
class ObservationError(DataProviderBase):
"""
Observation error covariance matrix provider for the LIRAS study.
This class works in the same way as the :class:`SensorNoiseAPriori` class
parts, which means that it takes the :code:`nedt` attributes of the given
sensors and concatenates these vectors into a combined diagonal matrix.
rh_mask = AltitudeMask(-1, 20e3)
rh_covariance = Diagonal(0.5, mask=rh_mask)
rh_covariance = SpatialCorrelation(rh_covariance, 2e3)
rh_a_priori = FunctionalAPriori("H2O",
"temperature",
a_priori_shape,
rh_covariance,
mask=rh_mask,
mask_value=-100)
rh_a_priori = MaskedRegularGrid(rh_a_priori,
21,
rh_mask,
quantity="altitude",
provide_retrieval_grid=False)
rh_a_priori.unit = "rh"
rh_a_priori.transformation = Composition(Atanh(0.0, 1.1),
PiecewiseLinear(rh_a_priori))
################################################################################
# Observation error
################################################################################
class ObservationError(DataProviderBase):
"""
Observation error covariance matrix provider for the LIRAS study.
This class works in the same way as the :class:`SensorNoiseAPriori` class
parts, which means that it takes the :code:`nedt` attributes of the given
sensors and concatenates these vectors into a combined diagonal matrix.
cloud_water_a_priori = FixedAPriori("cloud_water", -6, liquid_covariance,
mask = liquid_mask, mask_value = -20)
cloud_water_a_priori = ReferenceAPriori("cloud_water", liquid_covariance,
mask = liquid_mask, mask_value = -20,
a_priori = cloud_water_a_priori,
transformation = Log10())
cloud_water_a_priori = MaskedRegularGrid(cloud_water_a_priori, 7, liquid_mask,
"altitude", provide_retrieval_grid = False)
################################################################################
# Humidity
################################################################################
def a_priori_shape(t):
transformation = Atanh()
transformation.z_max = 1.2
transformation.z_min = 0.0
x = np.maximum(np.minimum(0.7 - (270 - t) / 100.0, 0.7), 0.2)
x = 0.5 * np.ones(t.shape)
return transformation(x)
z_grid = np.linspace(0, 20e3, 21)
rh_covariance = Diagonal(1.0)
rh_covariance = SpatialCorrelation(rh_covariance, 2e3)
rh_a_priori = FunctionalAPriori("H2O", "temperature", a_priori_shape, rh_covariance)
rh_a_priori = ReferenceAPriori("H2O", rh_covariance, transformation = Atanh(0.0, 1.2),
a_priori = rh_a_priori,
variable = "relative_humidity")
rh_a_priori = ReducedVerticalGrid(rh_a_priori, z_grid, "altitude",
provide_retrieval_grid = False)
def a_priori_shape(t):
transformation = Atanh()
transformation.z_max = 1.1
transformation.z_min = 0.0
x = np.maximum(np.minimum(0.5 - (270 - t) / 100.0, 0.5), 0.1)
return transformation(x)