class Radiation(EnergyBudget):
'''Base class for all band radiation models,
including grey and semi-grey model.
Input argument absorptivity is band absorptivity
(should same size as grid).
By default emissivity = absorptivity.
Subclasses can override this is necessary (e.g. for shortwave model)'''
def __init__(self,
absorptivity=None,
reflectivity=None,
albedo_sfc=0,
**kwargs):
super(Radiation, self).__init__(**kwargs)
if reflectivity is None:
reflectivity = np.zeros_like(self.Tatm)
if absorptivity is None:
absorptivity = np.zeros_like(self.Tatm)
self.absorptivity = absorptivity
self.reflectivity = reflectivity
self.albedo_sfc = albedo_sfc * np.ones_like(self.Ts)
self.flux_from_space = np.zeros_like(self.Ts)
self.flux_to_sfc = np.zeros_like(self.Ts)
self.flux_from_sfc = np.zeros_like(self.Ts)
self.flux_to_space = np.zeros_like(self.Ts)
self.heating_rate['Ts'] = np.zeros_like(self.Ts)
@property
def absorptivity(self):
return self.trans.absorptivity
@absorptivity.setter
def absorptivity(self, value):
# value should be a Field,
# or numpy array of same size as self.Tatm
try:
axis = value.domain.axis_index['lev']
except:
axis = self.Tatm.domain.axis_index['lev']
# if a single scalar is given, broadcast that to all levels
if len(np.shape(np.array(value))) is 0:
value = np.ones_like(self.Tatm) * value
elif value.shape != self.Tatm.shape:
raise ValueError(
'absorptivity must be a Field, a scalar, or match atm grid dimensions'
)
try:
self.trans = Transmissivity(absorptivity=value,
reflectivity=self.reflectivity,
axis=axis)
except:
self.trans = Transmissivity(absorptivity=value, axis=axis)
@property
def emissivity(self):
# This ensures that emissivity = absorptivity at all times
# needs to be overridden for shortwave classes
return self.absorptivity
@property
def transmissivity(self):
return self.trans.transmissivity
@transmissivity.setter
def transmissivity(self, value):
self.absorptivity = 1 - value
@property
def reflectivity(self):
return self.trans.reflectivity
@reflectivity.setter
def reflectivity(self, value):
# value should be a Field,
# or numpy array of same size as self.Tatm
try:
axis = value.domain.axis_index['lev']
except:
axis = self.Tatm.domain.axis_index['lev']
# if a single scalar is given, broadcast that to all levels
if len(np.shape(np.array(value))) is 0:
value = np.ones_like(self.Tatm) * value
elif value.shape != self.Tatm.shape:
raise ValueError(
'reflectivity must be a Field, a scalar, or match atm grid dimensions'
)
self.trans = Transmissivity(absorptivity=self.absorptivity,
reflectivity=value,
axis=axis)
def compute_emission(self):
return self.emissivity * blackbody_emission(self.Tatm)
def radiative_heating(self):
self.emission = self.compute_emission()
try:
fromspace = self.flux_from_space
except:
fromspace = np.zeros_like(self.Ts)
self.flux_down = self.trans.flux_down(fromspace, self.emission)
self.flux_reflected_up = \
self.trans.flux_reflected_up(self.flux_down, self.albedo_sfc)
# this ensure same dimensions as other fields
flux_down_sfc = self.flux_down[..., 0, np.newaxis]
flux_up_bottom = (self.flux_from_sfc +
self.flux_reflected_up[..., 0, np.newaxis])
self.flux_up = self.trans.flux_up(
flux_up_bottom, self.emission + self.flux_reflected_up[..., 1:])
self.flux_net = self.flux_up - self.flux_down
flux_up_top = self.flux_up[..., -1, np.newaxis]
# absorbed radiation (flux convergence) in W / m**2
self.absorbed = -np.diff(self.flux_net)
self.absorbed_total = np.sum(self.absorbed)
self.heating_rate['Tatm'] = self.absorbed
self.flux_to_sfc = flux_down_sfc
self.flux_to_space = flux_up_top
def _compute_heating_rates(self):
'''Compute energy flux convergences to get heating rates in W / m**2.'''
self.radiative_heating()
def flux_components_top(self):
'''Compute the contributions to the outgoing flux to space due to
emissions from each level and the surface.'''
N = self.lev.size
flux_up_bottom = self.flux_from_sfc
emission = np.zeros_like(self.emission)
this_flux_up = (np.ones_like(self.Ts) *
self.trans.flux_up(flux_up_bottom, emission))
sfcComponent = this_flux_up[..., -1]
atmComponents = np.zeros_like(self.Tatm)
flux_up_bottom = np.zeros_like(self.Ts)
# I'm sure there's a way to write this as a vectorized operation
# but the speed doesn't really matter if it's just for diagnostic
# and we are not calling it every timestep
for n in range(N):
emission = np.zeros_like(self.emission)
emission[..., n] = self.emission[..., n]
this_flux_up = self.trans.flux_up(flux_up_bottom, emission)
atmComponents[..., n] = this_flux_up[..., -1]
return sfcComponent, atmComponents
def flux_components_bottom(self):
'''Compute the contributions to the downwelling flux to surface due to
emissions from each level.'''
N = self.lev.size
atmComponents = np.zeros_like(self.Tatm)
flux_down_top = np.zeros_like(self.Ts)
# same comment as above... would be nice to vectorize
for n in range(N):
emission = np.zeros_like(self.emission)
emission[..., n] = self.emission[..., n]
this_flux_down = self.trans.flux_down(flux_down_top, emission)
atmComponents[..., n] = this_flux_down[..., 0]
return atmComponents
class GreyGas(Radiation):
'''Base class for all band radiation models,
including grey and semi-grey model.
Input argument absorptivity is band absorptivity
(should same size as grid).
By default emissivity = absorptivity.
Subclasses can override this is necessary (e.g. for shortwave model).
The following boundary and input values need to be specified by
user or parent process:
- albedo_sfc (default is zero)
- flux_from_space
- absorptivity
- reflectivity (default is zero)
These are accessible (and settable) as process attributes
Also stored in process.input dictionary
The following values are computed are stored in the .diagnostics dictionary:
- flux_from_sfc
- flux_to_sfc
- flux_to_space
- absorbed
- absorbed_total
- emission
- emission_sfc
- flux_reflected_up
(all in W/m2)
'''
def __init__(self, absorptivity=None, reflectivity=None, emissivity_sfc=1.,
albedo_sfc=0., **kwargs):
super(GreyGas, self).__init__(**kwargs)
newinput = ['reflectivity',
'absorptivity',
'emissivity_sfc',
'albedo_sfc',]
self.add_input(newinput)
if reflectivity is None:
reflectivity = np.zeros_like(self.Tatm)
if absorptivity is None:
absorptivity = np.zeros_like(self.Tatm)
self.absorptivity = absorptivity
self.reflectivity = reflectivity
self.emissivity_sfc = emissivity_sfc * np.ones_like(self.Ts)
self.albedo_sfc = albedo_sfc * np.ones_like(self.Ts)
# Initialize diagnostics
self.init_diagnostic('emission', 0. * self.Tatm)
self.init_diagnostic('emission_sfc', 0. * self.Ts)
self.init_diagnostic('flux_reflected_up')
@property
def absorptivity(self):
return self.trans.absorptivity
@absorptivity.setter
def absorptivity(self, value):
# value should be a Field,
# or numpy array of same size as self.Tatm
try:
axis = value.domain.axis_index['lev']
except:
axis = self.Tatm.domain.axis_index['lev']
# if a single scalar is given, broadcast that to all levels
if len(np.shape(np.array(value))) is 0:
value = np.ones_like(self.Tatm) * value
elif value.shape != self.Tatm.shape:
raise ValueError('absorptivity must be a Field, a scalar, or match atm grid dimensions')
try:
self.trans = Transmissivity(absorptivity=value,
reflectivity=self.reflectivity,
axis=axis)
except:
self.trans = Transmissivity(absorptivity=value, axis=axis)
self.input['absorptivity'] = value
@property
def emissivity(self):
# This ensures that emissivity = absorptivity at all times
# needs to be overridden for shortwave classes
return self.absorptivity
@property
def transmissivity(self):
return self.trans.transmissivity
@transmissivity.setter
def transmissivity(self, value):
self.absorptivity = 1 - value
@property
def reflectivity(self):
return self.trans.reflectivity
@reflectivity.setter
def reflectivity(self, value):
# value should be a Field,
# or numpy array of same size as self.Tatm
try:
axis = value.domain.axis_index['lev']
except:
axis = self.Tatm.domain.axis_index['lev']
# if a single scalar is given, broadcast that to all levels
if len(np.shape(np.array(value))) is 0:
value = np.ones_like(self.Tatm) * value
elif value.shape != self.Tatm.shape:
raise ValueError('reflectivity must be a Field, a scalar, or match atm grid dimensions')
self.trans = Transmissivity(absorptivity=self.absorptivity,
reflectivity=value,
axis=axis)
#self.input['reflectivity'] = value
def _compute_emission_sfc(self):
return self.emissivity_sfc * blackbody_emission(self.Ts)
def _compute_emission(self):
return self.emissivity * blackbody_emission(self.Tatm)
def _compute_fluxes(self):
''' All fluxes are band by band'''
self.emission = self._compute_emission()
self.emission_sfc = self._compute_emission_sfc()
fromspace = self._from_space()
self.flux_down = self.trans.flux_down(fromspace, self.emission)
self.flux_reflected_up = self.trans.flux_reflected_up(self.flux_down, self.albedo_sfc)
# this ensure same dimensions as other fields
self.flux_to_sfc = self.flux_down[..., -1, np.newaxis]
self.flux_from_sfc = (self.emission_sfc +
self.flux_reflected_up[..., -1, np.newaxis])
self.flux_up = self.trans.flux_up(self.flux_from_sfc,
self.emission + self.flux_reflected_up[...,0:-1])
self.flux_net = self.flux_up - self.flux_down
# absorbed radiation (flux convergence) in W / m**2 (per band)
self.absorbed = np.diff(self.flux_net, axis=-1)
self.absorbed_total = np.sum(self.absorbed, axis=-1)
self.flux_to_space = self._compute_flux_top()
def _compute_flux_top(self):
bandflux = self.flux_up[..., 0, np.newaxis]
return self._join_channels(bandflux)
def _flux_convergence_atm(self):
return self.absorbed
def _flux_convergence_sfc(self):
return ( self.flux_to_sfc - self.flux_from_sfc )
def _compute_radiative_heating(self):
self._compute_fluxes()
self.heating_rate['Tatm'] = self._join_channels(self.absorbed)
self.heating_rate['Ts'] = self._join_channels( self.flux_to_sfc -
self.flux_from_sfc )
def _from_space(self):
try:
fromspace = self.flux_from_space
except:
fromspace = np.zeros_like(self.Ts)
return self._split_channels(fromspace)
def _split_channels(self, flux):
'''Single channel for Grey Gas model.'''
return flux
def _join_channels(self, flux):
'''Single channel for Grey Gas model.'''
return flux
##### These routines may need fixing with flipped vertical grid...
def flux_components_top(self):
'''Compute the contributions to the outgoing flux to space due to
emissions from each level and the surface.'''
N = self.lev.size
flux_up_bottom = self.flux_from_sfc
emission = np.zeros_like(self.emission)
this_flux_up = (np.ones_like(self.Ts) *
self.trans.flux_up(flux_up_bottom, emission))
sfcComponent = this_flux_up[..., -1]
atmComponents = np.zeros_like(self.Tatm)
flux_up_bottom = np.zeros_like(self.Ts)
# I'm sure there's a way to write this as a vectorized operation
# but the speed doesn't really matter if it's just for diagnostic
# and we are not calling it every timestep
for n in range(N):
emission = np.zeros_like(self.emission)
emission[..., n] = self.emission[..., n]
this_flux_up = self.trans.flux_up(flux_up_bottom, emission)
atmComponents[..., n] = this_flux_up[..., -1]
return sfcComponent, atmComponents
def flux_components_bottom(self):
'''Compute the contributions to the downwelling flux to surface due to
emissions from each level.'''
N = self.lev.size
atmComponents = np.zeros_like(self.Tatm)
flux_down_top = np.zeros_like(self.Ts)
# same comment as above... would be nice to vectorize
for n in range(N):
emission = np.zeros_like(self.emission)
emission[..., n] = self.emission[..., n]
this_flux_down = self.trans.flux_down(flux_down_top, emission)
atmComponents[..., n] = this_flux_down[..., 0]
return atmComponents
class Radiation(EnergyBudget):
'''Base class for all band radiation models,
including grey and semi-grey model.
Input argument absorptivity is band absorptivity
(should same size as grid).
By default emissivity = absorptivity.
Subclasses can override this is necessary (e.g. for shortwave model)'''
def __init__(self, absorptivity=None, reflectivity=None,
albedo_sfc=0, **kwargs):
super(Radiation, self).__init__(**kwargs)
if reflectivity is None:
reflectivity = np.zeros_like(self.Tatm)
if absorptivity is None:
absorptivity = np.zeros_like(self.Tatm)
self.absorptivity = absorptivity
self.reflectivity = reflectivity
self.albedo_sfc = albedo_sfc*np.ones_like(self.Ts)
self.flux_from_space = np.zeros_like(self.Ts)
self.flux_to_sfc = np.zeros_like(self.Ts)
self.flux_from_sfc = np.zeros_like(self.Ts)
self.flux_to_space = np.zeros_like(self.Ts)
self.heating_rate['Ts'] = np.zeros_like(self.Ts)
@property
def absorptivity(self):
return self.trans.absorptivity
@absorptivity.setter
def absorptivity(self, value):
# value should be a Field,
# or numpy array of same size as self.Tatm
try:
axis = value.domain.axis_index['lev']
except:
axis = self.Tatm.domain.axis_index['lev']
# if a single scalar is given, broadcast that to all levels
if len(np.shape(np.array(value))) is 0:
value = np.ones_like(self.Tatm) * value
elif value.shape != self.Tatm.shape:
raise ValueError('absorptivity must be a Field, a scalar, or match atm grid dimensions')
try:
self.trans = Transmissivity(absorptivity=value,
reflectivity=self.reflectivity,
axis=axis)
except:
self.trans = Transmissivity(absorptivity=value, axis=axis)
@property
def emissivity(self):
# This ensures that emissivity = absorptivity at all times
# needs to be overridden for shortwave classes
return self.absorptivity
@property
def transmissivity(self):
return self.trans.transmissivity
@transmissivity.setter
def transmissivity(self, value):
self.absorptivity = 1 - value
@property
def reflectivity(self):
return self.trans.reflectivity
@reflectivity.setter
def reflectivity(self, value):
# value should be a Field,
# or numpy array of same size as self.Tatm
try:
axis = value.domain.axis_index['lev']
except:
axis = self.Tatm.domain.axis_index['lev']
# if a single scalar is given, broadcast that to all levels
if len(np.shape(np.array(value))) is 0:
value = np.ones_like(self.Tatm) * value
elif value.shape != self.Tatm.shape:
raise ValueError('reflectivity must be a Field, a scalar, or match atm grid dimensions')
self.trans = Transmissivity(absorptivity=self.absorptivity,
reflectivity=value,
axis=axis)
def compute_emission(self):
return self.emissivity * blackbody_emission(self.Tatm)
def radiative_heating(self):
self.emission = self.compute_emission()
try:
fromspace = self.flux_from_space
except:
fromspace = np.zeros_like(self.Ts)
self.flux_down = self.trans.flux_down(fromspace, self.emission)
self.flux_reflected_up = \
self.trans.flux_reflected_up(self.flux_down, self.albedo_sfc)
# this ensure same dimensions as other fields
flux_down_sfc = self.flux_down[..., 0, np.newaxis]
flux_up_bottom = (self.flux_from_sfc +
self.flux_reflected_up[..., 0, np.newaxis])
self.flux_up = self.trans.flux_up(flux_up_bottom,
self.emission+self.flux_reflected_up[...,1:])
self.flux_net = self.flux_up - self.flux_down
flux_up_top = self.flux_up[..., -1, np.newaxis]
# absorbed radiation (flux convergence) in W / m**2
self.absorbed = -np.diff(self.flux_net)
self.absorbed_total = np.sum(self.absorbed)
self.heating_rate['Tatm'] = self.absorbed
self.flux_to_sfc = flux_down_sfc
self.flux_to_space = flux_up_top
def _compute_heating_rates(self):
'''Compute energy flux convergences to get heating rates in W / m**2.'''
self.radiative_heating()
def flux_components_top(self):
'''Compute the contributions to the outgoing flux to space due to
emissions from each level and the surface.'''
N = self.lev.size
flux_up_bottom = self.flux_from_sfc
emission = np.zeros_like(self.emission)
this_flux_up = (np.ones_like(self.Ts) *
self.trans.flux_up(flux_up_bottom, emission))
sfcComponent = this_flux_up[..., -1]
atmComponents = np.zeros_like(self.Tatm)
flux_up_bottom = np.zeros_like(self.Ts)
# I'm sure there's a way to write this as a vectorized operation
# but the speed doesn't really matter if it's just for diagnostic
# and we are not calling it every timestep
for n in range(N):
emission = np.zeros_like(self.emission)
emission[..., n] = self.emission[..., n]
this_flux_up = self.trans.flux_up(flux_up_bottom, emission)
atmComponents[..., n] = this_flux_up[..., -1]
return sfcComponent, atmComponents
def flux_components_bottom(self):
'''Compute the contributions to the downwelling flux to surface due to
emissions from each level.'''
N = self.lev.size
atmComponents = np.zeros_like(self.Tatm)
flux_down_top = np.zeros_like(self.Ts)
# same comment as above... would be nice to vectorize
for n in range(N):
emission = np.zeros_like(self.emission)
emission[..., n] = self.emission[..., n]
this_flux_down = self.trans.flux_down(flux_down_top, emission)
atmComponents[..., n] = this_flux_down[..., 0]
return atmComponents