Exending the model

Adding new functions

The atmosphere class can be easily extended with custom atmospheric properties. The easiest way to do this, is to subclass from Atmosphere, and to add custom methods to your new, extended atmosphere object (below called MyExtendedAtmosphere).

>>> from ambiance import Atmosphere

>>> class MyExtendedAtmosphere(Atmosphere):
...     @property
...     def specific_humidity(self):
...         return 2*self.h  # Define your own function of altitude here...
...

>>> atmos = MyExtendedAtmosphere([0, 1, 100])
>>> atmos.specific_humidity
array([  0.,   2., 200.])

In the example above, an additional atmospheric property (specific_humidity) is added. By adding the decorator @property, we do not need to call the function using parentheses. The example shows a dummy function which simply multiplies the altitude by a factor of two, but you can replace the expression with any function you like. The variable self.h is the geometric altitude (self.H is the geopotential altitude, see also User guide).

Hint

When subclassing, all the existing functionality will still remain and you will be able to access the other properties like temperature or pressure.

Avoiding if-statements

Consider some atmospheric property \(p\) as a function of altitude \(h\). Now, if there is a kink or sudden change in the model \(p = p(h)\), we may need different expressions for the different regions. Lets consider the following graph as a hypothetical example model.

Up to some altitude, the property \(p\) is constant. At higher altitude the function looks like a sine-curve. With the Numpy library we can handle such models without the need for any if-statements

>>> from ambiance import Atmosphere
>>> import numpy as np

>>> class MyExtendedAtmosphere(Atmosphere):
...     @property
...     def specific_humidity(self):
...         # Get the array positions for altitudes below and above 1000 meter.
...         # Example:
...         # self.h = [0, 500, 900, 2000, 50e3]
...         # --> pos_lower = [1, 1, 1, 0, 0]
...         # --> pos_upper = [0, 0, 0, 1, 1]
...         pos_lower = (self.h < 1000).astype(int)
...         pos_upper = 1 - pos_lower
...
...         spec_humidity = pos_lower*222  # if h < 0
...         spec_humidity = spec_humidity + pos_upper*np.sin(self.h)  # if h >= 0
...
...         return spec_humidity
...

>>> atmos = MyExtendedAtmosphere([0, 500, 900, 2000, 50e3])
>>> atmos.specific_humidity
array([222.        , 222.        , 222.        ,   0.9300395 ,
        -0.99984019])

Additional arguments

You may of course define functions with additional arguments. In this case we omit the decorator @property. However, this means we cannot retrieve our atmospheric property by writing atmos.specific_humidity. Instead, we need to call the method using parentheses.

>>> from ambiance import Atmosphere
>>> import numpy as np

>>> class MyExtendedAtmosphere(Atmosphere):
...     def specific_humidity(self, model=1):
...         if model == 1:
...             return 2*self.h
...         else:
...             return np.log10(self.h)
...

>>> atmos = MyExtendedAtmosphere([1, 10, 100])
>>> atmos.specific_humidity()
array([  2.,  20., 200.])
>>> atmos.specific_humidity(model=2)
array([0., 1., 2.])