A Python module for generating lookup tables using SBDART, a discrete-ordinate radiative transfer model capable of estimating incoming and outgoing flux and radiance given plane-parallel cloud layers, aerosol properties, surface types, etc.
- SBDART from this repository (Released for Fortran 2003)
- python>=3.9
Install quickrad as a package by cloning this repository into a
directory listed in $PYTHONPATH. Otherwise, just import relevant
methods from quickrad.py.
SBDART should be installed by cloning the repository and
running make -si to compile with gfortran.
To point quickrad to your SBDART installation, you can move the
subsequent binary file sbdart to a directory in $PATH, or modify
the SBDART_PATH variable at the top of quickrad.py.
The configuration for executing quickrad is the same as for SBDART, where a string key is mapped to a numeric value, string, or list of numeric values. The options are the same as those documented with the original repo at rtdoc.txt. In order to assemble a config, create a dict mapping these string keys to desired values.
args = {"idatm":1, "zcloud":"10,14", "tcloud":"8,13", "iout":7}
tmp_dir = Path("tmp")
data = quickrad.parse_iout(
iout_id=args["iout"],
sb_out=quickrad.dispatch_sbdart(args, tmp_dir),
print_stdout=False,
)The most basic way to use quickrad is to call quickrad.parse_iout
with a dict of any configured parameters that are different from the
defaults, as shown above. This will execute the radiative transfer
model one time and return the results as documented in the
corresponding parsing method.
There are also multiprocessed methods for generating lookup tables
for a subset of outputs, currently quickrad.sflux_over_fields for
spectral flux and quickrad.flux_over_fields for broad-band flux
over a series of user-defined orthogonal axes.
These methods allow the user to define a number of fields (SBDART
parameter keys) and a nested list of coordinate values at which each
corresponding field is evaluated. SBDART is executed at each
combination of coordinate values (in parallel if workers>1). The
user can also provide a dict of arguments applying to every run.
The lookup table will have dimensions corresponding to
each label/coordinate pair in order, followed by the default
dimensions for each output type. For example, if the fields
["sza", "btemp"] and coords ([30,60,90], [280,290,300,310]) are
provided toquickrad.sflux_over_fields, the subsequent lookup table
will be 5D like (solar_zenith, surface_temp, wavelength, altitude,
flux_type) since the output for spectral flux has 3 default dimensions.
See get_lut.py for a worked example of generating a lookup table.