libraddask

Introduction

libRadtran (http://www.libradtran.org/) is a collection of tools for atmospheric radiative transfer computation. The main driver utility for libRadtran is an executable called uvspec. This utility reads an input file (generally *.INP) of keywords and keyword parameters that set up and execute the radiative transfer computation. A variety of radiative transfer solvers are provided with libRadtran, including various implementations of the classical Discrete Ordinates Radiative Transfer (DISORT) method. Further information and downloads are available at libRadtran website.

libraddask comprises three main parts:

1. libraddask.rad includes copies of the libRadtran files contained in libradtran-2.0.3/src_py, but converted to Python 3 status.

2. libraddask.rad also includes a number of utilities related to radiometry. This code was originally written by Derek Griffith (https://github.com/derekjgriffith/MORTICIA) for Python 2. The Python 3 converted files are included here. These include

   • Creating, reading and writing of MODTRAN-style .flt files for specification of sensor Spectral Response functions (SRFs)
   • Conversion of .flt class objects to lists of xarray.DataArray objects for use in the Spectra class.
   • The RadEnv class for computation and handling of Radiant Environment Maps (REMs).

3. Documentation and examples to use Dask for distributed libRadTran runs (i.e., on remote clusters).

Tools

The Python-3-converted libRadtran/src_py Python files are in the libraddask\libradtran folder. At the time of writing (20200420) the files in this folder are based on the libRadtran version 2.0.2 Python files, the port of the 2.0.3 files are still in process.

The utility files are in the rad/ folder.

Documentation

The documentation is provided in LaTeX format in the doc/ folder. The PDF version of the documentation might not be built in early beta releases, please build it yourself.

The doc/ folder also contains a Jupyter notebook that demonstrates the use of the libraddask package.

Installation and Requirements

libraddask requires Python 3.

A working installation of libRadtran is required to compute radiant environment maps and atmospheric transmittance. In the Monte carlo/statistical mode of operation, a compute cluster is generally required to achieve reasonable execution time of large numbers of runs.

Parallel computation is performed using the Dask package, which works in IPython/Jupyter notebooks as well as other Python launch modes.

libraddask has been developed and tested using the Anaconda distribution from Continuum Analytics.

Repository

The master repository for libraddask is hosted on GitHub at https://github.com/NelisW/libraddask.