Pyshdom performs 3D reconstruction of cloud microphysical properties from multi-angle, multi-spectral solar reflected radiation using a non-linear optimization procedure [1,2]. The core radiative transfer routines are sourced from the Fortran SHDOM (Spherical Harmonic Discrete Ordinate Method for 3D Atmospheric Radiative Transfer) code by Frank K. Evans . The python package was created by Aviad Levis, Amit Aides (Technion - Israel Institute of Technology) and Jesse Loveridge (University of Illinois).
At present pyshdom has the following features:
- Mie & Rayleigh scattering optical property calculations. Optical properties of other species (e.g. non-spherical ice or aerosol) can be included but must be calculated externally.
- Cloud data of varying complexity can be generated or read from LES output.
- Scalar radiative transfer from SHDOM with Perspective or Orthographic sensor geometries and Lambertian Surface.
- Each SHDOM solution is serial but independent wavelengths and pixel radiance calculations are parallelised in a shared memory framework.
- Local & Global optimization procedures for recovery of cloud microphysical properties (liquid water content, droplet effective radius, droplet effective variance) on 3D (or reduced order) grids from simulated (or observed) radiances.
- The calculation of optical properties and each SHDOM solution have been streamlined to minimize computational resources necessary at each iteration of the optimization routines.
- Implement vector SHDOM to include polarisation information in the optimization.
- Add additional sensor geometries (cross-track scan, push-broom) & expand to other surface types in SHDOM.
- Include a more flexible parallelisation scheme.
- Add useful regularisation options in the optimization procedure.
- Add further accelerations for computational efficiency.
- Include gaseous absorption for greater realism.
Updates in pyshdom3.0
- Code migration to python3
- Multispectral rendering and optimization
- Microphysical optimization
- Single scattering derivatives are exact (along broken ray paths)
- Main changes to SHDOM core Fortran code:
- Removal of global variables (property array)
- Particle mixing is done at runtime and not as an a-priori computation
- Mie computations are broken down to mono-disperse and poly-disperse
Installation using using anaconda package management
Start a clean virtual environment
conda create -n pyshdom python=3 source activate pyshdom
Install required packages
conda install anaconda dill tensorflow tensorboard pillow joblib
Install pyshdom distribution with (either install or develop flag)
python setup.py install
For basic usage follow the following jupyter notebook tutorials
- notebooks/Radiance Rendering [Single Image].ipynb
- notebooks/Radiance Rendering [Multiview].ipynb
- notebooks/Radiance Rendering [Multispectral].ipynb
For generating rendering and optimization scripts see the list below. The scripts folder contains another readme file with examples of how to run each script.
For info about command-line flags of each script use
python script.py --help
Usage and Contact
If you find this package useful please let me know at firstname.lastname@example.org, I am interested. If you use this package in an academic publication please acknowledge the appropriate publications (see LICENSE file).