DisORT++

A modern C++17 implementation of the DISORT (Discrete Ordinates Radiative Transfer) algorithm. DisORT++ solves the radiative transfer equation in a plane-parallel or pseudo-spherical atmosphere, computing fluxes, mean intensities, and angular intensities at arbitrary optical depths and directions.

Features

• Discrete-ordinate method with arbitrary even stream count
• Henyey-Greenstein, Rayleigh, and tabulated phase functions
• Delta-M and Delta-M+ scaling for forward-peaked scattering
• Thermal emission with Planck-function integration
• Lambertian and BRDF surface reflection (RPV, Cox-Munk, Ambrals, Hapke)
• Plane-parallel and pseudo-spherical beam geometry
• Nakajima-Tanaka and Buras-Emde intensity corrections
• OpenMP-parallelised spectral loop functions for multi-wavenumber calculations
• Python bindings via pybind11

Two solver interfaces

	DisortSolver	DisortFluxSolver
Output	Fluxes + intensities	Fluxes only
Surface	Lambertian + BRDF	Lambertian
Streams	Any even number >= 4	4, 6, 8, 10, 12, 14, 16, 32, 64
Performance	Baseline	30-50% faster

Installation

From PyPI

pip install disortpp

From source (Python module)

git clone https://github.com/NewStrangeWorlds/DisORT.git
cd DisORT
pip install .

From source (C++ library only)

cd DisORT
mkdir build && cd build
cmake .. -DCMAKE_BUILD_TYPE=Release
make -j$(nproc)

Dependencies (Eigen 3.4, pybind11) are fetched automatically via CMake FetchContent.

Quick start (Python)

import disortpp
import numpy as np

nlyr = 4
nstr = 16

cfg = disortpp.DisortConfig(nlyr, nstr)
cfg.flags.use_lambertian_surface = True
cfg.flags.comp_only_fluxes = True
cfg.allocate()

cfg.delta_tau = [0.1, 0.5, 1.0, 2.0]
cfg.single_scat_albedo = [0.9, 0.9, 0.9, 0.9]
cfg.set_henyey_greenstein(g=0.8)

cfg.bc.direct_beam_flux = np.pi
cfg.bc.direct_beam_mu = 0.5
cfg.bc.surface_albedo = 0.1

solver = disortpp.DisortSolver()
result = solver.solve(cfg)

flux_up = np.array(result.flux_up)
print(f"Upward flux at TOA: {flux_up[0]:.6f}")

Spectral loops

Solve many wavenumber bands efficiently in a single C++ call with OpenMP parallelisation:

bands = [(100.0, 200.0), (200.0, 300.0), (300.0, 400.0)]
results = disortpp.solve_spectral_bands(cfg, bands)
flux_spectrum = np.array([r.flux_up[0] for r in results])

Quick start (C++)

#include "DisortConfig.hpp"
#include "DisortSolver.hpp"

using namespace disortpp;

int main() {
    DisortConfig cfg(4, 16);
    cfg.flags.use_lambertian_surface = true;
    cfg.flags.comp_only_fluxes = true;
    cfg.allocate();

    cfg.delta_tau = {0.1, 0.5, 1.0, 2.0};
    cfg.single_scat_albedo = {0.9, 0.9, 0.9, 0.9};
    cfg.setHenyeyGreenstein(0.8);

    cfg.bc.direct_beam_flux = 3.14159;
    cfg.bc.direct_beam_mu = 0.5;
    cfg.bc.surface_albedo = 0.1;

    DisortSolver solver;
    DisortResult result = solver.solve(cfg);

    double flux_up_toa = result.flux_up[0];
}

CMake build options

Option	Default	Description
CMAKE_BUILD_TYPE	Release	Release (-O3) or Debug (-g -Wall)
BUILD_PYTHON_BINDINGS	OFF	Build the disortpp Python module
DISORTPP_MARCH_NATIVE	ON	Use -march=native (disable for portable binaries)

Running the tests

cd build
ctest --output-on-failure

License

GNU General Public License v3 (GPLv3). See LICENSE for details.