`nanodisort`¶

class nanodisort.DisortState[source]¶

DISORT solver state.

This class provides an interface to the CDISORT radiative transfer model. CDISORT is a C port of the original DISORT Fortran package. The main differences are:

Improved memory management. CDISORT allocates memory dynamically to use memory buffers of the right size, and internally encapsulates data in structs. This overall improves performance compared to the Fortran implementation.
Double precision. CDISORT operates entirely with double-precision arithmetics for improved numerical stability.
Correction of intensity fields by Buras-Emde algorithm, included by Robert Buras.
Pseudospherical geometry for direct beam source, included by Arve Kylling.
Solution for a general source term, included by Arve Kylling.

__init__() → None[source]¶: Initialize the DISORT state. All parameters are initialized with their default values.

allocate(self) → None¶: Allocate memory for arrays based on configured dimensions.

solve(self) → None¶: Run the DISORT solver.

print_state(pad: int | None = 0) → None[source]¶

Print to the terminal the full state of the DISORT solver.

Parameters:: pad (int or None, default: 0) – Minimal amount of padding space between variable name and first “=” sign. Set to None to automatically align “=” signs. By default, no padding is applied.

Boolean flags

usrtau: bool¶: Return radiant quantities at user-specified optical thicknesses.

usrang: bool¶: Return radiant quantities at user-specified polar angles.

lamber: bool¶: Isotropically reflecting bottom boundary.

planck: bool¶: Include thermal emission.

spher: bool¶: Pseudo-spherical geometry (vs plane-parallel).

onlyfl: bool¶: Return only fluxes (not intensities).

quiet: bool¶: Suppress output messages.

intensity_correction: bool¶: Apply intensity correction.

old_intensity_correction: bool¶: Apply old-style intensity correction (may be required for some tests).

Dimensions

nstr: int¶: Number of streams (discretization of the azimuth angle dimension).

nlyr: int¶: Number of computational layers (discretization of the vertical dimension).

nmom: int¶: Number of phase function moments. Set to -1 to disable scattering.

ntau: int¶: Number of user optical thicknesses.

numu: int¶: Number of user polar angles.

nphi: int¶: Number of user azimuth angles.

nphase: int¶: Number of phase angle grid points for Buras-Emde correction.

Boundary conditions

fbeam: float¶: Intensity of incident parallel beam.

umu0: float¶: Polar angle cosine of incident beam.

phi0: float¶: Azimuth angle of incident beam (degrees).

fisot: float¶: Intensity of top-boundary isotropic illumination.

fluor: float¶: Intensity of bottom-boundary isotropic illumination.

albedo: float¶: Albedo of bottom boundary.

btemp: float¶: Temperature [K] of bottom boundary.

ttemp: float¶: Temperature [K] of top boundary.

temis: float¶: Emissivity of top boundary.

brdf_type: BRDFType¶: Type of BRDF (0=none, 4=Hapke). Set lamber=False to activate.

Other scalar parameters

accur: float¶: Convergence criterion for azimuthal series.

wvnmlo: float¶: Wavenumber lower bound [cm⁻¹] for Planck function.

wvnmhi: float¶: Wavenumber upper bound [cm⁻¹] for Planck function.

Optical property arrays

dtauc: ndarray[tuple[Any, ...], dtype[float64]]¶: Optical thicknesses of computational layers [nlyr].

ssalb: ndarray[tuple[Any, ...], dtype[float64]]¶: Single-scattering albedo of computational layers [nlyr].

pmom: ndarray[tuple[Any, ...], dtype[float64]]¶: Phase function moments [nmom_nstr+1, nlyr].

phase: ndarray[tuple[Any, ...], dtype[float64]]¶: Phase function values [nlyr, nphase] for Buras-Emde correction.

mu_phase: ndarray[tuple[Any, ...], dtype[float64]]¶: Phase angle cosines [nphase] for Buras-Emde correction.

Other input arrays

umu: ndarray[tuple[Any, ...], dtype[float64]]¶: Polar angle cosines [numu].

phi: ndarray[tuple[Any, ...], dtype[float64]]¶: Azimuthal angles [degrees] [nphi].

utau: ndarray[tuple[Any, ...], dtype[float64]]¶: User optical thicknesses [ntau].

temper: ndarray[tuple[Any, ...], dtype[float64]]¶: Temperatures [K] at levels [nlyr+1].

Output arrays (read-only, always present)

rfldir: ndarray[tuple[Any, ...], dtype[float64]]¶: Direct beam flux (without delta-M scaling) [ntau].

rfldn: ndarray[tuple[Any, ...], dtype[float64]]¶: Diffuse downward flux (without delta-M scaling) [ntau].

flup: ndarray[tuple[Any, ...], dtype[float64]]¶: Diffuse upward flux [ntau].

dfdt: ndarray[tuple[Any, ...], dtype[float64]]¶: Flux divergence d(net flux)/d(optical thickness) [ntau].

uavg: ndarray[tuple[Any, ...], dtype[float64]]¶: Mean intensity including direct beam [ntau].

uavgdn: ndarray[tuple[Any, ...], dtype[float64]]¶: Mean diffuse downward intensity [ntau].

uavgup: ndarray[tuple[Any, ...], dtype[float64]]¶: Mean diffuse upward intensity [ntau].

uavgso: ndarray[tuple[Any, ...], dtype[float64]]¶: Mean direct solar intensity [ntau].

Output arrays (read-only, intensity)

uu: ndarray[tuple[Any, ...], dtype[float64]]¶: Intensity at user angles [numu, ntau, nphi].

u0u: ndarray[tuple[Any, ...], dtype[float64]]¶: Azimuthally averaged intensity [numu, ntau].

uum: ndarray[tuple[Any, ...], dtype[float64]]¶: Intensity (corrected) at user angles [numu, ntau, nphi].

Output arrays (read-only, special boundary condition)

albmed: ndarray[tuple[Any, ...], dtype[float64]]¶: Albedo of medium [numu].

trnmed: ndarray[tuple[Any, ...], dtype[float64]]¶: Transmissivity of medium [numu].

class nanodisort.BatchSolver[source]¶

Parallel batch DISORT solver.

Solves multiple independent spectral problems in parallel using a C++ thread pool. Shared geometry and control flags are configured once on the solver instance; spectrally-varying optical properties (optical depths, single-scattering albedos, phase function moments, beam intensities, surface albedos) are provided as arrays with a leading batch dimension.

Examples

>>> import nanodisort as nd
>>> import numpy as np
>>> solver = nd.BatchSolver(nthreads=4)
>>> solver.nstr = 8
>>> solver.nlyr = 1
>>> solver.nmom = 8
>>> solver.ntau = 2
>>> solver.usrtau = True
>>> solver.usrang = False
>>> solver.lamber = True
>>> solver.onlyfl = True
>>> solver.quiet = True
>>> solver.umu0 = 1.0
>>> solver.phi0 = 0.0
>>> solver.set_utau(np.array([0.0, 1.0]))
>>> nbatch = 10
>>> solver.allocate(nbatch)
>>> solver.set_dtauc(np.ones((nbatch, 1)))
>>> solver.set_ssalb(np.full((nbatch, 1), 0.9))
>>> pmom = np.zeros((9, 1, nbatch))
>>> pmom[0, :, :] = 1.0
>>> solver.set_pmom(pmom)
>>> solver.set_fbeam(np.full(nbatch, np.pi))
>>> solver.set_albedo(np.zeros(nbatch))
>>> solver.solve()
>>> rfldir = solver.rfldir  # shape (nbatch, 2)

Warning

Thread safety requires quiet=True and lamber=True. These are enforced at allocation time, which means that this class does not achieve feature parity with DisortState.

__init__(nthreads: int = 0) → None[source]¶

Initialize the batch solver.

Parameters:: nthreads (int, optional) – Number of worker threads. 0 (default) uses hardware concurrency.

allocate(self, nbatch: int) → None¶

Allocate memory for nbatch parallel DISORT problems.

Must be called after setting shared dimensions and flags.

solve(self) → None¶: Solve all batch problems in parallel.

Status

nbatch: int¶: Number of batch items.

nthreads: int¶: Number of worker threads.

allocated: bool¶: Memory allocation status.

solved: bool¶: Whether solve() has been called.

Boolean flags

usrtau: bool¶: Return radiant quantities at user-specified optical thicknesses.

usrang: bool¶: Return radiant quantities at user-specified polar angles.

lamber: bool¶: Isotropically reflecting bottom boundary (must be True).

planck: bool¶: Include thermal emission.

spher: bool¶: Pseudo-spherical geometry.

onlyfl: bool¶: Return only fluxes (not intensities).

quiet: bool¶: Suppress output messages (must be True).

intensity_correction: bool¶: Apply intensity correction.

old_intensity_correction: bool¶: Apply old-style intensity correction.

Dimensions

nstr: int¶: Number of streams.

nlyr: int¶: Number of computational layers.

nmom: int¶: Number of phase function moments.

ntau: int¶: Number of user optical thicknesses.

numu: int¶: Number of user polar angles.

nphi: int¶: Number of user azimuth angles.

Shared boundary conditions

umu0: float¶: Polar angle cosine of incident beam.

phi0: float¶: Azimuth angle of incident beam (degrees).

fisot: float¶: Intensity of top-boundary isotropic illumination.

fluor: float¶: Intensity of bottom-boundary isotropic illumination.

btemp: float¶: Temperature [K] of bottom boundary.

ttemp: float¶: Temperature [K] of top boundary.

temis: float¶: Emissivity of top boundary.

Other shared scalar parameters

accur: float¶: Convergence criterion for azimuthal series.

wvnmlo: float¶: Wavenumber lower bound [cm⁻¹].

wvnmhi: float¶: Wavenumber upper bound [cm⁻¹].

Shared input arrays

set_umu(self, umu: ndarray[dtype=float64, shape=(*), order='C', device='cpu']) → None¶: Set shared polar angle cosines [numu].

set_phi(self, phi: ndarray[dtype=float64, shape=(*), order='C', device='cpu']) → None¶: Set shared azimuthal angles [nphi].

set_utau(self, utau: ndarray[dtype=float64, shape=(*), order='C', device='cpu']) → None¶: Set shared user optical thicknesses [ntau].

set_temper(self, temper: ndarray[dtype=float64, shape=(*), order='C', device='cpu']) → None¶: Set shared level temperatures [nlyr+1].

Batched input setters

set_dtauc(self, dtauc: ndarray[dtype=float64, shape=(*, *), order='C', device='cpu']) → None¶: Set optical thicknesses [nbatch, nlyr].

set_ssalb(self, ssalb: ndarray[dtype=float64, shape=(*, *), order='C', device='cpu']) → None¶: Set single-scattering albedos [nbatch, nlyr].

set_pmom(self, pmom: ndarray[dtype=float64, shape=(*, *, *), order='F', device='cpu']) → None¶: Set phase function moments [nmom_nstr+1, nlyr, nbatch] Fortran-contiguous.

set_fbeam(self, fbeam: ndarray[dtype=float64, shape=(*), order='C', device='cpu']) → None¶: Set incident beam intensities [nbatch].

set_albedo(self, albedo: ndarray[dtype=float64, shape=(*), order='C', device='cpu']) → None¶: Set bottom boundary albedos [nbatch].

set_utau_batched(self, utau: ndarray[dtype=float64, shape=(*, *), order='C', device='cpu']) → None¶: Set per-batch user optical thicknesses [nbatch, ntau].

Output arrays (read-only, always present)

rfldir: ndarray[tuple[Any, ...], dtype[float64]]¶: Direct beam flux [nbatch, ntau].

rfldn: ndarray[tuple[Any, ...], dtype[float64]]¶: Diffuse downward flux [nbatch, ntau].

flup: ndarray[tuple[Any, ...], dtype[float64]]¶: Diffuse upward flux [nbatch, ntau].

dfdt: ndarray[tuple[Any, ...], dtype[float64]]¶: Flux divergence [nbatch, ntau].

uavg: ndarray[tuple[Any, ...], dtype[float64]]¶: Mean intensity [nbatch, ntau].

uavgdn: ndarray[tuple[Any, ...], dtype[float64]]¶: Mean diffuse downward intensity [nbatch, ntau].

uavgup: ndarray[tuple[Any, ...], dtype[float64]]¶: Mean diffuse upward intensity [nbatch, ntau].

uavgso: ndarray[tuple[Any, ...], dtype[float64]]¶: Mean direct solar intensity [nbatch, ntau].

Output arrays (read-only, intensity)

uu: ndarray[tuple[Any, ...], dtype[float64]]¶: Intensity [nbatch, numu, ntau, nphi].

class nanodisort.BRDFType[source]¶

BRDF model type for non-Lambertian surface reflection.

Pass to DisortState.brdf_type together with lamber=False to activate a bidirectional reflectance model.

NONE = 0¶

HAPKE = 4¶

__new__(value)¶

nanodisort¶

Boolean flags

Dimensions

Boundary conditions

Other scalar parameters

Optical property arrays

Other input arrays

Output arrays (read-only, always present)

Output arrays (read-only, intensity)

Output arrays (read-only, special boundary condition)

Status

Boolean flags

Dimensions

Shared boundary conditions

Other shared scalar parameters

Shared input arrays

Batched input setters

Output arrays (read-only, always present)

Output arrays (read-only, intensity)

`nanodisort`¶