One-dimensional experiment features#
Eradiate ships with an experiment dedicated to the simulation of radiative transfer on one-dimensional geometries. While the computation is not actually one-dimensional (Eradiate’s kernel is a 3D Monte Carlo ray tracer), scene setup ensures that the computed quantities are equivalent to what would be obtained with a proper 1D code. This guide introduces the features, configuration format and usage of this component.
Eradiate’s one-dimensional experiment is implemented by the
Instances are configured using the class’s constructor. The
operational mode, which defines how the spectral dimension of the
computation is handled, must be selected prior to instantiating the class.
The constructor only accepts keyword arguments, each of which can be passed an
object of an expected type or, alternatively and for relevant entries,
dictionaries. These dict-configurable entries all require a
which selects a programmatic component in Eradiate. In the following, the
type value corresponding to each feature is specified in brackets.
In addition to the
type parameter, an object configuration dictionary should
contain the parameters required to initialise the corresponding programmatic
element. Each allowed option is referenced, as well as the corresponding class
in the Eradiate codebase. Class documentation lists all their parameters and
Parameters can usually be omitted; in that case, they will be assigned
default values. Default values are documented in the reference documentation of
AtmosphereExperiment class and other Eradiate classes.
- Plane parallel [
By default, the surface and atmosphere are assumed translationally invariant in the X and Y directions. This approximation provides satisfactory accuracy in many situations.
- Spherical shell [
When this geometry configuration is used, the scene is built with a rotationally invariant symmetry. This configuration accounts for the roundedness of the planet. This feature is experimental and is being validated.
The one-dimensional experiment currently supports only one illumination type.
- Directional illumination [
An infinitely distant emitter which emits light in a single direction (angular Dirac delta distribution of incoming radiance). This type of illumination is used to simulate incoming Solar radiation. By default, a Solar spectrum is automatically selected.
In addition, this angular distribution can be associated a spectrum. A variety of pre-defined Solar irradiance spectra are defined (see Solar irradiance spectra for a complete list of shipped irradiance spectrum datasets).
This experiment currently supports the computation of radiative quantities at the top of the atmosphere. This parameter can be specified as a single measure, or as a list of measures.
- Distant radiancemeter [
This flexible measure records radiance exiting the scene. In practice, it outputs the top-of-atmosphere radiance under the set illumination conditions. The viewing directions for which radiance is computed can be controlled easily using various class method constructors.
When this measure is used, a number of derived quantities are computed. In the next paragraph, quantities available after post-processing are associated to the name of their corresponding field in the results dataset.
- TOA outgoing radiance [
This is the radiance reflected by the entire scene (surface and atmosphere), since the scene only contains infinitely distant illumination.
- TOA bidirectional reflectance distribution function (TOA BRDF) [
The TOA leaving radiance is post-processed together with scene illumination parameters to compute the TOA BRDF.
- TOA bidirectional reflectance factor (TOA BRF) [
The TOA BRDF normalised by the BRDF of a non-absorbing diffuse (Lambertian) surface.
- TOA outgoing radiance [
- Distant fluxmeter [
This measure records the flux leaving the scene (in W/m²/nm) over the entire hemisphere. It is mostly used to compute the scene albedo. The following quantities are available from the results dataset:
- Radiosity [
The flux leaving the scene in W/m²/nm.
- Albedo [
The total scene albedo.
- Radiosity [
An atmosphere can be optionally added to the scene. Currently, two types of atmosphere are supported.
- Homogeneous atmosphere [
The atmosphere has spatially invariant radiative properties.
- Heterogeneous atmosphere [
The atmosphere has spatially varying radiative properties along the altitude coordinate. The
HeterogeneousAtmosphereclass is configured by specifying a molecular component (
MolecularAtmosphere), describing absorption and scattering by atmospheric gases, and an arbitrary number of aerosol layers (
In this experiment, surfaces are smooth and their geometry is controlled by the
geometry parameter. Only the surface’s radiative properties can be selected.
The bidirectional scattering distribution function (BSDF) can be directly passed
surface parameter: Eradiate’s internals will wrap them in an
- Diffuse surface [
A diffuse or Lambertian surface reflects incoming radiation isotropically, regardless the incoming direction. This behaviour is modelled by the Lambert BRDF, parametrised by a reflectance parameter.
- Rahman-Pinty-Verstraete (RPV) surface [
This reflection model features an anisotropic behaviour and is commonly used for land surface reflection modelling. Eradiate implements several variants of it with 3 or 4 parameters.
- Black surface [
The black surface absorbs all incoming radiation, irrespective of incident angle or wavelength.
Digital elevation model [
Eradiate extends the standard functionalities of one-dimensional simulations with
digital elevation models (DEM). A three-dimensional surface can be defined with the
parameter. The DEM can be defined in several ways and it can be used with any BSDF type
mentioned in the surface section above. Since a DEM surface model always has a finite
horizontal extent, Eradiate adds horizontal elements to the edge of the DEM surface
to prevent rays from escaping under it. The remaining surface area outside of the DEM is
covered with the surface specified in the
Note that Eradiate does not adjust the horizontal level of the flat surface automatically. If the DEM contains elevation values below the flat surface level (0m per default), the 3D surface will intersect the flat surface and the areas below the flat surface’s level will be obscured by it.
There are three ways to define a DEM geometry:
- An xarray DataArray [
A DataArray defining a digital elevation model needs to have two coordinates named lat for the latitude and lon for the longitude, specified in degrees and the elevation data specified in kernel units of length.
- Triangulated meshes [
To define the DEM using a triangulated mesh users can supply either a .obj file or a .ply file. The mesh file will be interpreted as kernel units of length. In this case no constructor method is used. Instead the shape member of the dem class is directly defined with this mesh shape.
- Analytical functions [
Digital elevation models can be defined using functions, which take an x and y position and return the corresponding elevation value.
AtmosphereExperiment.run() method stores the computed results in the
results attribute as a dictionary mapping measure identifiers to a
Dataset object. Each data set has one variable for each
computed physical quantity (e.g. spectral irradiance, leaving radiance, BRDF
and BRF for the
distant measure). Results can then be easily exported to
files (e.g. NetCDF) and visualised using xarray’s integrated plotting
features or external plotting components.