Return to LP research page

ILLUMINA project summary

ILLUMINA is a monte carlo based radiative transfer model aimed to simulate the artificial light at night propagation into the environment. The model typically uses satellite data and locally acquired data as inputs. It comprises the calculation of the first two orders of scattering into the atmosphere and reflections on the ground based surfaces. Illumina can simulate both the sky radiance (clear and overcast conditions) and the radiance coming from a direct sight to lighting devices and lit surfaces. ILLUMINA calculations are made over a multiscale resolution grid which is maximal near the virtual observer. ILLUMINA accounts for various aerosol optical properties, relative humidity, ground level atmospheric pressure, ground reflectances spectrum, topography, lamps spectra, lamps angular emission functions, cloud base height, type of cloud, and subgrid obstacles. The radiance can be calculated toward any viewing angle, including downward. So that in addition to being able to simulate the night sky radiance, it can simulate a satellite view of the earth surface.

Current version

V2

1.  Heterogeneous modeling of artificial sky radiance

Martin Aubé, Ph.D.

1.1  Some publications and presentations



1.2  Downloading ILLUMINA v2

https://github.com/aubema/illumina

1.3  User's guides

1.4  Illumina v1 vs v2

1.5  Modeling and measurements contributions

Artificial sky radiance arise when light of artificial origin is travelling toward the sky. A part of this light is returned back to the Earth by the atmospheric constituents (molecules and aerosols). This phenomenon is highly correlated with an inapropriate and excessive use of the artificial light. In addition to limiting the access to the starry skies, light pollution acts negatively on fauna and the flora, on human health, and it also contributes to the growth of energy needs and thus to the production of greenhouse gases.

Professor Martin Aubé's group (GRAPHYCS) have developed hyper-spectral and multi-spectral detectors dedicated to the measurement of light pollution. GRAPHYCS also developed a sophisticated light pollution numerical model making it possible to simulate with precision the propagation of artificial light under a variety of conditions. The novelty of the model lies in the fact that it makes it possible to account for the heterogeneity of the environment in addition to being able to simulate the spectral behavior of the phenomenon. The present page is devoted to the description of this model named ILLUMINA.

The initial objective of the project ILLUMINA was to develop a new methodology to allow the detection of aerosols during night time. The method was based on the fact that aerosols are partly responsible (with molecules of the atmosphere) for artificial sky radiance. While promising, we focussed our efforts to study light pollution itself.

In this page, we dont want to discuss in detail the problems linked to light pollution, the reader is invited to visit the site www.darksky.org for further information on such topics.

We chose to follow a research track based on the development and the use of numerical modeling. We seek to develop a model allowing to predict the level of artificial sky radiance and the direct radiance from sources and lit surfaces according to a knowledge of the geographical distribution and physical properties of the light sources, a knowledge of the reflectance of the ground, of the topography, of the atmospheric content in aerosols, and of the subgrid obstacles properties. The model allow the simulation of the radiative properties of the atmosphere and thus we can compute the signal which would be detected by any instrument.


2.  Characteristics of the 3D radiative model ILLUMINA V2

  • 3D Calculation of single and double scattering with or without reflection on the ground
    To download animation
  • Multiscale model resolution (the finest is 20 m by default)
  • Uniformity of ground lambertian reflectance. We assume that the surfaces underneath lamps are the same.
  • Taking into account for luminaires total flux for each horizontal pixel. This information may be derived from VIIRS-DNB satellite imagery
  • Calculation of shadows from the topography
  • Correction for subgrid obstacles (trees, buildings) using mean distance between obstacles, mean obstacle height and the horizontal filling factor (in term of a blocking factor with respect to all azimutal angles)
  • Integration of the angular light pattern of the sources. We only consider vertical anisotropy.
  • Maximum of 256 different zones sources which can differ by their spectral power distribution, angular light pattern, post height, and obstacle properties.
  • Integration of the topography
  • Array size limit: 256 X 256 per layer with a default vertical resolution of 19 m.
  • Simulation of any observer position and viewing angle

3.  Model inputs

  • Horizontal resolution of the layer considered
  • Geographical distribution of lighting devices spectral flux
  • Up to 256 distinct angular light pattern can be simulated at the same time
  • Geographical distribution of the luminaire height
  • Ground reflectance
  • Digital elevation model
  • Angular light patterns (photometric file) for each kind of luminaire.
  • Aerosol optical properties (cross sections, phase function, optical thickness, angstrom coefficient)
  • Subgrid obstacles height, mean free path between obstacles and obstacle transparency
  • Ground level atmospheric pressure of the lowest elevation pixel

4.  Deriving aerosol optical properties from light pollution

This model can be used independently to simulate the fraction of light pollution due to the molecules and to atmospheric aerosols. By adjusting the aerosol content in an iterative way we seek to minimize the difference between modeling prediction and a measurement. The method rely on a certain number of simplifying assumptions concerning the population of aerosols and on the assumption of an horizontal isotropy of the aerosols concentrations on a scale comparable with the size of the modeling domain (typically about 100 km). In other words we suppose that at the time of an observation, the composition and the vertical profile of aerosols are horizontally uniform over the modeling domain.

A funny aspect of this new approach lies in the fact that we are taking advantage of the presence of a pollution type (light) to detect some other (aerosols).

5.  Outdated documents

GlossyBlue theme adapté par David Gilbert et Martin Aubé
Powered by PmWiki