PSFs for mapping artificial night sky luminance over large territories

TL;DR

This paper develops and evaluates point spread functions for mapping artificial night sky luminance over large areas, comparing model predictions with actual measurements and existing atlases to improve understanding of light pollution impacts.

Contribution

It introduces a method to model artificial night sky brightness using PSFs sensitive to key parameters, validated against real data and existing models.

Findings

The New world atlas overestimates sky brightness by 55%.

The Illuminav2 model underestimates it by 48%.

Model sensitivity varies with atmospheric and ground conditions.

Abstract

Knowledge of the night sky radiance over a large territory may be valuable informationto identify sites appropriate to astronomical observations or for assessing the impacts ofartificial light at night on ecosystems. Measuring the sky radiance can be a complex endeavourdepending on the desired temporal and spatial resolution. Similarly, modelling of artificialnight sky radiance for multiple points of a territory can represent a significant amount ofcomputing time depending on the complexity of the model used. The use of the convolutionof a point spread function with the light sources geographical distribution has been suggestedin order to model the sky radiance over large territories of hundreds of kilometres in size.We determine how the point spread function is sensitive to the main driving parameters ofthe artificial night sky radiance such as the wavelength, the ground reflectance, the obstaclesproperties, the Upward Light Output Ratio and the Aerosol Optical Depth using the Illuminav2 model. The obtained functions were used to model the artificial night sky brightness ofthe Mont-M\'egantic International Dark Sky Reserve for winter and summer conditions. Theresults were compared to the New world atlas of artificial night sky brightness, the Illuminav2 model and in situ Sky Quality Camera measurements. We found that the New world atlasoverestimates the artificial sky brightness by 55% whereas the Illumina model underestimatesit by 48%. This may be due to varying atmospherical conditions and the fact that the modelonly accounts for public light sources.