A framework to generate local spectral skies for spectral daylight simulations

Abstract

Spectral daylight simulations are becoming increasingly important in predicting non-visual responses to light and more accurate daylight colour and temporal patterns. However, the availability of locally measured spectral sky data is limited and current spectral sky models do not capture the local colour variability of skies. This paper presents a novel framework for generating high-accuracy, location-specific spectral sky data using typically measured atmospheric data at meteorological stations as inputs to physics-based radiative transfer programs like libRadtran. A comparison of measured and simulated data across the year demonstrates that using location-specific atmospheric profiles leads to significantly more accurate predictions of irradiances, spectral global and direct irradiances, and better captures the variability of clear skies regarding colour accuracy and appearance than simulations with default settings.