Modeling Skylight Angular Luminance Distribution from Routine Irradiance Measurements

Skylight is a nonuniform extended light source. Its intensity and spatial distribution vary as a function of insolation conditions. In addition to direct sunlight, sky-luminance angular distribution is the necessary and sufficient information required for calculating daylight penetration into any properly described environment. Because actual sky-luminance distribution data are available only in a handful of locations, it is essential to be able to estimate skylight distribution from routine measurements such as irradiance.

In a recent study, we evaluated six existing models designed to account for changing light spatial distribution as a function of insolation conditions. The conclusions of the study were:
1. The best possible performance of any of these models is limited by the random nature of cloud brightness patterns superimposed on homogeneous luminance distribution patterns for any given insolation condition.
2. The best models tested approached this ideal performance level; however, room for systematic improvement was noted.
3. The performance of empirically based models is satisfactory.
4. The key to a model's performance is its ability to adequately parameterize insolation conditions.

In this paper we present a new model that is consistent with points 2, 3, and 4 above. The model is experimentally derived from a large pool of data covering a wide range of insolation conditions, an relies on a parameterization of insolation conditions that has proven to be versatile and largely sit-independent. We address the random cloudiness issue in a separate paper.