Spatial Interpolation of Outdoor Illumination at Night using Geostatistical Modeling

There is a growing interest in studying spatial and temporal variation of artificial nighttime light in order to design safe and ecologically-balanced lighting strategies. Geostatistical analysis tools can aid in characterizing spatial illuminance patterns and interpolating values for locations where light samples were not taken. However, much of the research on spatial interpolation of light has studied sky brightness or ground illumination, without focusing on illuminance as perceived by the human eye. This paper investigates how geostatistical modeling can be used to visualize and predict outdoor illumination at night at the pedestrian scale. Field measurements of vertical illumination are taken across a campus environment and mapped using GIS software. Geostatistical models using kriging techniques are developed and cross-validated to evaluate prediction certainty. Results suggest that the use of geostatistical analysis to map and analyze the spatial distribution of illuminance across a campus area has potential to better predict the nighttime light environment and offers a promising method for producing spatially-complete maps of illuminance ranges to understand existing conditions and identify future needs for exterior lighting design. A discussion includes opportunities and improvements to the methodology for predicting illuminance, as well as considerations for the dynamic characteristics of nighttime lighting and the variable visual qualities of the surrounding built environment.