Anisotropic characteristic of artificial light at night – Systematic investigation with VIIRS DNB multi-temporal observations

Abstract

The released VIIRS DNB nightly images, also known as VIIRS DNB daily nighttime images, provide rich information for time series analysis of global socioeconomic dynamics. Anisotropic characteristic is a possible factor that influences the VIIRS DNB radiance at night and its time series analysis. This study aims to investigate the relationship between viewing angles and VIIRS DNB radiance of Suomi NPP satellite in urban areas. First, twenty-nine points were selected globally to explore the angle variation of Suomi NPP satellite views at night. We found that the variation of the satellite viewing zenith angle (VZA) is consistent (e.g. between 0° and 70°) since the range of VZA is fixed depending on the sensor design, and the range of viewing azimuth angle (VAA) increases with the increase of latitude. Second, thirty points in cities of Beijing, Houston, Los Angeles, Moscow, Quito and Sydney, were used to investigate the angle-radiance relationship. We proposed a zenith-radiance quadratic (ZRQ) model and a zenith-azimuth-radiance binary quadratic (ZARBQ) model to quantify the relationship between satellite viewing angles and artificial light radiance, which has been corrected by removing the moonlight and atmospheric impact from VIIRS DNB radiance products. For all the thirty points, the ZRQ and ZARBQ analysis have averaged R2 of 0.50 and 0.53, respectively, which indicates that the viewing angles are important factors influencing the variation of the artificial light radiance, but extending zenith to zenith-azimuth does not much better explain the variation of the observed artificial light. Importantly, based on the data analysis, we can make the hypothesis that building height may affect the relationship between VZA and artificial light, and cold and hot spot effects are clearly found in tall building areas. These findings are potentially useful to reconstruct more stable time series VIIRS DNB images for socioeconomic applications by removing the angular effects.