Impact of vegetation phenology on anisotropy of artificial light at night - Evidence from multi-angle satellite observations

Anisotropy of artificial light at night (ALAN) has been revealed from satellite observations, as Visible Infrared Imaging Radiometer Suite Day/Night Band (VIIRS DNB) provides multi-angle measurements of ALAN. However, the knowledge behind this phenomenon is very limited. In this study, we hypothesize that vegetation phenology impacts the anisotropy of ALAN, which is defined as the change in radiant intensity with viewing zenith angle (VZA). The time series Normalized Difference Vegetation Index (NDVI) quantifying the vegetation phenology and Change Index (CI) quantifying the ALAN anisotropy were extracted from VNP13A1 and VNP46A2 products, respectively. We analyzed the effect of vegetation phenology by comparing the anisotropy of ALAN at the same location across different seasons in eleven suburban study areas in North America. The anisotropy of ALAN was found to exhibit obvious seasonal dynamic which is consistent with that of NDVI, and these two variables showed significant positive correlation at both pixel scale (0.41 < r < 0.79) and regional scale (0.56 < r < 0.92). Furthermore, we found that the seasonality of the ALAN anisotropy was significantly correlated with the seasonality of vegetation over the eleven study areas (r = 0.75). All these results suggest that vegetation growth can reduce the anisotropy of ALAN. In other words, vegetation growth leads to a more even distribution of emitted light in different directions, which supports our hypothesis. These findings are potentially useful to improve the quality of time series nighttime light data by angular normalization considering vegetation phenology and better build City Emission Function (CEF) for modeling astronomic light pollution.