Robust method for lunar correction of nighttime VIIRS data

Moonlight reflection off the surface of ocean water with varying intensities contaminates the water-leaving radiances in night-time optical remote sensing data and hence limits the applicability of water color and other retrieved products. Correcting the moonlight/lunar reflection effects in night-time image data is challenging due to the lack of additional bands and large spatial and intensity variations across the scene. In this study, what we believe to be a novel algorithm is developed to correct night-time satellite measurements of water color for lunar reflections to retrieve water-leaving radiances from VIIRS Day/Night Band measurement data. The proposed algorithm involves correcting both the lunar illumination and lunar specular reflection components of moonlight reflection in VIIRS DNB satellite imagery. The algorithm employs the VIIRS products of single-band night-time radiances and moon illumination fractions for deriving an empirical formulation that relates the moon illumination fraction to the lunar component of radiance. The algorithm with a defined spatial filter, turbidity, and boundary conditions is tested on many night-time satellite image data. Qualitative comparisons with VIIRS Day/Night Band measurements showed that the new algorithm has good consistency in water-leaving radiance retrievals along various phases of the lunar cycle, and is capable of dealing with the dynamic lunar reflection intensities across the VIIRS DNB satellite image. This algorithm corrects the lunar reflection effect independently of ancillary data and simultaneous measurements; thus, it proves a robust tool for realizing the dynamic monitoring and inversion of water ocean color data rapidly.