Validation and Refinement of a Lunar Irradiance Model for Suomi NPP VIIRS Day-Night Band Quantitative Nighttime Applications

Monday, 15 December 2014: 10:50 AM
Steven D Miller1, Cindy Combs1, Sébastien Wagner2, Bartolomeo Viticchiè2, Andi Walther3 and Jeremy Solbrig4, (1)Cooperative Institute for Research in the Atmosphere, Fort Collins, CO, United States, (2)EUMETSAT, Darmstadt, Germany, (3)Cooperative Institute for Meteorological Satellite Studies, Madison, WI, United States, (4)Naval Research Laboratory, Monterey, CA, United States
The VIIRS Day-Night Band provides the first calibrated observations of nocturnal low-light visible/near-infrared (~500-900 nm response, 710 nm central wavelength) radiances, including reflected moonlight down to values of 3 × 10−5 W·m−2·sr−1. These novel measurements afford the first opportunity to attempt nighttime retrievals of optical depth for optically thick clouds when moonlight is available, thereby advancing our ability to observe the diurnal cycle of such structures as marine stratocumuli which are thought to play an important role in determining climate and climate feedbacks. In order to leverage the Day-Night Band measurements in this capacity, we must first convert the upwelling top-of-atmosphere radiances to equivalent values of reflectance. Doing so requires a detailed knowledge of the down-welling top-of-atmosphere lunar spectral irradiance which, unlike sunlight, varies significantly over the course of the ~29.5 day lunar cycle.

This research summarizes the ongoing development, validation, and refinement of a lunar irradiance model designed to convert Day-Night Band radiances to equivalent lunar reflectance. Comparisons between daytime and nighttime Day-Night Band reflectance for vicarious calibration targets offering radiometric stability (e.g., White Sands, Salar de Uyuni, Dome-C, and snow fields) confirms the model’s performance to within an expected ~10% uncertainty. An observed lunar-phase-dependent trend associated with the model’s assumption of a disk-averaged albedo was addressed via analysis of a version of the model adapted for comparison against Meteosat Second Generation SEVIRI lunar measurements. The analysis resulted in a phase-dependent 6th order polynomial correction to the model and expected model uncertainty improvements to within ~5%. Examples of lunar reflectance imagery for operational applications and the provisional quantitative application of Day-Night Band lunar reflectance to nighttime cloud optical property retrievals, bearing relevance to the diurnally resolved global climate data record, are shown.