Spectral darkening correction to CERES instruments

The Clouds and the Earth's Radiant Energy System (CERES) mission is instrumental in providing highly accurate radiation and cloud products, thus enhancing our understanding of the Earth-atmosphere system. The CERES instrument is a scanning thermistor bolometer with three sensors that measure broadband radiances in the shortwave (0.3-5 μm), total (0.3-200 μm) and window (8-12 μm) radiometric regions. Two identical CERES instruments each are on board Earth Observing System (EOS) satellites Terra [Flight Model 1 (FM1) and 2 (FM2)] and Aqua (FM3, FM4). The instrument has the capability of operating in cross-track or rotating azimuth plane (RAP) scan mode. In RAP mode, the instrument scans in elevation while rotating in azimuth, thus providing radiance measurements for a more extensive combination of viewing configurations. Operation in RAP mode was ideal in developing CERES angular distribution models that led to vastly improved radiance to flux conversions.

An apparent drop in reflected flux over ocean (~2% in the first 4 years of the mission), inconsistent with other datasets, suggests spectral darkening in the shortwave channel optics, attributable to contamination in the sensor optics. The spectral degradation in the reflected solar bands (shortwave and shortwave portion of the total channel) results in the artificial decreasing trend in reflected solar measurements, as well as a diverging trend between the daytime and nighttime longwave fluxes with time. Spectral darkening is observed to be greater when the instrument is operated in the RAP mode. This is likely because during part of the RAP scan cycle the scan plane is aligned with the direction of motion, which may enable more contamination to collect on the optics.

This presentation describes the strategy for characterizing and compensating for the spectral degradation. A direct comparison approach of nadir radiances between two instruments on the same platform is employed to determine temporally varying optimal spectral response functions for the SW sensor. Regression of daytime and nighttime differences of longwave and window radiance measurements is applied to correct for degradation in the shortwave region of the Total channel sensor. Validation results show that application of resulting degradation factors removes the shortwave trends previously detected, and shows no significant daytime–nighttime long wave flux differences.