Next-generation radiance unfiltering process for the Clouds and Earth’s Radiant Energy System instrument
Abstract. The filtered radiances measured by the Clouds and the Earth’s Radiant Energy System (CERES) instruments are converted to shortwave (SW), longwave (LW), and window unfiltered radiances based on regressions developed from theoretical radiative transfer simulations to relate filtered and unfiltered radiances. This paper describes an update to the existing Edition 4 CERES unfiltering algorithm (Loeb et al., 2001), incorporating the most recent developments in radiative transfer modeling, ancillary input datasets, and increased computational and storage capabilities during the past 20 years. Simulations are performed with MODTRAN 5.4. Over land and snow, the surface Bidirectional Reflectance Distribution Function (BRDF) is characterized by a kernel-based representation in the simulations, instead of the Lambertian surface used in the Edition 4 unfiltering process. Radiance unfiltering is explicitly separated into 4 seasonally dependent land surface groups based on the spectral radiation similarities of different surface types (defined by International Geosphere-Biosphere Programme); over snow, it is separated into fresh snow, permanent snow, and sea ice. It contrasts to the Edition 4 unfiltering process that one set of regressions for land and snow, respectively.
The instantaneous unfiltering errors are estimated with independent test cases generated from radiative transfer simulations in which the ‘true’ unfiltered radiances from radiative transfer simulations are compared with the unfiltered radiances calculated from the regressions. Overall, the relative errors are mostly within ±0.5 % for SW, within ±0.2 % for daytime LW, and within ±0.1 % for nighttime LW for both CERES Terra Flight Model 1 (FM1) and Aqua FM3 instruments. The unfiltered radiances are converted to fluxes and compared to CERES Edition 4 fluxes. The global mean instantaneous fluxes for Aqua FM3 are reduced by less than 0.42 Wm-2 for SW and increased by less than 0.47 Wm-2 for daytime LW; for Terra FM1, they are reduced by less than 0.31 Wm-2 for SW and increased by less than 0.29 Wm-2 for daytime LW, though regional differences can be as large as 2.0 Wm-2. Nighttime LW flux differences are nearly negligible for both instruments.
Viewed (geographical distribution)