Determination of unfiltered radiances from the clouds and the Earth's Radiant Energy System instrument

A new method for determining unfiltered shortwave (SW), longwave (LW), and window radiances from filtered radiances measured by the Clouds and the Ear th's Radiant Energy System (CERES) satellite instrument is presented. The m ethod uses theoretically derived regression coefficients between filtered a nd unfiltered radiances that are a function of viewing geometry, geotype, a nd whether cloud is present. Relative errors in instantaneous unfiltered ra diances from this method are generally well below 1% for SW radiances (std dev approximate to0.4% or approximate to1 W m(-2) equivalent flux), less th an 0.2% for LW radiances (std dev approximate to0.1% or approximate to0.3 W m(-2) equivalent flux), and less than 0.2% (std dev approximate to0.1%) fo r window channel radiances. When three months (June, July, and August of 1998) of CERES Earth Radiation Budget Experiment (ERBE)-like unfiltered radiances from the Tropical Rainf all Measuring Mission satellite between 20 degreesS and 20 degreesN are com pared with archived Earth Radiation Budget Satellite (ERBS) scanner measure ments for the same months over a 5-yr period (1985-89), significant scene-t ype dependent differences are observed in the SW channel. Full-resolution C ERES SW unfiltered radiances are approximate to7.5% (approximate to3 W m(-2 ) equivalent diurnal average flux) lower than ERBS over clear ocean, as com pared with approximate to1.7% (approximate to4 W m(-2) equivalent diurnal a verage flux) for deep convective clouds and approximate to6% (approximate t o4-6 W m(-2) equivalent diurnal average flux) for clear land and desert. Th is dependence on scene type is shown to be partly caused by differences in spatial resolution between CERES and ERBS and by errors in the unfiltering method used in ERBS. When the CERES measurements are spatially averaged to match the ERBS spatial resolution and the unfiltering scheme proposed in th is study is applied to both CERES and ERBS, the ERBS all-sky SW radiances i ncrease by approximate to1.7%, and the CERES radiances are now consistently approximate to3.5%-5% lower than the modified ERBS values for all scene ty pes. Further study is needed to determine the cause for this remaining diff erence, and even calibration errors cannot be ruled out. CERES LW radiances are closer to ERBS values for individual scene types-CERES radiances are w ithin approximate to0.1% (approximate to0.3 W m(-2)) of ERBS over clear oce an and approximate to0.5% (approximate to1.5 W m(-2)) over clear land and d esert.