Analysis of Temporal Sampling Errors in CERES Data Products

The Clouds and the Earth's Radiant Energy System (CERES) Experiment is the latest and most accurate satellite-based instrument designed to measure the Earth's global energy budget. With improvements in instrument calibration accuracy and stability, coupled with the development of new angular directional models, temporal sampling is the largest remaining error source for CERES regional mean fluxes. To improve sampling of the diurnal cycle, CERES was designed as a 3-satellite constellation with instruments aboard the sun-synchronous Terra and Aqua satellites and the temporally precessing TRMM spacecraft. However, the failure of the TRMM instrument in April 2000, and launch delays for Terra and Aqua resulted in single-satellite coverage for most of the mission to date. Pre-launch error analysis using two months of data from a single GOES satellite demonstrated that monthly zonal mean shortwave flux errors due to temporal sampling could exceed 10 Wm2 for Terra for data processed in the manner of the ERBE experiment. The new CERES data processing system was designed to reduce temporal sampling errors by using geostationary data to improve the interpolation of fluxes between the times of CERES observation. Pre-launch studies demonstrated that this new technique reduces instantaneous interpolation errors by over 50% for both longwave and shortwave fluxes relative to the ERBE processing. However, a final error analysis of the monthly mean fluxes that includes possible error sources from the geostationary data was not completed. This paper will focus on a new analysis of the temporal sampling error budget for CERES. The pre-launch study has been expanded to include a twelve months of ISCCP DX data covering all longitudes. New estimates of the global and seasonal error budget will be presented for the sampling patterns of each of the CERES satellites. Results suggest that the global sampling errors may have been overestimated by using only Western Hemispheric data during convectively active months. In addition, an expanded error analysis of monthly mean fluxes will be presented that includes errors in the calibration and narrowband-to-broadband conversion of the geostationary data used in the new CERES temporal interpolation algorithm.