Tools for Predicting Uncertainty and confidence Intervals in Radiometric Data Products

Abstract

Spaced-based observations of atmospheric energetics, such as those provided by NASA’s Clouds and the Earth’s Radiant Energy System (CERES), produce data products intended to be shared with the larger scientific community and merged with other complemental)’ data sets. Meaningful fusion of complementary data requires a well-founded common statistical basis for cited precision and accuracy. A high—level numerical model is available capable of predicting the dynamic opto-electrothermal behavior of CERES like radiometric channels. The paper reports use of this model to explore the sensitivity of data products to variations in individual optical, thermal and electronic parameters. The optical/thermal radiative part of the model is based on the Monte Carlos Ray-Trace (MCRT~ method in which millions of rays are traced. Several hours of execution time on a large computer are required to simulate a single scan across the Earth’s surface, thus making it impractical to run the simulation for every possible variation of each parameter. A key element of the research involves an effort to determine the minimum number of simulations required to produce statistically meaningful results.