Cloud Identification for ERBE Radiative Flux Retrieval

Abstract

Derivation of top of atmosphere radiative fluxes requires the use of measured satellite radiances and assumptions about the anisotropy of the Earth's radiation field. The primary modification of the Earth's anisotropy is caused by variations in cloud properties. These variations occur rapidly in space and time and provide a challenge for the accurate derivation of radiative flux estimates. The present paper discusses the application of a maximum likelihood estimation (MLE) technique to the problem of cloud determination for coarse resolution broadband satellite data. This methodology is developed in concert with new empirical models of the angular dependence of radiance, and is tested against simulated satellite observations. It is argued that the new angular dependence models are a more complete description of the Earth's radiation field than any previously available models. When used to determine cloud conditions for the inversion of satellite-measured radiances to fluxes, simulations predict that the MLE approach gives substantial improvements over both a Lambertian Earth assumption and the clear/cloud threshold used in the inversion of Nimbus 3 and Nimbus 7 Earth Radiation Budget scanner data. The MLE methodology will be used in the operational processing of the Earth Radiation Budget Experiment (ERBE) scanner data. The present paper serves to document both the philosophy and the form of the MLE methodology. Validation studies using both ERBE and Nimbus 7 radiation budget data will be the subject of future papers by several ERBE Science Team investigators.