Modeling polarized radiances toward the development of an aerosol retrieval method
Date
2005-10
Authors
Stephens, Graeme L., 1952-, author
Lebsock, Matthew, author
Department of Atmospheric Science, Colorado State University, publisher
Journal Title
Journal ISSN
Volume Title
Abstract
Polarized radiances reflected from aerosol laden atmospheres were modeled. An atmospheric aerosol model was defined which corresponds to a clean oceanic environment composed mainly of sulfate particles. This model specifies an aerosol size distribution and optical properties. Fifteen atmospheric scenes composed of varying solar zenith angles and aerosol optical depths were defined to explore the scene dependent nature of the top of the atmosphere total and polarized radiances. The sensitivity of the forward model to aerosol optical depth was examined for these fifteen cases. A comprehensive error assessment was also performed for each of the fifteen cases. This assessment included the explicit modeling of errors due to aerosol model assumptions. The sensitivities were combined with the error estimates to produce signal to noise ratios for each scene. The signal to noise ratio demonstrates a significant viewing angle dependence for all of the fifteen cases. It was found that the angles with the largest signal to noise ratio for the total radiance are in the backscatter direction while angles in the sun glint demonstrated the lowest signal to noise ratio. It was also observed that the model sensitivity tends to decrease with optical depth while errors are shown to increase with optical. Consequently, the signal to noise ratio tends to decrease strongly with optical depth. Finally, it was found that the signal to noise ratio for the total radiances tends to be about three times as large as the polarized radiances. The total error estimates are used to develop an optimal estimation two-channel optical depth retrieval. Due to the greater signal to noise ratio of the total radiances, the polarized radiances were not used in the retrieval. Synthetic data were created to test the retrieval functionality. Two sources of bias were demonstrated. First, an a priori bias was shown which biases the retrieval towards the a priori initial guess. A second source of bias is introduced through the necessity to assume an aerosol model. It is demonstrated that these assumptions may bias the retrieval either high or low. Viewing geometries with small signal to noise ratios are shown to have a larger bias than those with large signal to noise ratio. It was concluded that the ideal multi-angular retrieval will utilize viewing geometries with large signal to noise ratios to limit the degree of the above biases. Finally, the retrieval is applied to a small sampling of POLDER II radiance data. The retrieved optical depths tend to be in qualitatively good agreement with the POLDER optical depths.
Description
October, 2005.
Includes bibliographical references (pages 107-109).
Includes bibliographical references (pages 107-109).
Rights Access
Subject
Atmospheric aerosols