Advanced signal processing techniques for the analysis of solar radiometer data in the presence of temporally varying aerosol optical depths

Abstract

Solar radiometers, which are used for remote sensing of atmospheric aerosols and absorbing gases, have traditionally been calibrated by the Langley method. Temporally variable conditions, however, can significantly bias the zero-airmass intercepts obtained by this method. In this dissertation, a number of new signal processing techniques are developed to better characterize aerosol variability and use it to obtain improved intercepts under a broad range of conditions. The techniques include (1) an extension of Forgan's method, using correlation between optical depths at different wavelengths to model temporal variations; (2) spectral/fractal analysis and filtering to identify systematic atmospheric variations and distinguish them from noise; and (3) error correction using correlation between results from different data sets. These techniques, along with some preliminary adjustments and an algorithm for estimating ozone content, are incorporated into an iterative processing scheme that both calibrates the instrument and provides improved estimates of each optically significant atmospheric constituent. Finally, the characterization of aerosol variability is further enhanced by analyzing data taken with a customized radiometer that measures diffuse skylight as well as direct sunlight.