Aerosol Retrievals from Individual AVHRR Channels. Part I: Retrieval Algorithm and Transition from Dave to 6S Radiative Transfer Model

Abstract

The present second-generation aerosol retrieval algorithm over oceans used at NOAA/National Environmental Satellite, Data, and Information Service (NESDIS) separately retrieves two values of aerosol optical depth, τ1 and τ2, from Advanced Very High Resolution Radiometer (AVHRR) channels 1 and 2 centered at λ1 = 0.63 (operational) and λ2 = 0.83 μm (experimental), respectively. From these, an effective Ångström exponent α, related to particle size, can be derived as α = −ln(τ1/τ2)/ln(λ1/λ2). The single-channel lookup tables, relating reflectance to optical depth in the retrievals, have been precalculated with the Dave (1973) scalar radiative transfer (RT) model. This first part of a two-part paper describes the retrieval algorithm, with emphasis on its RT modeling related elements, and documents the transition to the Second Simulation of the Satellite Signal in the Solar Spectrum (6S; 1997) RT model. The new 6S RT model has the capability to account for reflection from wind-roughened sea surface, offers a wide choice of flexible aerosol and gaseous absorption models, and allows easy convolution with the sensor's spectral response. The value of these new features for aerosol remote sensing from AVHRR is discussed in detail. The transition effect is quantified by directly applying the Dave- and 6S-based algorithms to four large datasets of NOAA-14 AVHRR measurements, collected between February 1998 and May 1999 over the latitudinal belt of 5°–25°S. Statistics of the differences (δτ = τDave − τ6S and δα = αDave − α6S) are as follows: averages − 〈δτ1〉 < 1 × 10−3, 〈δτ2〉 ≈ −4 × 10−3, and 〈δα〉 ≈ +8 × 10−2; and standard deviations are στ1 ∼ 6 × 10−3, στ2 ∼ 4 × 10−3, and σα ≈ 9 × 10−2. These are found to be well within a few percent of typical values of τ and α and their respective ranges of variability, thus ensuring a smooth transition and continuity in the operational aerosol retrieval. On the other hand, the 6S model provides a much more flexible RT modeling tool compared to the previously used Dave code.