Improved Water Vapor Retrieval Algorithm for the SAGE III Experiment

The Stratospheric Aerosol and Gas Experiment III (SAGE III) scheduled for flight on Russian Spacecraft METEOR 3M-1 set for launch in December 2000 aims to retrieve the vertical profiles of water vapor concentration from cloud top to 50 km. Simulation studies have been conducted to investigate several approaches for the implementation of the inversion algorithm.

The purpose of this paper is to demonstrate the performance of the recently adopted least squares algorithm which employs a modification of the Levenberg-Marquardt approach. In contrast to the earlier algorithms, this new retrieval uses the measured slant path absorption at the 29 sub-channels (without removing the contributions of aerosol and ozone) as inputs to the iterative procedure. Results of simulated retrieval will be presented to illustrate the two major improvements accomplished with the non-linear least squares algorithm: (1) lesser sensitivity to potential systematic bias in the ozone cross sections, and (2) smaller effect due to variations in the level of aerosol loading.

In addition, formal error analysis for the retrieval will be discussed to allow quantitative evaluation of (1) average kernel, (2) weighting function, and (3) various error components.