Fast wavelet radiative transfer model for inversion of IASI radiances

IASI is a Fourier Transform Interferometer that covers the spectral range 3.5-16 μm; it is being developed to meet the accuracy specifications issued by the Third Meteorological Organization for operational use in numerical weather prediction; the total amount of ozone and some information about its vertical distribution, fractional cloud cover and cloud top temperature/pressure are mission objectives as well. The main mathematical problem for retrieving the atmospheric profile of temperature and chemical concentration is the inversion of the radiative transfer equation (RTE). In Amato et al. (1999) (and references therein) an algorithm has been proposed for the RTE numerical inversion where the atmosphere is divided into N layers and the related geophysical parameters (temperature and chemical constituents) are represented by their average values. Even though the procedure has been proved effective in several circumstances, it is far from being used for (real-time) operational purposes (e.g., numerical weather prediction) mainly because of the computational time needed for its solution. The bottleneck is the computation of radiance and, in particular, of the Jacobian at every iteration. In Amato (1998) a quasi-analytical method was proposed for both that dramatically cuts computational time and contemporarily heavily increases the accuracy of the computations. However it relies on a line-by-line radiative transfer model for computing optical depths that is the current bottleneck of the method. The aim of the present paper is to introduce a method for the fast computation of radiances and the Jacobian