Radiative Transfer in highly variable cloud scenarios

Abstract

New technologies permit going continuously down into the spatial scale when observing the Earth's atmosphere from spaceborne instruments. At the same time, new computers are equipped with large main memories that allow to allocate bigger data arrays of optical properties. Historically, strongly approximative approaches have being used for the computation of the radiative transfer in remote sensing and climate/weather modeling applications. The main reasons are: On the one hand, to keep the computational time and memory save under some required limit, and on the other hand, to come up with easy-to-use analytical expressions that simplify the inversion process. The new generation of applications will have to deal with more complexity in optical fields and with the general solution of the radiative transfer in order to be able to reproduce the real radiance and flux fields. Within this scenario, we present results of the "exact" radiative transfer through high spatially resolved stochastically generated cloud fields embedded in realistic atmospheres. For this purpose, a versatile three-dimensional radiative transfer Monte Carlo code (MoCaRT) together with a robust cloud generator (IAAFT) able to add subscale variability have been used. The "exact" results are compared with those obtained using restrictive methods such as plane parallel and independent pixel approximations. The differences and deficiencies of these approximations are pointed out.