DART: a 3D model for simulating satellite images and studying surface radiation budget

DART (Discrete Anisotropic Radiative Transfer) is a radiative transfer model that simulates remotely acquired images. It was originally developed to work in the short wavelengths (0.3-3 mu m) within 3D natural scenes that are represented as matrices of rectangular cells containing trees, shrubs, grass, soil, etc. DART was recently modified to extend its domain of application and to improve its accuracy. This paper summarizes the major features of DART and presents the changes that were implemented for improving its accuracy. Presently, this model works with natural and urban landscapes, on the whole optical domain (thermal infrared included) and with a multispectral approach that uses optical data bases from 0.3 mu m up to 15 mu m. It simulates radiative transfer in the whole 'atmosphere-Earth' system and it accounts for the instrumental transfer function. Three major changes allowed us to improve DART accuracy by a factor of three: more accurate simulation of single and multiple scattering, use of a scheme that oversamples DART cells and a better account of the direction of radiation that gives rise to multiple scattered radiation.