Fast Monte Carlo-assisted simulation of hyperspectral earth backgrounds

A physics-based calculation method has been developed for rapidly synthesizing radiometrically accurate spectral imagery of the Earth incorporating 3-D cloud effects. The method combines measured surface reflectance imagery with 3-D cloud radiance calculations and 1-D atmospheric radiation transport. Cloud property images are pre-calculated with the MCScene Monte Carlo code [1]; a data fusion-based method is used to speed convergence. These images are combined on-the-fly with surface and upper atmospheric descriptions using a radiation transport equation incorporating atmospheric spatial dependence. The method enables arbitrary combinations of sensor and sun location, cloud field and surface to be assembled on-the-fly and provides hyperspectral resolution with minimal computational effort. The simulations have a realistic appearance and agree well with much more time-consuming direct MCScene calculations.