Simulation of hyperspectral radiance images with quantification of adjacency effects over rugged scenes

Abstract

Simulation of hyperspectral remote sensing images can be applied to the generation of simulated datasets and validation of the data processing algorithms, as well as to deriving the ground reflectance of rugged areas from remote sensing images. Realistic simulations require capabilities dealing with rugged scenes and accurate modeling of the adjacency effects. Estimation of adjacency effects remains a difficulty because of the complex phenomena induced by rugged terrains. To find the optimal tradeoff between the accuracy and the computation time, this paper describes a method of simulating hyperspectral radiance images over rugged scenes with quantitative estimation of the adjacency effects under the assumptions of Lambertian surface and single scattering. In the quantification of adjacency effects, molecular/aerosol scattering phase functions, topographic features and ground heterogeneity are taken into account. Radiance images are generated by several sequential processing steps, which create atmospheric effects using a radiative transfer model, model topographic effects based on digital elevation model and simulate the instrument response function of a given sensor. The method has been applied to the complete scene simulation and the validity of various assumptions is checked using experimental datasets from Hyperion.