The Study Of At-sensor Radiance Image Simulation In The Spectrum Of Solar Radiation

Image simulation is an essential technology wildly used to generate analog images based on the analysis of remote sensor imaging processing and the model developed to simulate this procedure. It also can be used to construct and evaluate a new remote sensing device in a virtual environment,assess the technological feasibility in an earth observation system after involving the new device. It also has some other advantages, such as providing much more data than the real system in the same condition of atmosphere and observation, which is a significant auxiliary method for remote sensing research. Therefore, image simulation is not only a very important tool for remote sensor designer, operator and algorithm developer, but also has considerable significance of remote sensing device design, algorithm research, and quantitative analysis.This thesis investigates the problem of "The study of at-sensor radiance image simulation in the spectrum of solar radiation". In respect to this target, the influence factors of at-sensor radiance in the spectrum of solar radiation has been explored and analyzed. In particular the key factors that have an effect on the atmospheric transmittance has been figured out, include vapor, aerosol, observation geometry and so on, which is an essential support for the study of complex surface scenes simulation, the creation of general atmosphere lookup tables, and the study of at-sensor radiance image simulation. On this basis, to address the problem that the shadow effect caused by complex mountainous topography and pixels mutual occlusion makes the effective illumination of each pixels fairly different and the surface reflectivity changing. The thesis starts with the surface reflectance simulation: first computes the surface scene shadow factor based on the ray tracing model POV-ray, and then combines it with high spatial resolution surface reflectance images through space mean resampling algorithm, and then employs the Spatial filtering method, to accomplish the low resolution reflectivity simulation under consideration of terrain impact. Through precision validation with the real surface reflectance, this approach has an error probability less than 1%.Subsequently, a general atmospheric lookup table has been created in order to cope with the problem that it is significantly time consuming to conduct a point-by-point radiance simulation based on the atmospheric radiative transfer model under the conditions that different pixels have distinct atmospheric composition and observation geometry. At present, there are several issues with the general atmospheric lookup tables, such as specific sensor dependence, insufficient consideration of atmospheric parameters, poor universal. After exploring numerous key factors like atmospheric parameter, observation geometry and radiative transfer path, the thesis takes a further step to consider a more reasonable configuration of threshold (Maximum and Minimum) and step size for each factor and to create a new general atmospheric lookup table, and then acquires the atmospheric parameters with multidimensional gravity Lagrange Interpolation algorithm. Finally, combined the reflectivity and the atmospheric parameters obtained from previous steps, to achieve the target of image simulation for at-sensor radiance in the spectrum of solar radiation. Through precision validation with the output of MODTRAN, this approach has an error probability no more than 7%.