Sensitivity analysis of atmospheric compensation algorithms for multispectral systems configuration

This study evaluates a series of atmospheric correction techniques developed at RIT called Total Inversion. The ability to convert remotely sensed image data to physically meaningful scientific units, such as surface reflectance, has been demonstrated f hyperspectral systems. This capability, however, has not been proven with the use of multispectral satellite-based remote sensing systems. The goal of this study is to determine the feasibility of adapting the Total Inversion techniques for multispectral s understanding the capabilities and limitations of these techniques for operational use. This means that the algorithmic process being used must be image based, have practical run times, require little or no user intervention and produce consistent results within acceptable error tolerances. Three tasks were performed to study the feasibility of using Total Inversion for multispectral sensors. Task one evaluated the potential for using a pre-built set of lookup tables (LUTs) for use with the radiative transfer based spectral ma atmospheric correction methods. Task two is a sensitivity analysis for using independent ancillary estimates for elevation and water vapor inputs. Task three of this study focused on the comparison of two algorithms for the estimation of aerosol visibility. These included the regression intersection method (RIM) for spectral fitting and the non-linear least squares spectral fit method (NLLSSF). For all these tasks the study ut existing image data and ground truth to enable evaluation and demonstration of quantitative performance of various approaches specifics and rationales of these tasks are covered in the Project details section.