An Integrated Correction Method for Geometric, Atmospheric, and Terrain Effects in Hyperspectral Remote Sensing Images

Reflectance inversion is a critical challenge in quantitative hyperspectral remote sensing, especially in rugged terrain. The under-/over-correction of reflectance caused by such terrain and its impacts on subsequent information extraction have not been well quantified or effectively eliminated. To improve surface spectral reflectance inversion under varying illumination, atmospheric, and terrain conditions, an integrated correction method for geometric, atmospheric, and terrain effects (ICGAT) is proposed. The method incorporates a coupled radiative transfer model that accounts for direct solar irradiance, diffuse sky irradiance, surface-atmosphere coupling irradiance, and irradiance from neighboring slopes along the sun–terrain-target path. The geometric positioning and irradiance components are first computed, followed by an iterative retrieval of reflectance. Geometric resampling is subsequently applied to the reflectance, and an uncertainty quantification is finally conducted in the process. Tests show that ICGAT significantly improves spectral consistency between neighboring shaded and sunny slopes, reducing the influence of terrain relief on reflectance, as well as retrieving high-accuracy surface reflectance in flat areas. Meanwhile, the uncertainty propagation analysis could reveal not only the fidelity of the retrieved reflectance but also the major sources of uncertainty.