BRDF Correction in Landsat Directional Reflectance Data by Combining Optimal BRDF Archetype With Spectral Characteristics

The Landsat optical images have provided a unique long-term record of the terrestrial surface and have always been an important data source for earth observation. However, due to variable solar and view geometries, Landsat’s observations have directional reflectance effects over non-Lambertian surfaces across all spectral bands, known as bidirectional reflectance distribution function (BRDF) effects, which will become a significant source of noise for fine applications. In this study, we developed the Combining Optimal BRDF Archetype with Spectral Characteristics (COBASC) method to effectively reduce BRDF effects in Landsat directional reflectance data. COBASC builds a high-accuracy BRDF parameter lookup table (BRDF LUT) for Landsat pixels by determining the optimal BRDF archetypes of Landsat pixel samples and combining them with the spectral characteristics of the pixel samples, considering landscape heterogeneity at the Landsat scale. Based on the BRDF LUT built, high-accuracy BRDF parameters for Landsat pixels are obtained to implement BRDF correction. Experimental results show that under varying degrees of BRDF effects in Landsat observations, the proposed method consistently and effectively reduces the directional reflectance differences caused by the BRDF. Moreover, the accuracy of the BRDF-corrected results is higher than that achieved by current mainstream methods, thereby demonstrating the superiority of the proposed method in correcting BRDF effects in Landsat directional reflectance data. The COBASC has the potential to be used flexibly for BRDF correction on other medium- to high-spatial-resolution satellite remote sensing images.