“Clear water” is a scale-dependent concept, so it is more likely to successfully find the “clear waters” from images with smaller scale than that with larger scale data. In this study, an optimal spectral relationships of Moderate-resolution Imaging Spectroradiometer (MODIS) 250 m and 1 km resolution data at near-infrared bands (OSRLM) is constructed for converting pseudo “clear water” reflectance at 859nm to that at 748 and 869 nm. Due to the scale effects on satellite-observed pseudo “clear water” reflectance, the satellite-observed pseudo “clear water” reflectance is >5.18% larger than that of OSRLM model-derived. An atmospheric correction model for MODIS 1km data using pseudo “clear water” reflectance of MODIS 250m data (ACMM) was developed for improving the performance of traditional “clear water” atmospheric correction model (CWAC). The model validation results indicate that the ACMM model has a better performance than CWAC model. By comparison, using ACMM model decreases ~19.18% uncertainty from CWAC model in deriving water-leaving reflectance in Taihu Lake, China. The significantly reduced uncertainty in water-leaving reflectance estimation due to partly removal of scale effects on “clear water”. These findings imply that satellite-derived aerosol scattering contribution at smaller scale usually has better performance than that at larger scale.
