Snow grain size retrieval method based on radiative transfer model

Although the Asymptotic Radiative Transfer (ART) model has been widely applied in various algorithms for retrieving snow grain size, most existing methods are limited to pure snow pixels and fail to account for the influence of mixed pixels in large-scale regions. This study integrates the ART model with a mixed-pixel decomposition algorithm, incorporating the impact of snow grain size variation, and proposes a novel mixed-pixel decomposition approach aimed at simultaneously retrieving snow grain size and snow-covered area. To enhance computational efficiency, prior knowledge and the Stochastic Gradient Descent (SGD) algorithm were introduced into the unmixing process. Experimental results demonstrate that the proposed algorithm achieves higher accuracy in retrieving snow grain size and snow-covered area compared to existing methods, thereby enabling large-scale snow grain size retrieval using remote sensing technology.