Unique Perturbation Methods Exploitation for Semi-Supervised Remote Sensing Image Semantic Segmentation

To effectively tackle the challenges posed by spatial-scale differences and spatial heterogeneity, this paper presents a distinctive coarse pixel-scale ground “truth" dataset by upscaling sparsely distributed in situ measurements. This dataset is a valuable resource for validating and correcting global surface albedo products, enhancing reference data accuracy by 6.04 %. Remarkably, it substantially enhances 17.09 % in regions with strong spatial heterogeneity.