Using the near-space platform to conduct radiometric calibrations of ocean color sensors is a promising method for refining calibration precision, but there is knowledge gap about the radiance contributions above near-space over the open ocean. We used the radiative transfer (RT) model (PCOART) to assess the contributions (L_R) of the upwelling radiance received at the near-space balloons to the total radiance (L_t) measured at the top of the atmosphere (TOA). The results indicated that the L_R displayed distinct geometric dependencies with exceeding 2% across most observation geometries. Moreover, the L_R increased with wavelengths under the various solar zenith angles, and the L_R values fell below 1% only for the two near-infrared bands. Additionally, the influences of variations in oceanic constituents on L_R were negligible across various azimuth angles and spectral bands, except in nonalgal particle (NAP)-dominated waters. Furthermore, the influences of aerosol optical thicknesses (AOTs) and atmospheric vertical distributions on L_R were examined. Outside glint-contaminated areas, the atmosphere-associated L_R variations could exceed 2% but declined substantially as AOTs increased under most observation geometries. The mean height of the vertically inhomogeneous layer (h_m) significantly influenced L_R, and the differences in L_t could exceed 5% when comparing atmospheric vertical distributions following homogeneous versus Gaussian-like distributions. Finally, the transformability from near-space radiance to L_t was examined based on a multiple layer perceptron (MLP) model, which exhibited high agreement with the RT simulations. The MAPD averaged 0.420% across the eight bands, ranging from 0.218% to 0.497%. Overall, the radiometric calibration utilizing near-space represents a significant innovation method for satellite-borne ocean color sensors.

Reliable multimodal image analysis is challenging when complementary sensors exhibit class-dependent strengths and local prediction uncertainty. This study presents a reliability-aware hybrid decision-fusion framework for fine-grained tidal-flat land-cover mapping as a demanding application of multisource image processing. A pyramid scene parsing network (PSPNet) with a 50-layer residual network (ResNet-50) backbone extracts multi-scale spatial context from high-resolution multispectral patches, while a Random Forest models 166-dimensional hyperspectral reflectance signatures. Instead of applying uniform score averaging, the framework combines validation-derived class reliability, pixel-level confidence, and branch disagreement to preserve trusted spatial predictions and selectively introduce spectral correction in uncertain regions; morphological refinement improves local mask continuity. Experiments used GF-2, GF-7, ZY1-02E/F, and ground-reference data acquired from 2022 to 2024, comprising 23 classes, 12,666 image samples, and 452,156 spectral samples. The method achieved 94.64% overall accuracy, 92.85% mean class recall (MCR), 85.47% mean intersection over union (mIoU), 91.72% macro-averaged F1 score (Macro-F1), and 85.88% Small-object F1. Relative to the PSPNet and Random Forest branches, mIoU increased by 4.73 and 3.84 percentage points, while Small-object F1 increased by 5.36 and 6.92 points. The results demonstrate that reliability-aware hybrid fusion can improve both spectrally ambiguous surface classification and spatially fragmented object recognition in complex multimodal scenes.