This paper presents a model to describe the polarization patterns of celestial light, which includes sunlight and skylight, when refracted by wavy water surfaces. The polarization patterns and intensity distribution of refracted light through the wave water surface were calculated. The model was validated by underwater experimental measurements. The experimental and theoretical values agree well qualitatively. This work provides a quantitative description of the repolarization and transmittance of celestial light transmitted through wave water surfaces. The effects of wind speed and incident sources on the underwater refraction polarization patterns are discussed. Scattering skylight dominates the polarization patterns while direct solar light is the dominant source of the intensity of the underwater light field. Wind speed has an influence on disturbing the patterns under water.

Optical remote sensing of aquatic vegetation in shallow water is an essential aid to ecosystem protection, but it is difficult because the spectral characteristics of the vegetation are sensitive to external features such as water background effects, atmospheric effects, and the structural properties of the canopy. A global sensitivity analysis of an aquatic vegetation radiative transfer model provides invaluable background for algorithm development for use in optical remote sensing. Here, we use the extended Fourier Amplitude Sensitivity Test (EFAST) method for the modelling. Four different cases were identified by subdividing the ranges of water depth and leaf area index (LAI) involved. The results indicate that the reflectance of emergent vegetation is affected mainly by the concentrations of chlorophyll a + b in leaves (Cab), leaf inclination distribution function parameter (LIDFa) and LAI. The parameter LAI is influential in sparse vegetation cases whereas Cab and LIDFa are influential in dense vegetation cases. Canopy reflectance for submerged vegetation is dominated by water parameters. Relatively, LAI and Cab are highly sensitive vegetation parameters. The analysis is extended to vegetation index as well, which takes the Sentinel-2A as the reference sensor. It shows that NDAVI (Normalized Difference Aquatic Vegetation Index) is suitable for retrieving LAI in all cases except deep-sparse for emergent vegetation, whereas NDVI (Normalized Difference Vegetation Index) would be better in the deep-sparse case. NDVI, NDAVI and WAVI (Water Adjusted Vegetation Index), respectively, are suitable for retrieving Cab, Car and LIDFa in dense cases. For submerged vegetation, the sensitivity of LAI to NDAVI is relatively high only in the shallow-sparse case. The adjustment factor L in SAVI and WAVI fails to suppress the sensitivity to water constituent parameters. The sensitivity of LAI and Cab to NDVI in deep cases is relatively higher than that to the other indices, which may provide clues for the construction of inversion algorithms in macrophyte remote sensing in the aquatic environment using spectral signatures in the visible and near infrared regions.

The integration of the Crop Growth Model (CGM), Radiative Transfer Model (RTM), and Machine Learning Algorithm (MLA) for estimating crop traits represents a cutting-edge area of research. This integration requires in-depth study to address RTM limitations, particularly of similar spectral responses from multiple input combinations. This study proposes the integration of CGM and RTM for crop trait retrieval and evaluates the performance of CGM output-based RTM spectra generation for multiple crop traits estimation without biased sampling using machine learning models. Moreover, PROSAIL spectra as training against Harmonized Landsat Sentinel-2 (HLS) as testing was also compared with HLS data only as an alternative. It was found that satellite data (HLS, 80:20) not only consistently performed better, but PROSAIL (train) and HLS (test) also had satisfactory results for multiple crop traits from uniform training samples in spite of differences in simulated and real data. PROSAIL-HLS has an RMSE of 0.67 for leaf area index (LAI), 5.66 µg/cm² for chlorophyll ab (Cab), 0.0003 g/cm² for dry matter content (Cm), and 0.002 g/cm² for leaf water content (Cw) against the HLS only, with an RMSE of 0.40 for LAI, 3.28 µg/cm² for Cab, 0.0002 g/cm² for Cm, and 0.001 g/cm² for Cw. Optimized machine learning models, namely Extreme Gradient Boost (XGBoost) for LAI, Support Vector Machine (SVM) for Cab, and Random Forest (RF) for Cm and Cw, were deployed for temporal mapping of traits to be used for wheat productivity enhancement.

Infrared small target detection has long faced significant challenges due to the extremely small size of targets, low contrast, and the scarcity of annotated data. To tackle these issues, we propose a wavelet-recalibrated semi-supervised network (WRSSNet) that integrates synthetic data augmentation, feature reconstruction, and semi-supervised learning, aiming to fully exploit the potential of unlabeled infrared images under limited supervision. We construct a dataset containing 843 visible-light small target images and employ an improved CycleGAN model to convert them into high-quality pseudo-infrared images, effectively expanding the scale of training data for infrared small target detection. In addition, we design a lightweight wavelet-enhanced channel recalibration and fusion (WECRF) module, which integrates wavelet decomposition with both channel and spatial attention mechanisms. This module enables adaptive reweighting and efficient fusion of multi-scale features, highlighting high-frequency details and weak target responses. Extensive experiments on two public infrared small target datasets, NUAA-SIRST and IRSTD-1K, demonstrate that WRSSNet achieves superior detection accuracy and lower false alarm rates compared to several state-of-the-art methods, while maintaining low computational complexity.

Inversion of aerosol optical depth (AOD) over bright land surface by optical remote sensing is particularly challenging because surface reflectance dominates the satellite signal. A structure function method is suitable and can effectively solve aerosol optical depth inversion in high reflectance areas. How to select d (the pixel distance between two pixels) value is one of the key problems with the structure function method. We present a method based on geostatistics, where variogram theory is referenced, to determine the pixel distance in structure function model for aerosol optical depth inversion. This method was validated by the Moderate Resolution Imaging Spectroradiometer (MODIS) 1 km resolution level 1B data from Beijing, China. The results indicate that the relationship between variogram and d in four different directions can be fitted by exponential function, of which correlation coefficients are all above 0.9. Compared with the MODIS aerosol product (MOD 04 product), the inversion AOD has higher accuracy, with an absolute error of −0.00187 instead of −0.00854 and a relative error of 0.99% instead of 4.35%, based on AERosol RObotic NETwork (AERONET) observations. For validation, we applied this new method separately to both Beijing–Tianjin–Hebei region and MODIS 500 m resolution images for region and resolution validation. The results show that inversion AOD has higher accuracy and more efficient pixels.