Zhihua Mao

For addressing difficult detail extraction and low operating efficiency in monitoring sea ice in a large area with wide-field-of-view images from the Chinese Gaofen-1 satellite, a lightweight, high-precision sea ice segmentation network adaptive multistatistic fusion attention (AMFA) module using DeepLabV3+ as the base architecture (AMFA-DeepLab) is proposed. First, the module replaces the backbone network with a lightweight MobileNetV2 to ensure feature extraction capability and greatly reduce model computational complexity using inverted residuals and depthwise separable convolution. Second, to solve the problems of fragmented ice texture blurring and speckle noise interference in optical images, an AMFA is designed and introduced into the decoder side. This module innovatively integrates the global median pooling branch and adapts the recalibrated feature weight through a dynamic channel mixing mechanism, effectively enhancing the model’s capability of capturing fine sea ice edge features and its antinoise robustness in complex backgrounds. Experimental results based on the dataset from Liaodong Bay in the Bohai Sea of China show that the intersection over union of AMFA-DeepLab reaches 92.15% and the F1-score reaches 95.91%, increases of 3.06%, and 1.68%, respectively, compared with those of the baseline model. In addition, only 5.85 million model parameters are needed, the training time is shortened to 4.42 h, and the inference speed is 281.76 frames per second. Visualized analysis and generalization test further demonstrates that this model can accurately eliminate clutter interference from coastal land and seawater and extract the fine filamentous structure of drift ice in the scene of complex melting ice. This research overcomes the precision bottleneck while achieving an ultimate lightweight model, providing efficient technical support for operational dynamic monitoring of sea ice disasters based on Chinese GaoFen-1 satellites.

Sea Surface Skin Temperature (SSTskin) derived from satellites and its diurnal variation are crucial for climate research, yet conventional ocean models, which primarily solve for the foundation or bulk SST, are not designed to simulate the very thin skin layer temperature (SSTskin). Consequently, specialized parameterizations or coupled model components are often required to obtain SSTskin. This study aimed to capture SSTskin diurnal warming events and evaluate the performance of the improved one-dimensional mixed-layer model (PWP: Price-Weller-Pinkel) in simulating SSTskin. Using high-frequency Himawari-8 satellite observations, a typical diurnal warming event was detected in the coastal waters off northwestern Australia, with the maximum SSTskin diurnal variation reaching 3 °C. The reliability of Himawari-8 data was validated using iQuam in situ observations, showing a mean bias of −0.28 °C. The improved PWP model (incorporating an SSTskin parameterization scheme), forced by ERA5 datasets, was used to simulate SSTskin and its diurnal variation at 90 (0.25° × 0.25°) grid points. Results indicated that the PWP model reproduced the diurnal variation cycle consistently with observations, accurately matched regions with significant warming, and achieved a mean bias of −0.37 °C. However, in low-wind-speed areas (3 °C) occurred due to rapid thinning of the mixed layer and the absence of horizontal diffusion in this one-dimensional model. The improved PWP model, with its relatively stable SSTskin parameterization scheme, provides a computationally efficient tool for studying vertical processes in the upper ocean. Future work should evaluate vertical mixing schemes under low wind speed conditions to enhance the capability of numerical models to simulate SSTskin.

Water replenishment projects (WRP) are widely regarded by lake managers as an effective method for ecological restoration. But studies on multiple water receiving lakes indicate that multi-source water supplementation may exacerbate spatial heterogeneity in water quality. The primary causes are the uneven quality of external water sources and the morphology of the lakes themselves. Hangzhou’s West Lake, which relies on a single water source, also exhibits ecological differences between its lake areas. To explain this phenomenon from a hydrodynamic perspective, this study employs Delft3D model simulations to determine the spatiotemporal heterogeneity in the impact of artificial water replenishment on local water renewal times (LWRT) across different lake areas. The results indicate that, compared to natural conditions, WRP shortens the typical LWRT at most sites in West Lake and reduces the seasonal variation amplitude of LWRT. The WRP impedes local water exchange capacity in certain lake areas during specific months, potentially creating stagnant zones. Moreover, the spatiotemporal heterogeneity of its impact on West Lake’s hydrodynamic conditions amplifies differences in flow rhythms between lake regions, leading to fragmented water movement. This may ultimately compromise the integrity of the lake’s overall ecological environment. Managers of receiving lakes should consider increasing monitoring frequency and the number of monitoring stations to address more complex water quality conditions. Future designers of WRPs should incorporate the prevention of fragmented hydrodynamic conditions as an evaluation criterion in diversion scheme planning. This study provides references and evidence for lake ecological management and research on the impact mechanisms of WRPs on shallow lakes.

This paper addresses the fault-tolerant trajectory tracking control problem of Euler-Lagrange (EL) systems subject to actuator faults and component faults (including system uncertainties). A novel finite-time Second-Order Sliding Mode Observer with Nonlinear feedback (SOSMON) incorporating adaptive fault estimators is proposed. Unlike linear feedback or complex sliding mode designs, the proposed SOSMON’s nonlinear feedback terms effectively handle the unknown high-order nonlinear terms in EL systems. To reconstruct the two faults, an adaptively weighted Radial Basis Function Neural Network (RBF NN) technique is introduced to estimate component faults without prior knowledge of fault characteristics, and an Adaptive actuator Fault Estimator (AFE) is designed. Then, accurately estimated faults allow for the simple design of a Sliding Mode Fault Tolerance Controller (SMFTC). Finite-time convergence of the observer and closed-loop stability are proved via Lyapunov theory. Finally, comparative simulations on an Unmanned Surface Vessel (USV) demonstrate the superiority of the proposed method over conventional observers and the Back Propagation (BP) NN-based estimator regarding estimation accuracy and convergence speed.

Artificial water replenishment has improved the ecological environment of West Lake by introducing external clean water, but pollution issues still persist in some local regions. However, whether enhancing water exchange through internal water diversion within the lake can improve local water quality remains unverified. This study employed the Delft3D hydrodynamic model to simulate the spatiotemporal distribution of local water renewal time in West Lake, revealing that regions with prolonged water renewal times exhibited diminished resilience to water quality disturbances. This study utilized the Random Forest algorithm to determine the responsiveness of West Lake’s water transparency to parameters such as local water renewal time, and further discussed the impact of reducing local water renewal time on transparency under different water quality conditions. The results indicate that the sensitivity of West Lake’s transparency to water quality parameters closely resembles that of lakes with rainwater storage. The primary mechanism by which external water diversion improves transparency is through pollutant dilution, whereas enhanced local water exchange capacity contributes minimally to this effect. This conclusion demonstrates that localized internal water diversion within the lake is only suitable for preventing ecological issues such as regional eutrophication and algal blooms, but cannot effectively improve the overall lake ecosystem. Furthermore, this study identifies key factors affecting water transparency in artificially managed waters, highlighting priority monitoring indicators for similar water bodies. It also provides evidence to support research on aquatic optics and the development of remote sensing algorithms for such waters.

Nearshore bathymetry is critical for coastal management and ecology. While airborne hyperspectral remote sensing provides high-resolution image data, obtaining rapid and accurate bathymetric inversion in coastal areas lacking in situ data remains challenging. The widely used Hyperspectral Optimization Process Exemplar (HOPE) achieves high accuracy but suffers from computational inefficiency, making it impractical for large-scale, high-resolution datasets. By contrast, HOPE-Pure Water (HOPE-PW) offers computational efficiency but exhibits limitations in capturing fine-scale spatial patterns of bottom reflectance (ρ), and its applicability in transitional waters between Case I and II types requires further validation. Against this background, we employed machine learning-based substrate classification (support vector machine, random forest, maximum likelihood) in Wenchang coastal waters, China, to constrain ρ estimation in HOPE-PW, with validation using ICESat-2 data that extends its conventional application scenarios. Results demonstrate that when constrained by the optimal classifier (random forest), HOPE-PW achieves comparable accuracy to HOPE in shallow water while reducing runtime by 56% and memory usage by 68%. However, HOPE-PW exhibits slight underestimation in deeper areas, likely because simplification reduces sensitivity to water optical properties. Future research will focus on this issue. This study proposes an efficient and reliable framework for monitoring and evaluating water depth in areas lacking in situ data, offering a practical solution for integrated coastal zone management.

Sea surface temperature (SST) diurnal variation is an important part of air-sea energy exchange, and its response to the Madden–Julian Oscillation (MJO) also helps in studying the development mechanism of the MJO. This study investigates SST diurnal variation in the North Indian Ocean by employing hourly data from the FY-4A geostationary satellite, in conjunction with reanalysis products and the MJO index. Validation against in situ SST measurements shows a bias of –0.33°C, an absolute bias of 0.73°C, an RSD of 0.33°C, an RMSE of 0.94°C, and a correlation coefficient of 0.90. The results demonstrate consistent accuracy throughout the diurnal cycle. Diurnal warming is most pronounced in the northern Arabian Sea and northern Bay of Bengal, especially from March to May prior to the monsoon onset. Wind speed and shortwave flux emerge as key drivers, although wind speed exerts a stronger regional influence on diurnal warming. Furthermore, the SST diurnal variation responds to the MJO within 1–2 days, as its active phase increases cloud cover and intensifies wind, thereby suppressing daytime warming. Despite the high temporal resolution of FY-4A, which allows for detailed sub-daily observations, monsoonal circulation and cloud cover can reduce data availability, complicating quantitative studies of SST variability and MJO feedbacks. Therefore, integrating multiple data sources is crucial for comprehensive analyses. Overall, these findings underscore the potential of geostationary satellites for monitoring diurnal SST fluctuations and emphasize the necessity of accounting for complex atmospheric–oceanic interactions in the North Indian Ocean when examining the role of diurnal variation in broader climate processes.

Lidar technology is increasingly applied to the inversion of oceanic biological parameters and optical properties based on empirical and semi-empirical bio-optical models. However, these models cannot be directly applied to inland waters due to their complex composition, and research on the biological parameters and optical properties of inland waters remains limited. In this study, the Fernald method was employed to retrieve the vertical distribution of optical properties in Qiandao Lake for the first time using shipborne lidar data obtained in June 2019. By quantifying the depth-resolved optical contributions of biological components, the vertical distributions of chlorophyll-a concentration were mapped with greater precision. The lidar-estimated optical properties exhibited characteristic spatiotemporal distributions, which were closely related to water quality. At the inflow of Xin’an River, the attenuation and scattering coefficient showed a gradual increase with depth. At the north–south-oriented reservoir area and the outflow of Qiandao Lake, an apparently continuous subsurface layer with the maximum signal occurred at approximately 3.5 m. The vertical distributions of chlorophyll-a profiles were consistently classified as subsurface chlorophyll maxima, with the maximum value of chlorophyll-a concentration fluctuating between 4 and 12 μg/L. The subsurface phytoplankton layer was observed at water depths ranging from 1.5 to 3.5 m, with a thickness of 3 to 6 m. Furthermore, the influences of lidar ratio S_p(z) and reference value b_bp(z_m) were discussed as significant sources of inversion error in the Fernald method. These results indicate that lidar technology holds great potential for the long-term monitoring of lakes.

Airborne bathymetry LiDAR, such as Mapper5000, experienced varied water-bottom return intensities. Under the high bottom return signal condition, owing to the clear water body, high bottom reflectance, low depth, or high gain setting of the detector, the peak power of the received signal from the bottom often exceeds the linear dynamic range of the 532 nm detector assembly. The resulting signal saturation distorted the recorded bottom return waveforms and resulted in undesired depth bias exceeding 2.0 m in certain extreme cases. In this study, we developed a low-complexity peak position restoration method for this “saturation in-water range bias” correction based on an intersection point of double linear approximations of the leading and trailing edges of the saturated bottom waveform (DLI). The validation of corrected results with consistent sonar depth measurements indicated that the DLI method demonstrated considerable robustness to the effect of pulse stretching under saturation conditions and reduced the mean absolute depth bias from 1.46 m to 0.47 m. Compared with Gaussian and exponentially modified Gaussian fitting algorithms, the DLI was faster and the most effective at reducing depth bias under strong saturation conditions. The success of DLI benefited from the small width of the transmitted laser pulse and the sharp slope of the leading edge of the saturated bottom waveform. However, the interference of volume backscattering signals at shallow depths could affect the performance of DLI, which should be carefully removed.