Jie Guo

Hypoxia has become increasingly frequent in the semi-enclosed Bohai Sea since the early 2000s, posing significant risks to marine ecosystems. To address the limitations of existing dissolved oxygen models—particularly their poor predictive ability and lack of interpretability—we developed a two-month lead probabilistic forecasting framework for summer hypoxia using only multi-source remote sensing and reanalysis data, supplemented by in situ observations for validation. Environmental conditions in June were used to predict hypoxia probability in August via machine learning; among the seven algorithms tested, the optimized Random Forest model achieved the best performance (F1 = 0.76 and AUC = 0.92 on the independent test set). The model successfully reproduced observed hypoxia patterns in 2019 (validated against numerical simulations) and 2022 (validated against field measurements), capturing an increase in hypoxic area from 8229 km² to 13,866 km², which is consistent with intensifying thermal stratification under climate warming. SHAP-based interpretability analysis identified reduced wind speed and enhanced thermal stratification as the dominant physical drivers, highlighting the critical role of suppressed vertical mixing in limiting bottom-water oxygen supply. This study demonstrates that a physics-informed, interpretable machine learning approach based solely on satellite and reanalysis data can deliver reliable, early, and physically consistent hypoxia forecasts, offering a scalable solution for environmental monitoring of data-limited coastal seas.

Seasonal hypoxia in bottom waters of the Bohai Sea poses an escalating threat to marine ecosystems, yet monitoring it via satellite remote sensing continues to be challenging due to the inaccessibility of bottom layers. However, surface bio-optical signals do not instantaneously reflect variation in bottom-water dissolved oxygen (DO); instead, a distinct temporal lag exists between surface biological activity and its influence on bottom DO. Leveraging this insight, an inversion framework was established, integrating multi-source remote sensing data with decision tree-based machine learning models to estimate bottom-water DO concentration. We evaluated multiple lag intervals for satellite-derived bio-optical variables and adopted a 14-day lag as representative of the delayed impact of surface processes on bottom DO. An optimized feature set selected via a genetic algorithm (GA) was used to train the XGBoost model, which achieved high predictive performance (R² = 0.86, RMSE = 0.79 mg/L, MAPE = 8.89%). Interpretability analysis identified the sea surface temperature as the dominant driver of bottom-water DO variation in the Bohai Sea. The framework successfully reproduced the spatiotemporal variability in bottom DO from 2022 to 2024 in the Bohai Sea and captured the locations of summer hypoxic zones. Further analysis demonstrated that incorporating physically based bottom-layer variables substantially enhances model accuracy (R² = 0.89, RMSE = 0.68 mg/L, MAPE = 7.85%), underscoring their critical role in regulating bottom-water DO concentrations. Building on the established inversion framework and integrating extended in situ and satellite observations, we reconstruct the long-term temporal distribution of bottom DO in the Bohai Sea from 2014 to 2025, revealing the considerable potential of satellite data for monitoring bottom-water DO conditions in coastal seas.

Freshwater ice cover on inland water bodies is a sensitive indicator of climate change, and its monitoring is hampered by the decline of in situ observation networks since the 1980s. Previous applications of GNSS-Reflectometry (GNSS-R) have focused mainly on sea ice detection, while its capability for seasonal freshwater ice monitoring has remained insufficiently explored. The aim of this study is to assess the sensitivity of spaceborne L-band bistatic measurements to the processes of freshwater ice cover formation, growth, and break-up on large inland lakes, and to compare the scattering behavior with that of sea ice. A full annual cycle (September 2023 – August 2024) was analyzed using data from the GNOS-II instrument onboard the Feng-Yun 3E (FY-3E) satellite at two test sites: Great Slave Lake (Canada) for freshwater ice and the Sea of Okhotsk for sea ice. Statistical moments of the Doppler spectrum – width and kurtosis coefficient – were computed from delay-Doppler maps. The results were validated against meteorological records from the Yellowknife station, MODIS imagery, and ERA5 ice thickness reanalysis. A monotonic decrease in Doppler spectrum width with decreasing air temperature was established for freshwater ice, consistent with a reduction of effective surface roughness. For freshwater ice, peak power and kurtosis coefficient show a distinct dependence on ice thickness, reflecting signal attenuation within the ice volume and dominant reflection at the ice-water interface; for sea ice these parameters remain stable, since reflection occurs at the air-ice boundary owing to high brine salinity. The findings demonstrate that L-band bistatic reflectometry provides an effective all-weather tool for operational monitoring of freshwater and sea ice, including detection of freeze-up and break-up phases, ice thickness estimation, and support for Arctic navigation.

The central Bohai Sea (CBS) is the distribution center and wintering grounds for economically important species of fish, shrimp, and crabs migrating from the Yellow Sea and the BS. However, the frequency of hypoxia in the CBS has gradually increased, posing a threat to its ecology. Therefore, we analyzed data from an on-site investigation of the cold-water mass coverage area in the southern part of the BS in the spring, summer, and autumn of 2022. We investigated the characteristics of seasonal variation in water quality parameter, the main characteristics and leading factors affecting the distribution of bottom hypoxia using stratification data and the Nutritional Status Quality Index. The “boot-shaped” distribution of hypoxia in summer was primarily the result of the intrusion of cold and highly saline water from the northern part in the study area, as well as the intrusion of high-temperature and low-salinity water from the Yellow River estuary (YRE) and the high-salinity water in the northeast corner of the study area, which had altered the stratification effect of the region. This is also the main reason that affects the accuracy of the prediction for occurrence of hypoxia stations in summer. The results show that the cold-water mass in the northern part of the Bohai Sea invades the cold-water mass in the southern part in summer 2022. Thus, this study provides novel insights into the formation and distribution of hypoxia in the CBS.

The central Bohai Sea (CBS) serves as a distribution center and wintering ground for the migration of economically important species of fish, shrimp, and crabs from the Yellow Sea and the BS. The frequency of hypoxia has gradually increased, posing a threat to the ecology of the CBS. Data from an on-site investigation of the cold water mass coverage area in the southern part of the BS in the spring, summer, and autumn of 2022 were an-alyzed to determine the relevant factors using stratification data and the nutritional status quality index. The results indicated that stratification was the leading cause of hypoxia. The 'boot-shaped' distribution of hypoxia in summer was primarily the result of the intrusion of cold and highly saline water in the northern part of the study area, as well as the intrusion of high-temperature and low-salinity water from the Yellow River estuary and the high salinity water in the northeast corner of the study area. The study found that the cold water mass in the northern part of the Bohai Sea invades the cold water mass in the southern part. This study provides novel insights into the formation and distribution of hypoxia in the CBS.

Recurrent algal blooms in coastal regions have become a pressing environmental concern, significantly impacting the ecological integrity of the Bohai Sea due to their sudden occurrence and rapid development. Addressing this issue neces sitates high-resolution and continuous remote sensing methods for dynamic monitoring. This study confronts two challenges: the inconsistency in band configurations and spectral widths across multi-source satellite sensors when generating long-term obser vation datasets, and the limited applicability and low accuracy of traditional empirical models for Chlorophyll-a (Chl-a) retrieval under complex optical conditions. To overcome these limitations, a harmonized long-term dataset was developed using the consis tent spectral response characteristics of geostationary ocean color imager (GOCI)-I and GOCI-II sensors. In addition, a Chl-a inver sion framework was proposed by integrating a Quasi-analytical algorithm version 6 (QAA-v6)-optimized physics-informed neural network (QPINN) with a spatiotemporal attention mechanism. A weighted transformation approach was employed to standardize reflectance data from both sensors, enabling a unified dataset with consistent spectral and spatial parameters. Spatiotemporal features were extracted using a Transformer architecture and combined with QPINN to suppress anomalies, enforce physical constraints, and improve model generalization. The resulting QPINN-Transformer model achieved a high inversion accuracy, with a coefficient of determination R² = 0.91,MAE = 0.3750. Long term application of the model from 2014 to 2020 revealed a strong spatial association between Chl-a concentrations and the central hypoxic zone in the Bohai Sea. These findings offer methodolog ical advancements for the dynamic evaluation of eutrophication, inform land-based pollution mitigation strategies, and support assessments of marine carbon sequestration potential.

Chlorophyll-a (Chl-a) is essential for assessing water quality and aquatic ecosystems. Traditional in situ monitoring suffers from sparse stations and limited samples, and machine learning often overfits under such conditions. Transfer learning offers a solution by leveraging pretrained knowledge, but domain discrepancies remain a major challenge. To improve cross-domain Chl-a inversion under few-shot settings, this study proposes GSA-ICPO-Former, which integrates a Transformer architecture, an Improved Crested Porcupine Optimizer (ICPO), and a Gradient Sensitivity-based Adaptive Fine-Tuning method (GSA). The Transformer captures global spectral dependencies through self-attention; ICPO enhances hyperparameter optimization and prevents overfitting; GSA adjusts model parameters to adapt source-domain knowledge to target-domain features. After ICPO optimization, the model's $R^{2}$ in the source domain improved from 0.73 to 0.84, and RMSE decreased from 2.39 to 1.86 $\mu$g/L, representing improvements of 15.07% and 22.18%, respectively. In the target domain, compared with ICPO-Former, GSA-ICPO-Former raised $R^{2}$ from 0.54 to 0.79 and reduced RMSE by 32.97%, demonstrating enhanced cross-domain adaptability. Ablation studies confirmed the effectiveness of each optimization component. The model also showed robust performance across different coastal environments, including both nearshore (Tangdao Bay) and open-sea (Rongcheng) regions. It accurately tracked Chl-a distribution with limited samples and showed promise in dynamic algal bloom monitoring. Overall, the proposed model effectively bridges domain gaps and maintains high prediction accuracy in low-sample, cross-regional scenarios, providing a practical tool for satellite-based coastal water quality monitoring.