Dissolved inorganic nitrogen (DIN) and dissolved inorganic phosphorus (DIP) are the two dominant nutrients influencing seawater quality. Due to the non-optically active nature of nutrients and their regional variability, relying solely on optical data is inadequate for achieving high-precision remote sensing retrieval in complex marine environments. We developed a novel remote sensing algorithm for DIN and DIP using MODIS remote sensing reflectance (Rrs) products and the XGBoost machine learning framework. Beyond optical inputs, our model integrates sea surface temperature (SST) and spatiotemporal information, including a shoreline-based pixel location descriptor, which significantly enhances model performance. We generated monthly average distributions of DIN and DIP concentrations across China’s coastal waters from 2012 to 2022. The findings highlight extensive high-nutrient zones in the Bohai Bay and Changjiang River (Yangtze River) Estuary−Hangzhou Bay regions, with a notable declining trend in nutrient concentrations. The Zhujiang River (Pearl River) Estuary also exhibits elevated nutrient levels, albeit with minimal changes. This study pioneers the incorporation of dual-coordinate information in nutrient retrieval for complex marine environments, significantly improving model accuracy and addressing stripping artifacts associated with single-coordinate systems. Moreover, the results provide unprecedented spatiotemporal insights into nutrient distributions in China’s coastal waters, offering valuable support for marine environmental management and policy-making.

Recently, Randomly Wired Neural Networks (RWNNs) using random graphs for Convolutional Neural Network (CNN) construction have shown efficient layer connectivity, but may limit depth, affecting approximation, generalization, and robustness. In this work, we increase the depth of graph-structured CNNs while maintaining efficient pathway usage, which is achieved by building a feature-extraction backbone with a depth-first search, employing edges that have not been traversed for parameter-efficient skip connections. The proposed Efficiently Pathed Deep Network (EPDN) reaches maximum graph-based architecture depth without redundant node use, ensuring feature propagation with reduced connectivity. The deep structure of EPDN, coupled with its efficient pathway usage, allows for a nuanced feature extraction. EPDN is highly beneficial for processing remote sensing images, as its performance relies on the ability to resolve intricate spatial details. EPDN facilitates this by preserving low-level details through its deep and efficient skip connections, allowing for enhanced feature extraction. Additionally, the remote-sensing-adapted EPDN variant is akin to a special case of a multistep method for solving an Ordinary Differential Equation (ODE), leveraging historical data for improved prediction. EPDN outperforms existing CNNs in generalization and robustness on image classification benchmarks and remote sensing tasks. The source code is publicly available at https://github.com/AnonymousGithubLink/EPDN.