Eutrophication risk assessment using an integrated Sparrow Search Algorithm-Deep Neural Network and Copula: A case study of the Central Yunnan Plateau Lakes, China

Traditional trophic state index assessment methods tend to weaken the recognition of interactions and joint risks among water quality factors, making it difficult to comprehensively capture the multivariate dependencies in complex aquatic environments. To overcome this limitation, this study coupled the Sparrow Search Algorithm-Deep Neural Network (SSA-DNN) and Copula function models, establishing an integrated framework from remote sensing-based water quality retrieval to eutrophication risk assessment. This approach achieves an organic integration of remote sensing-driven retrieval and multivariate modeling. The results show that, compared with traditional machine learning methods (Random Forest and Support Vector Regression) and empirical models, the SSA-DNN model achieved higher accuracy in water quality parameter retrieval, with testing set R² values of 0.81, 0.81, and 0.80 for chlorophyll-a (Chl-a), total phosphorus (TP), and total nitrogen (TN), respectively. Between 2021 and 2024, Chl-a, TP, and TN concentrations in Dianchi and Qilu Lakes remained high, highlighting severe eutrophication issues. Copula-based risk assessment further revealed that TN was the key factor influencing eutrophication risk probability, with the maximum risk probability increasing by up to 68 %. The proposed integrated framework enables effective monitoring of water quality status and provides scientific and systematic technical support for ecological risk assessment in typical plateau lake regions.