The black-odor phenomenon has been widely reported worldwide and recognized as a global ecological risk for aquatic environments. However, driving factors for black-odor-related microorganisms and potential self-remediation strategies are still poorly understood. This study collected eight water samples (sites A–H) disturbed by different factors from the Jishan River located in Jinmen, Hubei Province, China. Black-odor-related environmental factors and functional bacterial structure were further measured based on the basic physicochemical parameters. The results indicated that different types of disturbed conditions shape the distribution of water quality and microbial community structures. Site B, which was disturbed by dams, had the worst water quality, the lowest abundance of functional microbes for Mn, Fe, and S biotransformation, and the highest abundance of functional microbes for fermentation. The natural wetlands surrounding the terminus of the river (site H) were keys to eliminating the black-odor phenomenon. Potential black-odor-forming microorganisms include Lactococcus, Veillonella, Clostridium sensu stricto, Trichococcus, Rhodoferax, Sulfurospirillum, Desulfobulbus, and Anaeromusa-Anaeroarcus. Potential black-odor-repairing microbes include Acinetobacter, Mycobacterium, and Acidovorax. pH and COD were paramount physiochemical factors contributing to blackening-odor-related microorganisms. This study deepens our understanding of driving factors for black-odor-related microorganisms and provides a theoretical basis for eradicating the black-odor phenomenon.

To realize the estimation of rainfall retention capacity for small reservoirs considering spatial soil moisture, a rainfall retention capacity model and its parameter schemes have been developed in this study. An iterative trial solution method considering potential rainfall and soil moisture for the model constructed was proposed for efficient computation. The rainfall retention capacity of 32 pilot small reservoirs located in ungauged basins of Hunan province was calculated starting from 21 August 2023. In addition, a continuous calculation was carried out from 1 August to 30 September 2023 using the proposed method for Heping reservoir. The results show that the Pearson’s correlation coefficients between rainfall retention capacity and available reservoir capacity and soil moisture, are 0.36 and −0.64, respectively. Using Heping reservoir as an example, this study effectively characterized the dynamic change in its rainfall retention capacity, which ranged from 123.6 mm to 68 mm in August 2023. The analysis indicates the rainfall retention capacity of the pilot small reservoirs calculated is reasonably related to the soil moisture, supporting risk visualization for small reservoirs within rain-affected regions. Furthermore, the impact of the antecedent precipitation on rainfall retention capacity can also be dynamically quantified in real time, which provides reference for the continuous management of small reservoirs.

Recurrent neural networks (RNNs) are widely used in reservoir operation because they can represent temporal dependencies, but their performance depends strongly on hyperparameter selection. This study compares eight evolutionary strategies—genetic algorithm (GA), particle swarm optimization (PSO), simulated annealing (SA), differential evolution (DE), ant colony optimization (ACO), artificial electric field algorithm (AEFA), slime mould algorithm (SMA), and white shark optimization (WSO)—for optimizing the hidden-node number and initial learning rate of long short-term memory (LSTM) and gated recurrent unit (GRU) models. Under common experimental settings, ACO required the shortest computation time in all reported dataset–architecture blocks, whereas the most accurate strategy depended on the dataset, architecture, and metric. For the total dataset, ACO–LSTM and AEFA–GRU produced the lowest reported MAEs (609.876 and 582.942 m³/s, respectively); WSO–LSTM was best for flood-season MAE (948.326 m³/s), and GA–GRU was best for non-flood-season MAE (516.088 m³/s). An independent chronological holdout experiment further verifies the search boundaries and validation-fitness procedure and shows that the fitness response also varies with hydrological period and RNN architecture. The results support objective- and dataset-specific optimizer selection, subject to the limitations of a single-reservoir case study and a restricted two-variable search.