<p>Accurate water quality forecasting plays a vital role in environmental monitoring and water resources management. However, the nonlinear dynamics of aquatic systems and the influence of diverse environmental factors present substantial challenges for time series modeling. To address these issues, this study proposes a novel Environment-Aware Hybrid State Space Model (AEMamba) for multivariate water quality time series forecasting. The proposed model integrates a trend–residual decomposition module with a selective state space modeling strategy. Specifically, the input sequence is first decomposed into long-term trends and short-term fluctuations using smoothing filters, which enables differentiated modeling of the two components. An environment-aware controller is then introduced to dynamically adjust the state transition parameters according to external environmental variables, thereby improving the model’s adaptability to changing conditions. In addition, a gated fusion mechanism is designed to combine a lightweight convolutional branch with a state space branch, enabling the model to capture both local dynamics and global dependencies. Extensive experiments on three real-world water quality monitoring datasets demonstrate that AEMamba consistently outperforms several state-of-the-art models, including long short-term memory (LSTM), gated recurrent unit (GRU), Informer, and Mamba, across multiple evaluation metrics, showing superior stability and generalization ability. Overall, this study provides an effective and reliable solution for water quality forecasting under complex environmental conditions.</p>