论文部分内容阅读
富营养化模型是进行湖泊水环境质量预测和管理的重要工具,然而模型客观存在的误差一直是应用者关心的重要问题.数据同化作为连接观测数据与数值模型的重要方法,可以有效提高模型的准确性.集合卡尔曼滤波(EnKF)是众多数据同化算法中应用最为广泛的一种,可进行非线性系统的数据同化,并能有效降低数据同化的计算量.本研究以太湖作为具体实例,选择Delft3D-BLOOM作为富营养化模型,在数值实验确定EnKF集合数为100、观测误差方差为1％、模拟误差方差为10％的基础上分别进行模型状态变量同化以及状态变量与关键参数同步同化.结果显示,仅同化状态变量时,模型预测精度有所增加;同时同化状态变量和关键参数时,可显著提升模型在湖泊水环境质量预测中的精度.该研究为应用集合卡尔曼滤波以提高复杂的湖库富营养化模型模拟精度提供了有效的方法.“,”Numerical eutrophication model is an important tool to predict and manage the ecosystem of lakes and reservoirs.However,the objective errors of the model are always vital problems the users concerned.Data assimilation,which connects observations and model simulations,can effectively improve the accuracy of models.Ensemble Kalman filter(EnKF),which is one of the most widely used methods for data assimilation,is suitable for nonlinear system and has high computation efficiency.In this research,the Delft3D-BLOOM was taken as the eutrophication model,and Lake Taihu was taken as the study case.After numerical testing,the ensemble size was set to 100,the observation error variance was set to 1％,and the simulation error variance was set to 10％.Two data assimilation modes,assimilation of model state variables and synchronous assimilation of both state variables and key parameters,were examined.The results showed that the fitness between model simulation and observation was slightly improved when the state variable was updated.When both the state variables and parameters were assimilated,the fitness was significantly improved.The study provides a promising approach in using EnKF to improve the simulation accuracy of complex eutrophication models.