Time-varying parameterization of a reservoir water quality model using data assimilation

Complex and multi-dimensional water quality (WQ) simulation models often suffer from structural misspecification, calibration difficulties, and constant parameterization, which can be unrealistic under evolving conditions. Dual state-parameter updating schemes based on data assimilation enable adaptive parameter and state updates by integrating new observations. This approach provides an automatic and adaptive calibration framework that not only improves model flexibility but may also alleviate process misspecifications. This study pioneers the adjustment of temporal evolution of parameters in a WQ simulation model of a deep stratified reservoir using a dual state-parameter updating framework implemented via the ensemble Kalman filter. It focuses on improving the evaluation of temperature profile and dissolved oxygen (DO) profile of the Wadi Dayqah Reservoir in Oman and uses the CE-QUAL-W2 WQ simulation model. To effectively update parameters, the study identified the optimal configuration of parameter, state, and observation vectors and evaluated multiple parameter evolution strategies. Overall, incorporating time-evolving parameter adjustment through a dual state-parameter updating framework significantly improved WQ estimation in the reservoir, achieving a noticeable reduction in the root mean square error of the temperature and DO profiles estimations by 16.8% and 43%, respectively.