Automatic calibration of the two-dimensional hydrodynamic and water quality model using sequential uncertainty...

Reservoir hydrodynamic and water quality modeling, in conjunction with the monitoring programs, is one of the essential tools for controlling the pollution of these types of water bodies. The complexity of the model, data scarcity, and the variable nature of natural phenomena lead to uncertainty in models, which should be considered in the calibration process of these models. Uncertainty-based automatic calibration is one of the methods that can be effective in achieving a high-reliability model. In this paper, the Sequential Uncertainty Fitting (SUFI-2) algorithm was used for the automatic calibration of the two-dimensional hydrodynamic and water quality model (CE-QUAL-W2) for the reservoir under parameter uncertainty conditions. To this end, the CE-QUAL-W2 model was developed to simulate the temperature and water surface elevation of the Karkheh Dam reservoir (western Iran). The parameters affecting temperature were regarded as uncertain parameters in the calibration process, including the coefficients of longitudinal eddy viscosity, longitudinal eddy diffusivity, Chezy coefficient or Manning, wind sheltering, solar radiation absorbed in the surface layer, extinction coefficient for pure water, and the experimental coefficients of wind speed function. The developed method demonstrated a high potential for matching the simulated temperature and water surface elevation for the reservoir with the measured data. Averagely, 69% of the simulated temperature and 90% of the simulated water surface elevation were located within the 95% confidence interval. The SUFI-2 algorithm also showed better performance in terms of the convergence rate compared with the particle swarm optimization (PSO) algorithm, which indicated a lower number of calls (80 calls compared to 2000 calls) and could reduce the total root-mean-square error by 9.6%.