Performance comparison of three multi-objective optimization algorithms on calibration of hydrological model

In this essay, the comparison of three multi-objective optimization approaches MOPSO, NSGA-II and MOSCEM-UA has been carried out in the auto calibration of hydrological model- Hymod. By carrying out the calibration on two objectives of high flow and low flow objective functions, the Pareto front can be drawn. The performance of the three optimization algorithms is analyzed depending on three criterions that are the optimization time cost, Pareto front spacing rate and the dominating rate. Through analyzing the comparison results of MOSCEM-UA and NSGA-II with MOPSO method, the performance of convergence rate, Pareto non-dominant spacing rate and iteration speed are ideally expected. The simulation result with MOPSO algorithm is reasonable in the high flow and low flow process. The prediction area drawn from the optimization result indicates the reliability of the model. Meanwhile, the model uncertainty is also discussed to some extent.