Using two-stage genetic algorithms to solve the nonlinear time series models for ten-day streamflow forecasting

Streamflow forecasting is of utmost importance for the management of water resources. A higher accuracy in flow prediction can lead to a more effective and comprehensive application of water resources. The characteristics of hydrological data can be classified as non-steady and nonlinear. This study used two-stage genetic algorithms to solve complex nonlinear time series models. Ten-day streamflows of the Wu-shi river in Taiwan were taken as an example. Compared with the traditional linear time series, the analysis verified that nonlinear time series models by two-stage genetic algorithms are superior.