Application of Empirical Mode Decomposition Method to Synthesize Flow Data: A Case Study of Hushan Reservoir in Taiwan

Although empirical mode decomposition (EMD) was developed to analyze nonlinear and non-stationary data in the beginning, the purpose of this study is to propose a new method&mdash
based on EMD&mdash
to synthesize and generate data which be interfered with the non-stationary problems. While using EMD to decompose flow record, the intrinsic mode functions and residue of a given record can be re-arranged and re-combined to generate synthetic time series with the same period. Next, the new synthetic and historical flow data will be used to simulate the water supply system of Hushan reservoir, and explore the difference between the newly synthetic and historical flow data for each goal in the water supply system of Hushan reservoir. Compared the historical flow with the synthetic data generated by EMD, the synthetic data is similar to the historical flow distribution overall. The flow during dry season changes in significantly (&plusmn
0.78 m3/s)
however, the flow distribution during wet season varies significantly (&plusmn
0.63 m3/s). There are two analytic scenarios for demand. For Scenario I, without supporting industrial demand, the simulation results of the generation data of Method I and II show that both are more severe than the current condition, the shortage index of each method is between 0.67&ndash
1.96 but are acceptable. For Scenario II, no matter in which way the synthesis flow is simulated, supporting industrial demand will seriously affect the equity of domestic demand, the shortage index of each method is between 1.203 and 2.12.