Inspection method for random simulation of runoff sequence based on sample entropy and statistical parameters

The traditional inspection method of runoff simulation sequences is based on statistical parameters. While the method can reflect the coincidence of statistical parameters of runoff simulation series and observed data at different time intersections, it ignores the characteristics of runoff as time series (e.g., nonlinearity, dissipation and complexity) which might be required in hydrological engineering. By introducing the concept of sample entropy to quantify complexity, we proposed a new method based on sample entropy and statistical parameters for inspecting random simulation runoff sequences, which can supplement the previous test method. To test the new method, we used the monthly runoff simulation series of Luohe Hydrological Station in Huaihe River Basin, Tanghe Hydrological Station in Yangtze River Basin and Heishiguan Hydrological Station in Yellow River Basin as examples. Our evaluation revealed that this new inspection method can effectively find anomalies (ignored by statistical parameters) about flow value and distribution location of the monthly runoff; and estimate the error values of flow and distribution location in simulation sequences of three basins. Compared to the previous inspection method, this new method can screen out the simulation sequences of higher accuracy more effectively, which offers a new way to improve the accuracy of simulation sequences.