Nonstationarity Analysis and Simulation of Karst Spring Flow in Dry Season

Hydrologic series has high nonlinear and nonstationary properties because of heterogeneity and anisotropy of karstic aquifer. Taking spring flow sequence day by day of Maoshuikeng station in Houzhai area for example, make the sequence nearly stationary by getting the natural logarithm. Based on the stationary sequence, AR (2), an auto regression model, is developed to simulate and predict the karst spring flow. The simulation results showed that AR(2) model is suitable to short-term (3 days) prediction.