Li Xiangyang¹, Chu Yongwei¹, Ren Lixin²

(1.The Heihe Basin Administrative Bureau of the Yellow River Conservancy Commission, Lanzhou 730000, china 2.  Bureau of the Upper-Reach Hydrology and Water Resources  of the Yellow River Conservancy Commission, Lanzhou 730030, China£©

Abstract:  Through analyzing the flow changeable regularity on Yingluoxia station of the Heihe River, the annual runoff of the station is divided into three kinds according to frequency 25% and 75%: plenty water year x_i£¾55.0, medium water 42.1¡Üx_i¡Ü55.0, less water year x_i< 42.1. By figuring out the shifting frequency, it is found that the annual runoff will have all sorts of possible changes from a state to another state, but the dataset of shifting frequency will be 81.8%. Thereby changeable process of annual runoff are random and mutuality. To forecast the process, the pre-atmosphere cyclic factor and average annual flow forecasting model as well as the time series composition model of average annual flow are set up. It has a highl forecasting precision for annual runoff. Through error evaluation of the two models, it can be found that both models are top level ones. Tests show they are rather steady.

Key words:  runoff forecast; modeling study; Heihe River