Abstract: By analyzing the monthly precipitation trend of Ankang Station from 1950 to 2014, we found that the precipitation at Ankang changes periodically. Having determined the delay time and embedded dimension of the nonlinear system of precipitation at Ankang using autocorrelation function method and C-C correlation integral method, we reconstructed the phase space of the precipitation series, and identified the chaotic characteristics of precipitation by using the G-P correlation dimension method and the maximum Lyapunov exponent method. Results reveal that the G-P correlation dimension algorithm indicates no chaos, while the maximum Lyapunov exponent method suggests chaos. With the existing 780 monthly precipitation data, we can forecast up to seven months of precipitation. The research finding offers scientific support for the runoff forecast in Hanjiang River and its downstream areas.

Key words: precipitation characteristics, chaos theory, C-C method, G-P correlation dimension, forecast time scale, Ankang Station in upper Hanjiang River