Signal processing for the annual runoff process by wavelet and R/S analysis: A case study of the Tarim headwater basin

The paper studies signal processing for the annual runoff processes of the Tarim headwater basin. By utilizing the wavelet and R/S analysis, the main findings are generalized as follows: (1) The annual runoff processes of the three headwaters of the Tarim River are complex nonlinear systems, which show nonlinear trends, as well as fractals, and long-term correlation. (2) In the time scale of 16 (2⁴) years from 1957 to 2002, the annual runoff in the Aksu and Yarkand Rivers show an increasing trend in general, while in the Hotan River a slightly decreasing trend is observed for the same time period. If the time scale reduces to 8 (2³) or 4 (2²) years, the annual runoff in each river does not show an apparent trend of either increasing or decreasing. (3) The time series of annual runoff in each river present a long-term correlation characteristic. The Hurst exponent in the period from 1989 to 2002 indicates that the annual runoff in the Aksu and Yarkand River will show an increasing trend, and that in the Hotan River will show a decreasing trend in the years after 2002.