Theme 5-10-Stochastic modeling of sediment load of upper Yangtze river (China).

Abstract

Temporal and spatial sediment load characteristics of Yangtze River, which is Asia's longest river, have been examined in this paper. Annual and monthly sediment load characteristics in temporal domain have been modelled. The annual sediment load data from 1950 to 1990 and monthly sediment load data from 1950 to 1969 have been used. Statistical tests such as turning point test, Kendall’s rank correlation test and Anderson’s correlogram test have been applied for randomness and trend identification. The periodicity in the sediment load data was analyzed by harmonic analysis and stochastic component was modeled by auto-regressive model.

The results indicate that the annual sediment load series is trend free at 5 percent significance level and the monthly means and standard deviations of sediment load show periodicity. The month to month correlation structure is non stationary. Using AR(1) model for dependent stochastic component 100 years of monthly sediment data have been generated. The monthly means of observed and generated data match almost perfectly.