Relation And Forecasting Of Runoff And Sediment In The Lancang River

The change of river water flux directly affects sediment transport and evolution of river bed.Human activities interfere natural relation between water and sediment,and thus have an important impact on the whole river environment.Seeking for the natural nexus of water and sediment in the Lancang River has a great scientific significance and application value for knowing the river motion and clarifying the influence of human activities on the river environment.Based on the study of water and sediment relation and prediction of sediment flux in the middle reaches of Lancang River,detailed conclusions have been drawn as following:(1)After analyzing water and sediment natural changes in Jiuzhou and Gajiu station,the formula of sediment flux derivative was established where the coefficient and the exponent were influenced by the characteristics of system and the characteristics of local sediment transportation.High-order-power function of sediment flux and runoff was obtained by integral,which promotes the relation between sediment flux and runoff from the experience to the theoretical level.(2)Based on the Mann-Kendall method and wavelet analysis,the variations of flow and sediment load were analyzed.It is found that the water and sediment fluxes had a trend turning point in 1998 caused by the rainstorm.Affected by the construction of the power station,sediment flux also had an another turning point in 2005.The main periods of runoff are 18,8 and 4 years,and the main periods of sediment flux are 18,13 and 8 years.Both discharge and sediment had the same first main period.(3)According to water and sediment data measured,it is found that the peak of water and sediment were unsynchronous.Established the optimized Back-Propagation Neural Network(BP-NN)in which the unsynchronization was considered and the Set Pair Analysis(SPA)in which the Wavelet Denoise was considered.Using the optimized BP-NN and WD SPA models forecasting the monthly and annual sediment flux.The results show that the optimized BP-NN can ameliorate the peak offset and improve the prediction precision.The wavelet neural network(WNN)has the same accuracy with the optimized BP-NN.WNN is more accurate in the numerical prediction while the optimized BP-NN is more suitable for the trend prediction.WD SPA model has the good prediction precision and is good at mutation forecasting.Comparing the results of WD SPA and BP-NN can quantitatively analyze the effects of human activities.