In the current study, the efficiency of Wavelet-based Least Square Support Vector Machine （WLSSVM） model was examined for prediction of daily and monthly Suspended Sediment Load （SSL） of the Mississippi River. For...In the current study, the efficiency of Wavelet-based Least Square Support Vector Machine （WLSSVM） model was examined for prediction of daily and monthly Suspended Sediment Load （SSL） of the Mississippi River. For this purpose, in the first step, SSL was predicted via ad hoc LSSVM and Artificial Neural Network （ANN） models; then, streamflow and SSL data were decomposed into sub- signals via wavelet, and these decomposed sub-time series were imposed to LSSVM and ANN to simulate discharge-SSL relationship. Finally, the ability of WLSSVM was compared with other models in multi- step-ahead SSL predictions. The results showed that in daily SSL prediction, LSSVM has better outcomes with Determination Coefficient （DC）=o.92 than ad hoc ANN with DC=o.88. However unlike daily SSL, in monthly modeling, ANN has a bit accurate upshot. WLSSVM and wavelet-based ANN （WANN） models showed same consequences in daily and different in monthly SSL predictions, and adding wavelet led to more accuracy of LSSVM and ANN. Furthermore, conjunction of wavelet to LSSVM and ANN evaluated via multi-step-ahead SSL predictions and, e.g., DCLssVM=0.4 was increased to the DCwLsSVM=0.71 in 7- day ahead SSL prediction. In addition, WLSSVM outperformed WANN by increment of time horizon prediction.展开更多
基金supported by the University of Tabriz under grant No. 1117394325
文摘In the current study, the efficiency of Wavelet-based Least Square Support Vector Machine （WLSSVM） model was examined for prediction of daily and monthly Suspended Sediment Load （SSL） of the Mississippi River. For this purpose, in the first step, SSL was predicted via ad hoc LSSVM and Artificial Neural Network （ANN） models; then, streamflow and SSL data were decomposed into sub- signals via wavelet, and these decomposed sub-time series were imposed to LSSVM and ANN to simulate discharge-SSL relationship. Finally, the ability of WLSSVM was compared with other models in multi- step-ahead SSL predictions. The results showed that in daily SSL prediction, LSSVM has better outcomes with Determination Coefficient （DC）=o.92 than ad hoc ANN with DC=o.88. However unlike daily SSL, in monthly modeling, ANN has a bit accurate upshot. WLSSVM and wavelet-based ANN （WANN） models showed same consequences in daily and different in monthly SSL predictions, and adding wavelet led to more accuracy of LSSVM and ANN. Furthermore, conjunction of wavelet to LSSVM and ANN evaluated via multi-step-ahead SSL predictions and, e.g., DCLssVM=0.4 was increased to the DCwLsSVM=0.71 in 7- day ahead SSL prediction. In addition, WLSSVM outperformed WANN by increment of time horizon prediction.
