Author(s): Xiekang Wang; Tonghuan Liu; Shuyou Cao
Linked Author(s):
Keywords: Rtificial neural networks; Water and sediment; Small watershed
Abstract: In recent years, artificial neural networks model has become a novel tool to simulate nonlinear processes in many scientific fields. With the development of economics and address of environmental protection, water and soil loss and related researches that have an intricate nonlinear relationship among impacted factors become an important and urgency problem. This paper mainly aims at several traditional forecast problems on water and sediment nonlinear processes in small watershed using artificial neural networks based on available experimental and field data. The adopted neural network models in this study have been founded considering the physical characteristics of predicting problem and model function of artificial neural networks, in addition, the different algorithms related neural networks model have been took into accounted. At the same time, the advantage or disadvantage between artificial neural networks and traditional statistic or physical dynamics methods has been analyzed and discussed in order to compare the accuracy and precision of forecast problem. Several water and sediment problems including overland flow, soil erosion and bedload transport rate based on neural network models have been simulated, these comparisons between observation and prediction indicate artificial neural network models proposed in this study are useful and reliable to forecast the water and sediment processes. However, the further study is necessary in order to extend the other small watershed or larger scale drainage basin.
Year: 2004