Author(s): Nophakhun Somsin; Pichaid Varoonchotikul
Linked Author(s):
Keywords: Flood forecasting; Artificial Neural Network; Time marching scheme
Abstract: An artificial neural network model was developed to analyse and forecast the behaviour of the Chumporn River in Thailand during year 1990. The model makes use of the upstream river gauging stations (X. 46& X. 64) and predicts the water level of the river at station X. 158 at Khlong Tha Taphao, Chumporn. Model predictions up to six hour ahead are very accurate (i. e. coefficient of efficiency is more than 95%) when the model is used with a hourly time horizon. Increasing the time horizon decreases the accuracy of the model and also amplifies the phase errors. The time horizon of the input data and the response time of the river basin limit performance of the model.
Year: 2002