Paper Details

[ Downloads: 0 | Abstract Viewed: 80 | Pages: 9 ]

Abstract of the Article

Note: English CIVILICA is in its Trial Period so Full Texts can not be provided! Persian users can download it here

Combination of the Forecasts of Different Rainfall-Runoff Models

Authors:
[ Mohsen Poorreza Bilondi ] - MSc. Student of Hydraulic Structures, Shahid Chamran University
[ Reza Tareghian ] - MSc. Student of Hydraulic Structures, Shahid Chamran University
[ Mehdi Ghomshi ] - Associate Professor, School of Water Science Engineering, Shahid Chamran University

Abstract:

Rainfall-runoff models are used as components in river flood forecasting systems. In such systems, a rainfall-runoff model may have been selected on its accuracy, its familiarity to the user, its ease of use, the type of the catchments and the available data. To date, a lot of rainfall-runoff models have been developed and applied, but it can’t be claimed that there is a one model which can perform satisfactorily at all times or under all conditions. To solve this problem, methods of combining the results of different rainfall-runoff models have been proposed. In this investigation, using rainfall, runoff and watershed characteristics information of two South Fork and North Fork rivers watershed, river flow was forecasted by combining the results of rainfall –runoff models. SLM, LPM, LVGFM and SVRC models were used as the rainfall-runoff models. Simple average, linear & nonlinear regression and neural networks methods were applied to combine the results of four different rainfall-runoff models. When the neural network combination method developed, not only networks were trained and tested using back propagation algorithm, but also different transfer functions and hidden layer nodes were introduced. The results showed that combination methods could predict river flow with higher correlation and les error. Furthermore neural networks and nonlinear regression combination methods predicted better than the others.

Keywords: