Click for new scientific resources and news about Corona[COVID-19]

Paper Information

Journal:   JOURNAL OF WATER AND SOIL CONSERVATION (JOURNAL OF AGRICULTURAL SCIENCES AND NATURAL RESOURCES)   2011 , Volume 18 , Number 1; Page(s) 163 To 180.
 
Paper: 

STREAM FLOW FORECAST USING ANN MODELS AND INVESTIGATION OF MODEL’S PERFORMANCE BY SOI INPUTS

 
 
Author(s):  ANVARI TAFTI S., SAGHAFIAN B., MORID S.*
 
* DEPT. OF WATER STRUCTURE, TARBIAT MODARES UNIVERSITY
 
Abstract: 

Traditional modeling techniques such as regression and time series, often fail in the modeling of nonlinear hydrological processes like stream flow forecasts. Therefore, nonlinear techniques such as Artificial Neural Networks (ANNs) could be more efficient tools for forecasting these processes. This study aims to forecast the next 1- to 3-months flow of Karoon basin. For these forecasts, first four input models of ANN have been constructed by Principal Component Analysis (PCA) and Cross Correlation (CC) techniques. Then, using Cross Validation (CV) method and model performance evaluation indices like determination coefficient, mean absolute error and root mean square error, the best input model was identified. Also, to determine the effectiveness of each input variables and principal components, their single and compound role on ANN performance was investigated. Finally, the effect of SOI index at t, t-1 and t-2 times on the best model performance was evaluated. Results showed that thermal variables have the greatest influence on Karoon basin streamflow. Also, SOI index as input variable can improve the performance of model averagely about 5% in MAE and RMSE and 4% in R2 values.

 
Keyword(s): STREAM FLOW FORECAST, ANN, SOI, KAROON BASIN
 
References: 
  • ندارد
 
  Persian Abstract Yearly Visit 53
 
Latest on Blog
Enter SID Blog