Rainfall-Runoff Modeling Using ANN for Wainganga River in Godavari Basin of India

摘要

The rainfall-runoff model is required to ascertain relationship between rainfall and runoff. Hydrologists are often confronted with problems of prediction and estimation of runoff due to highly nonlinear and complex relationship. In this study daily rainfall data (1999-2(X)8) of Balaghat. Seoni and Keohiri stations and daily runoff observations of Kumhari outlet were used in model development. The Artificial Neural Network (ANN) is a method of computation inspired by studies of nervous systems in living organisms. Being a robust tool for modelling many of complex non-linear hydrologic processes it recently used in forecasting of daily stream flow through ANN in MATLAB, However for large area watershed, rainfall of previous days is also included in forecasting models. Highest value of coefficient of correlation between estimated and observed ninoff was found to be 0.936 and 0.828 during training and validation respectively by using ANN. The ANN model with LM algorithm, 7 Numbers of neurons and 60% and 40% length of record for training and validation is found best for model development in runoff estimation.