Record Details

Short-term rainfall prediction using ANN and MT techniques

DSpace at IIT Bombay

View Archive Info
 
 
Field Value
 
Title Short-term rainfall prediction using ANN and MT techniques
 
Creator Mandal, T
Jothiprakash, V
 
Subject rainfall prediction
data-driven techniques
artificial neural networks
model tree
Koyna Dam Station
 
Description Most of the rainfall prediction models use atmospheric weather data, which are somewhat difficult to access by common water resources managers. On the other hand, data-driven techniques are finding wider application in forecasting many hydrological variables. The data-driven technique predicts the future variable better if there is a well-defined pattern with or without noise in the data set. In the present study, this ability of data-driven techniques, such as artificial neural networks (ANNs) and model tree (MT), has been applied to predict the next time step rainfall using lagged time series of observed rainfall data. The models were trained and tested with 47 years of daily rainfall measurements at the Koyna Dam, Maharashtra, India. Among various available training algorithms, multilayer perceptron, radial basis function (RBF) and time lagged recurrent networks (TLRN) have been attempted. It is found that TLRN has captured the pattern in a better way. In case of MT, various trials on pruning and smoothening have been carried out and found that un-pruned and un-smoothed MT performed better. It is found that both ANN and MT models have performed equally better, indicating that they are promising techniques for short-term rainfall prediction. However, the agreeable result shows that the data-driven techniques may be explored further for prediction of short-term rainfall data from the observed rainfall series.
 
Publisher Indian Society for Hydraulics
 
Date 2012-09-12T11:01:00Z
2012-09-12T11:01:00Z
2012
 
Type Article
 
Identifier ISH Journal of Hydraulic Engineering, 18(1) 20-26
2164-3040
http://dspace.library.iitb.ac.in/jspui/handle/100/14399
 
Language English