Please use this identifier to cite or link to this item:
Title: Predictability of sea surface temperature anomalies in the Indian Ocean using artificial neural networks
Authors: Tripathi, K C
Das, I M L
Sahai, A K
Keywords: Artificial neural networks;Error back propagation;Sea surface temperature;Linear regression;Indian Ocean
Issue Date: Sep-2006
Publisher: CSIR
Abstract: Artificial Neural Networks (ANN) have been used to access the predictability of sea surface temperature (SST) anomalies for the small area of Indian Ocean Region (27º to 35º S and 96º to 104º E). Twelve networks, corresponding to each month, have been trained on the area average SST values. The performance of the networks has been evaluated and found that the models have been able to predict the anomalies with a reasonably good accuracy. The performance of ANN models has been compared with the Linear Multivariate Regression model to justify the use of a nonlinear model. It has been found that whenever the dependence of present anomalies on the past anomalies show a nonlinear relationship, the linear model such as regression models fails to make any forecast. These are the months of June, September, October and November. In such cases the nonlinear ANN models have been proved to be fairly useful and make relatively better forecasts. When the dependence is linear, the performance of the ANN models is similar to the regression models. In such cases, use of ANN models only leads to increase in complexity without significant improvement in the performance.
Page(s): 210-220
ISSN: 0379-5136
Appears in Collections: IJMS Vol.35(3) [September 2006]

Files in This Item:
File Description SizeFormat 
IJMS 35(3) 210-220.pdf156.71 kBAdobe PDFView/Open

Items in NOPR are protected by copyright, with all rights reserved, unless otherwise indicated.