Long Range Forecast on South West Monsoon Rainfall using Artificial Neural Networks based on Clustering Approach

Abstract

The purpose of this study is to forecast Southwest Indian Monsoon rainfall based on sea surface temperature, sea level pressure, humidity and zonal (u) and meridional (v) winds. With the aforementioned parameters given as input to an Artificial Neural Network (ANN), the rainfall within 1 0 x1 0 grids of southwest Indian regions is predicted by means of one of the most efficient clustering methods, namely the Kohonen Self-Organizing Maps (SOM). The ANN is trained with input parameters spanning for 36 years (1960-1995) and tested and validated for a period of 9 years (1996-2004). It is further used to predict the rainfall for 6 years (2005-2010). The results show reasonably good accuracy for the summer monsoon periods June, July, August and September (JJAS) of the validation years.

Application of Neural Network in Drought Forecasting; An Intense Literature Review

“…After development of ANN based model Krishna, [44], Malik et al, [52], have described the dependency on monsoon of world's most population and a great impact on the livelihood of the Indian families where cultivation is a major source of livelihood. Baboo and Shereef, [08], Jillella S.S. et al, [35], have also observed that the Neural Networks package supports different types of training or learning algorithms on which most useful algorithm is Back Propagation Neural Network (BPN) technique. They have applied Curve fitting and Extrapolation methods with back propagation and found that used model is the most important for prediction of weather.…”

“…After development of ANN based model Krishna, [44], Malik et al, [52], have found that the world's most population, specially the country like India, are depend on the cultivation, where the livelihood of the people are mostly dependent on the monsoon. Further, Baboo and Shereef, [8], Jillella S.S. et al, [35], have also observed that the impact of the monsoon on the livelihood of the Indian families, for prediction of the monsoon rainfall they have applied the Neural Networks package which supports different types of training or learning algorithms on which most useful algorithm is Back Propagation Neural Network (BPN) technique. They have applied Curve fitting and Extrapolation methods with back propagation and found that used model is the most important for prediction of weather.…”

Implementasi Metode Kohonen Untuk Prediksi Curah Hujan (Studi Kasus : Kota Pontianak)

“…Beberapa studi tentang penerapan jaringan syaraf tiruan untuk peramalan telah dilakukan [11,12,13,14,15]. Banyaknya aplikasi yang menyajikan sistem prediksi terkini juga merupakan salah satu kemajuan teknologi dibidang kecerdasan buatan.…”

An Evolving Neuro-Fuzzy System with Online Learning/Self-learning

“…Nowadays artificial neural networks (ANNs) are widely used in Data Mining tasks, prediction tasks, identification and emulation tasks etc. under conditions of uncertainty, nonlinearity, stochasticity and chaoticity, various kinds of disturbance and noise [1][2][3][4][5][6][7][8][9][10]. They are universal approximators and are able to learn using data which characterize the object under study.…”

“…under conditions of uncertainty, nonlinearity, stochasticity and chaoticity, various kinds of disturbance and noise [1][2][3][4][5][6][7][8][9][10]. They are universal approximators and are able to learn using data which characterize the object under study.…”