Dynamic time series method by Artificial Neural Network (ANN) for Predicting of Rainfall in India


J.Ashok, G.Kalaiarasi, S.Purushothaman, P.Surendar, Associate Professor,
Department of ECE, VSB Engineering College, Tamilnadu, India.


This Research Paper proposes Ann Neural Network for Rain forecasting. Mathematical Model like Arima, Sann provide Date Prediction with Time Series Calculation with small data sets. This Proposed Research uses Dynamic Time Series method for data sets from the earlier period rain fall data as a input for the system. India is country of farming; the 60 percent of land of agricultural is depending upon rainfall water. Only 40 percent land depend on ground water. Looking at the greater need of farming on rainfall, in this research we take rain water data of the past few years (2002 To 2010) and predict the future rain fall. This research uses the data of Indian metrological data collected from all parts of India as a uncooked data for analysis. We take all the state of India which vastly producing agricultural foods example paddy, wheat. We supply the data of 8 years of rain in all states and forecast the rainwater for upcoming Years by mat lab NARX (Nonlinear auto regression with external exogenous input) Ann network by Levenberg- Marquardt algorithm. Even though the climate change is major issue in rain prediction, computer accuracy will be near to the future figures prediction. Our consequences are compared with the arithmetical methods like ARIMA. By mat lab neural networks predictions are resourceful then other results.