INTELLIGENT PREDICTION OF RAINFALL USING GPS AND METEOROLOGICAL DATA

Abstract

Information about the amount of rainfall and it's variability (spatial and temporal) is important for many real life applications such as hydrological cycle. Further, prediction of rainfall is economically important for countries thriving on agro-based economy. Moreover, rainfall is a critical factor behind many natural disasters such as landslide, flush flood etc. specifically in hilly terrain. However, accurate prediction of rainfall is still remains elusive, specifically in a low-cost user friendly way. So, an intelligent low cost system for prediction of precipitation will be in great use for user community including prediction of many natural disasters. Soft computation methods can construct computationally intelligent systems within a specific domain, adapt themselves and learn to do better in changing environments. The guiding principle of soft computing is to exploit tolerance for imprecision, uncertainty, robustness, partial truth to achieve tractability, and better rapport with reality. ANN based models are specifically useful where the dynamic processes and their interrelations for a given phenomenon are not known with sufficient accuracy as in case of rainfall prediction. The ANN based implicit model is capable to model without prescribing physical process, catching the complex nonlinear relation of input and output, and solving without the use of differential equation. In addition, ANN implicit model could learn and generalize from examples to produce a meaningful solution even when the- input data contain errors or is incomplete. Thus in this study, ANN based implicit model have been proposed to be developed for forecasting of rainfall; keeping in view of the limitations associated with the ANN based modeling.