Abstract

Natural water source is one of the influencing factors for crop cultivation and production in agricultural sector. At the same time most of the agro industrial development interdepends the natural resources and its impact for its progress. Hence, there exists an ever growing demand for effective rainfall prediction system- The ongoing demand for precise forecast methods has led to thedevelopment of computerized rainfall forecastsscenarios. This investigation makes an attempt to achieve precise rainfall predictions using Neural Network (NN) approach. This researchintroduces a novel exhaustive search based Maximum Frequency Weighted Feature Selection (MFWFS) using approach to identify the significant weather parameterfor prediction. The effects of feature selectionin model performance were also investigated. This was done by examining the performance of the neural networksprediction model using both complete and reduced parameters. A meticulous comparison of the overall performance indicated that the back propagation algorithm approach based neural network model outperformed better than existing methods. The proposed NN architecture achieved 0.0499 error rate and 95.01% prediction accuracy using effective weather parameters. This investigation introduces acompact interactive graphical user interface (GUI) based tool developed using C# and net platform to enable users to conduct meteorological assessment on their own ease. This tool enables users to train and test the NN model for various input options and to visualize the resultsby processing the rainfall forecast scenarios.