An Approach for Rainfall Prediction Using Long Short Term Memory Neural Network

Weather forecasting is utilized in aiding up any organization for decision making in the context of disaster prevention. Rainfall prediction is one of the prominent and challenging tasks in the field of weather forecasting. Various techniques including physical, statistical and hybrid methods are employed for rainfall prediction. These methods further involve machine learning and deep learning techniques which have the potential to be harnessed. This paper incorporates Long Short-Term Memory (LSTM) based Recurrent Neural Network (RNN) for the prediction of rainfall. A standard dataset is used for the training and testing of the developed model. The time series rainfall data is pre-processed using Additive Seasonal Decomposition for enhancement of predictive analysis. This pre-processed data is then fed into the model. Artificial Neural Network (ANN) has been implemented for benchmarking the LSTM model. The parameters and factors taken into consideration for improvement in accuracy and performance evaluation include learning rate, epochs, number of hidden layer neurons, loss and Root Mean Square Error (RMSE). The proposed work finds extensive usage in several civil and defense applications such as disaster prediction and prevention, operational planning and many others.