A Comparative Study and Analysis of Time Series Forecasting Techniques for Indian Summer Monsoon Rainfall (ISMR)

Abstract

The importance of monsoon rains cannot be looked over, as it has an impact on activities all year round, from agricultural to industrial. In the domains of water resource management and agriculture, accurate rainfall estimation and forecast is extremely useful in making crucial decisions. This study presents various deep learning approaches such as Multi-layer Perceptron, Convolutional Neutral Network, Long Short-Term Memory Networks, and Wide Deep Neural Networks to forecast the Indian summer monsoon rainfall (ISMR) (June–September) based on seasonal and monthly time scales. For modeling purposes, the ISMR time series data sets are divided into two categories: (1) training data (1871–1960) and (2) testing data (1961–2016). Statistical analyses reveal ISMR’s dynamic nature, which couldn’t be predicted accurately by statistical and mathematical models. Therefore, this study provides a comparative analysis that demonstrates the effectiveness of various algorithms to forecast ISMR. Moreover, it also weighs the result with established existing models.