Modelling and Prediction of Annual Rainfall Using ARIMA

Abstract

Rainfall is a key climatic parameter that regulates hydrological cycles and ecosystem balance. Analyzing annual rainfall variability aids in understanding long-term climatic trends and supports effective resource and risk management. This study employs the Autoregressive Integrated Moving Average (ARIMA) model to forecast annual rainfall in Kollam district, Kerala. Historical rainfall data were subjected to time series analysis to examine trends, patterns, and stationarity. The series was found to be non-stationary and was transformed through differencing to achieve stationarity. The optimal ARIMA model was selected based on the Bayesian Information Criterion (BIC) and validated using statistical performance indicators. The model achieved a predictive accuracy of 83.89% for the year 2024. Forecasts for the period 2024–2040 indicate realistic rainfall variations, demonstrating the potential of ARIMA modeling for reliable rainfall prediction and climate-related planning in monsoon-prone regions such as Kerala.