Time Series SARIMA Modelling and Forecasting of Monthly Rainfall and Temperature in the South Asian Countries

Abstract

This study attempted to examine the future behaviour of monthly average rainfall and temperature of South Asian countries by using the Seasonal Autoregressive Moving Average model. Mann–Kendall trend test with Sen's Slope Estimator, to find the trending behaviour of all data series. The study has also been attempted to compare the above methods with the help of actual data. The monthly average rainfall and temperature of South Asian countries except Afghanistan and Maldives viz. Bangladesh, Bhutan, India, Nepal, Pakistan, Sri Lanka data from January, 1961 to December, 2016 have been collected from World Bank Group, Climate change knowledge portal. For estimating the trending behaviour, a non-parametric model such as the Mann–Kendall test was used with Sen's slope estimation to determine the magnitude of the trend. Box–Jenkins methodology was also used to develop the model and estimate the forecasting behaviour of rainfall and temperature in South Asian countries. Forecasting is carried out for both monthly rainfall and the average temperature of all the countries using best fitted models based on the data series. The monthly data from January, 1961 to December, 2010 are considered for validation of the model can be regarded as in-sample forecast and the data from January, 2011 to December, 2016 are used as out-sample forecast. The forecasting values with 95% confidence limit from January, 2011 to December, 2021 using best-fitted models for both rainfall and temperature. We conclude that climate change occurs for both rainfall and temperature in South Asian countries from the study period. The selected model can be used for forecasting both rainfall and temperature of respective countries from January, 2011 to December, 2021. As the climatic data analysis is valuable to understand the variation of global climatic change, this study may help for future research work on rainfall and temperature data.