Time-Series Analysis of Monthly Rainfall Data for the Nuwara Eliya District, Sri Lanka

Abstract

One of the most important climatic element is Rainfall, that directly influences on the agriculture. As a real world practice, the rainfall data has a seasonal trend with short term and long term oscillations; especially, monthly rainfall forecasting is significant to make decisions in management of agricultural scheme and manage daily human activities. In this current study monthly average rainfall of Nuwara eliya district, Sri Lanka since 1996 to 2015 were considered. According to the unit root study, original observations are non-stationery. So, we moved 1st difference for further study. Furthermore, Seasonal Autoregressive Integrated Moving Average (SARIMA) model was fitted for analyzing and validation 228 monthly observations. According to minimum Akaike information criteria (AIC), SARIMA (1, 0, 0) (1, 0, 2) (12) is selected as a best model for forecasting rainfall in this selected region.