Forecasting Precipitation Using SARIMA Model: A Case Study of Mt. Kenya Region

Abstract

Precipitation estimates are an important component of water resources applications, example, in designing drainage system and irrigation. The amount of rainfall in Kenya fluctuates from year to year causing it to be very hard to predict it through empirical observations of the atmosphere alone. Our objective was to determine the forecasted values of precipitation in Mt. Kenya region and also to determine the accuracy of the SARIMA model in forecasting precipitation in the same region. This research considers a univariate time series model to forecast precipitation in Mt. Kenya region. We fitted the SARIMA model to our data and we picked the model which exhibited the least AIC and BIC values. Finally, we forecasted our data after following the three Box-Jenkins methodologies, that is, model identification, estimation of parameters and diagnostic check. Having three tentative models, the best model had two highly significant variables, a constant and with p-values< 0.01 respectively. This model passed residual normality test and the forecasting evaluation statistics shows ME=-0.0053687, MSE=0.96794, RMSE=0.98384 and MAE= 0.75197. Indeed, SARIMA model is a good model for forecasting precipitation in Mt. Kenya region