Sharma, M.K. (Ass. Prof.Gebretsadikan, Amaha2022-03-282023-11-092022-03-282023-11-092010-10http://10.90.10.223:4000/handle/123456789/30939In thi s study an attempt is made to explore the practical procedures in time domain univariate Box-Jenkins methodology for modeling and forecasting monthly rai nfa ll in Tigray region. In particular, the study employs Seasonal Autoregressive Integrated Moving Average (SARIMA) model for monthly rain fall data collected by the Nat ional Meteoro logy Agency at Mekele station for the period from January 1975 to December 2009. Through the various model identification, estimation and diagnostics methods, we developed models that can adequately fit to the data. Residual analysis which is the important tool for diagnost ic checks shows that there was no violation of assumptions in relation to model adeq uacy. Further model selection was performed at the forecasting stage usi ng forecas ting accuracy methods based on the va lidation period. The point forecast results showed a very close match with the pattern of the actual data and better forecast ing accuracy in the validation period. Accordingly, the more parsimonious SARIMA (0, 0, I) x ( I, I, 4)12 model is found appropriate to describe the observed data. Therefore, the res ults of the study indicate that SARIMA model of Box-Jenkins methodology allows in capturing mure complex description of the seasonality, autocorrelation structure and non- stationary of the series and appears to be reasonably good in forecast ing the monthly rainfall series. Future forecast results of the model show that there seems to be no trend of increasing or decreasing pattern over the period from January 20 I 0 to September 20 11 . Key words: Monthly rainfa ll, SARIMA, Box-Jenkins, Forecasting.enMonthly rainfallSARIMABox-JenkinsForecasting.Thesis