PAPanga, S.M.Nyimvua, I.S.Mbalawata
The prediction of rainfall on monthly time scale has been attempted by a number of researchers by using different time series techniques at different time periods around the world. It is challenging to forecast rainfall at monthly time scale because of spatial and temporal random variation caused by a numbers of dynamic and environmental factors. In this paper, an attempt has been made to develop a Seasonal Autoregressive Integrated Moving Average (SARIMA) Model to analyze long term monthly rainfall data of Dar es Salaam region in Tanzania for the period of fifty three years (1961 to 2014). Rainfall observations were discovered to have Seasonality and also non-stationarity and hence differencing and Seasonal differencing was used to attain stationarity. Rainfall data were found to have two seasons namely October to December (OND) and March to May (MAM) .The analysis exhibited that the Seasonal ARIMA model which is satisfactory in describing the monthly rainfall data in Dar es Salaam Tanzania is SARIMA (2, 1, 1)(1, 1, 1)12. The model was then used for predictions of monthly rainfall values from January 2015 to December 2024. The forecasting results showed that monthly rainfall values have a decreasing trends, hence that may be a threat to agriculturists and water managers in the region. The study will be useful to decision makers for the region of Dar es Salaam to establish priorities and strategies based on the impacts posed by the variation of rainfall.
