Forecasting the Rainfall in Muang, Nan Province

Abstract

Rainfall is an important factor for agricultural cultivation. Study on past rainfall records using statistical methods to forecast the rainfall is essential for the planning and devising a policy for agricultural extension. Therefore, the objective of this research was to study the appropriate forecasting model for the rainfall in Muang, Nan province. The data used in this research was the average monthly rainfall from the Hydrology and Water Management Center for Upper Northern Region - Royal Irrigation Department during January 1998 to August 2013 (188 values). The time series data were divided into two categories. The first 180 values from January 1998 until December 2012 were used for the modeling by Box-Jenkins method, simple seasonal exponential smoothing method, and combined forecasting method. The last 8 values from January to August 2013 were, on the other hand, used for checking the accuracy of the forecasting models via the determination of the lowest root mean square error. Research findings indicated that for all the forecasting methods that had been studied, combined forecasting method is the most suitable for this time
series and the forecasting model is: Ŷt = 2.71022 ‒ 0.14376Ŷ1t + 1.11493Ŷ2t where Ŷ1t and Ŷ2t represent the single forecasts at time t from Box-Jenkins method and simple seasonal exponential smoothing method, respectively.