Abstract: In this paper, a time series analysis is adopted as a
technique for forecastingofmonthlyrainfall and number of
rainy days. The data sets are recorded by the meteorological department of
Chiang Mai from January 1982 to August 2006. The Box-Jenkins technique is used
for identifying theparametersofanAutoregressiveIntegratedMoving Average(ARIMA)model.TheAkaikeinformationcriterion,the Schwarz’s Bayesian criterion
and the mean square error are used throughout to test for simplification of any
particular model. The periodogram analysis is used to confirm the existence of a
seasonal period in the ARIMA model. The results from time
series analysis showed that the fitted model for the rainfall is theand the fitted model for the number
of rainy days is the with no constant included in the
models. The results from the periodogram analysis showed that the rainfall and
the number of rainy day series contained significant periodic component of 12
months which are the same results of the seasonal phenomena of ARIMA Box and
Jenkins models. The ARIMA with seasonal model possibly predicts the monthly
rainfall and the number of rainy days one year ahead with acceptable accuracy.
Keywords and phrases: autocorrelation, Box-Jenkins model, partial autocorrelation, periodogram analysis, rainfall, time series analysis.
