Using ARIMA Models to Predict Annual Rainfall Values in KASAB Station in Northwestern Syria During the period from 1960 to 2011

Abstract

     The water sector in Syria faces a major challenge, as precipitation changes are associated with declining precipitation values due to climate changes. Rainfall is one of the difficult and complex elements of hydrological cycle, due to the numerous and overlapping meteorological factors that cause rainfall.

The research problem is that predicting rainfall amounts on the Syrian coast is generally undetermined. The research aims to predict the future amounts of annual rainfall by developing a statistical model based on time series analysis using the statistical software Minitab.

Data from 51 years of rainfall were used, with 43 years during the period from 1960 to 2011  for model developing, and eight years for testing. The most suitable model was found to be ARIMA(3,1,5), according to the Akaki criterion after conducting the required tests. The test results showed good performance accuracy. The ARIMA model (4,1,5) was adopted to predict annual rainfall values for the next twenty years.