A Time Series Approach for Precipitation in Turkey

Abstract

The climate that has a dynamic structure varies continuously. Basic climatological features of any region depend on many meteorological variables, such as precipitation and temperature records. The changes in these two variables reveal important clues about climate variations. Turkey is one of the sensitive areas for climate variation around the world. In this study, statistical analysis is applied to historical data set consisted of monthly precipitation totals (mm.) which are provided by the Turkish State Meteorological Service(TSMS) for 1999-2010 period in Turkey. Furthermore, a sinusoidal model for each region obtained by using a time series method is proposed in order to obtain a sinusoidal form for the precipitation to each region. Besides a basic form of Frequency Domain Techniques, usual Box-Jenkins ARIMA approach is applied in this study. In conclusion, it is observed that the sinusoidal model predictions are more appropriate than ARIMA techniques and usual generalized linear model (GLM) techniques.