Change point analysis of mean annual air temperature in Iran

Abstract

The existence of change point in the mean of air temperature is an important indicator of climate change. In this study, Student's t parametric and Mann–Whitney nonparametric Change Point Models (CPMs) were applied to test whether a change point has occurred in the mean of annual Air Temperature Anomalies Time Series (ATATS) of 27 synoptic stations in different regions of Iran for the period 1956–2010. The Likelihood Ratio Test (LRT) was also applied to evaluate the detected change points. The ATATS of all stations except Bandar Anzali and Gorgan stations, which were serially correlated, were transformed to produce an uncorrelated pre-whitened time series as an input file for the CPMs and LRT. Both the Student's t and Mann–Whitney CPMs detected the change point in the ATATS of (a) Tehran Mehrabad, Abadan, Kermanshah, Khoramabad and Yazd in 1992, (b) Mashhad and Tabriz in 1993, (c) Bandar Anzali, Babolsar and Ramsar in 1994, (d) Kerman and Zahedan in 1996 at 5% significance level. The likelihood ratio test shows that the ATATS before and after detected change points in these 12 stations are normally distributed with different means. The Student's t and Mann–Whitney CPMs suggested different change points for individual stations in Bushehr, Bam, Shahroud, and Gorgan. However, the LRT confirmed the change points in these four stations as 1997, 1996, 1993, and 1996, respectively. No change points were detected in the remaining 11 stations.