%0 Journal Article %T Assessment of homogenization of climate data and trend of temperature %J Journal of the Earth and Space Physics %I Institute of Geophysics, University of Tehran %Z 2538-371X %A Rafati, Somayeh %A Karimi, Mostafa %D 2018 %\ 04/21/2018 %V 44 %N 1 %P 199-214 %! Assessment of homogenization of climate data and trend of temperature %K Temperature trend %K Data homogenization %K PMFred algorithm %K Temperature trend pattern %R 10.22059/jesphys.2017.61674 %X Climate data series usually contain artificial shifts due to inevitable changes in observing instrument or observer, location, environment and observing practices/procedures taking place during the period of data collection. Data discontinuities also arise from the continuously evolving technology of climate monitoring. It is important to detect artificial changepoints in climate data series, because these artificial changes could considerably bias the results of climate trends and variability analysis. Thus, corrections and homogenization of climate data are imperative for the assessment of observed climate trends. In this study homogenization of mean monthly temperature was assessed for 33 synoptic stations in Iran using PMFred algorithm and also linear trend estimates were obtained using this algorithm. The p value of the linear trend was determined by the t-test statistic of the slope parameter. The p value is the probability for an estimated positive trend to be greater than zero, or for an estimated negative trend to be smaller than zero. The probability for the estimated trend to be within these intervals is 95%. Linear trend was estimated for raw and homogenized data in order to evaluation of homogenization effect on trend analysis. Linear trend was normalized via half of confidence interval (95% confidence level) so that absolute value of significant trend (at this confidence level) would be greater than one. Then distribution map of mean monthly temperature trend was provided. This study showed that assessment of homogenization using an absolute test can lead to wrong results without the usage of adjacent stations data comparison, if there is no complete and reliable metadata. Because absolute homogenization tests could not realize between natural and artificial shifts and thus should not be used automatically and without subjective qualitative check. Thus adjacent stations data along with metadata (if it exists) was used for the detection of artificial shifts. Mean monthly temperature data was recognized homogeneous in Tehran, Shiraz, Esfahan, Hamedan-Nojeh, Tabriz, Khoy, Oromieh, Sabzevar, Shahrood, Babolsar and Bandar-Anzali stations and it was recognized inhomogeneous in Zanjan, Saqez, Sanandaj, Kermanshah, Khoram-Abad, Shahrekord, Ahvaz, Abadan, Yazd, Bandar-Abbas, Bam, Kerman, Zahedan, Zabol, Mashhad, Torbat-Heydarieh, Gorgan, Ramsar, Rasht, Qazvin, Birjand and Arak Stations. The results showed that the estimates could be biased by the unaccounted shifts in the series as expected. In the other words, it was observed negative trend before adjustment in mean monthly temperature in many stations which have inhomogeneous data, while they showed positive trend after adjustment (Torbat-Heydarieh, Birjand, Zabol, Gorgan, Bandar-Abbas, Khoram-Abad, Shahrekord, Ahvaz, Zanjan, Rasht, Qazvin, Saqez stations). Estimation of linear trend for homogenized data revealed that mean monthly temperature has increased significantly in most stations in Iran. Also, it has not been increased significantly in northwest, except Tabriz station and in Sabzevar- Shahrud to Bandar-Abbas, in a north-south direction. Also a north-south pattern was observed in intensity of increased trend in Iran. That is temperature has not increased in the northwest, while it has increased in north to central and southwest of Iran relatively severely (about 0.003 degrees Celsius in each month). It has not increased significantly in east of this region. Also, it has increased in east of Iran severely.  %U https://jesphys.ut.ac.ir/article_61674_a9d1b5326aceb702c9654dfeb3e1074a.pdf