Classification of atmospheric circulation patterns in the Middle East and Iran

The main idea of ​​atmospheric circulation patterns and their classification is to convert a set of atmospheric parameters into a univariate catalog, with cases of similar characteristics grouped together. Easier interpretation of weather conditions would be the advantage of the classification.Many researchers to classify atmospheric circulation patterns have used the principal component analysis (PCA) method. Esteban (2006) in a study of atmospheric circulation patterns in Western Europe, categorized daily synoptic circulation patterns into 20 groups and showed sea level pressure and 500-hPa geopotential height configurations throughout the year. Smith and Sheridan (2018) clustered geopotential height anomalies at levels of 100 and 10-hPa and sea level pressure to study the patterns of tropospheric and stratospheric potential vorticity. In Iran, Raziei (2007) by PCA classification and clustering identified 18 atmospheric circulation patterns for the Middle East and Iran. He showed that patterns with meridional and northwest flow often cause drought and patterns with southwest flow cause wetting. Hanafi (1399) identified the air types in the northwestern region of the country (Maragheh station), and showed that the high geopotential systems of Saudi Arabia and North Africa governed by hot and dry conditions and the Mediterranean and polar cold air masses are involved in creating cold periods.Atmospheric circulation patterns identified for Iran are based on monthly average atmospheric data and with a resolution of 2.5°. Often due to the importance of winter as the main rainy season in Iran, the identification of atmospheric circulation patterns of other seasons has less studied. On the other hand, most studies have conducted regionally in a province/small part of the country. In this study, we have classified the synoptic patterns affecting Iran and the Middle East in the period of 1990-2019, using the PCA method for all seasons and with a higher accuracy using data with 0.5° resolution on a daily time scale. Then, composite maps of the MSLP (Mean Sea Level Pressure) and 500-hPa geopotential height were prepared and analyzed to obtain the frequency of monthly and seasonal occurrence of each investigated pattern. Previous studies have focused mainly on 500-hPa and MSLP parameters. However, in the present study we have used the 850-hPa moisture flux besides the already applied quantities, which can be considered a main and important quantity for the rain occurrence.

Materials and methods  :

Geopotential height data of 500-hPa received from the ECMWF (ERA5) at one hour intervals during the period 1990-2019 in the area of 20E-65°E longitude and 10N-55°N latitude. Due to the data volume and software and hardware limitations, data with 0.5° resolution extracted. By calculating the daily average in each grid, the data matrix consisted of 8281 columns (grids) and 10957 rows (number of days). Then the MSLP, 10m wind speed, and 850-hPa temperature and relative humidity maps/data were provided/extracted.The PCA analysis is a method for extracting important variables from a large set of variables in a data set. The main goal is to reduce number of components by finding the ones that explain correlations between the variables. We selected the first nine components covering a large portion of the described variances and ignored other minor components. In the present study, the R language programming were used to analyze the principal components with S array and select the components. To classify air types, K-means clustering was used, so that it presents the most alternating patterns of atmospheric circulation in the study area during the year. After selecting the main components and determining the scores of the components, all days (10957 days) were classified into 18 groups. Sea level pressure and 500-hPa geopotential height maps were prepared and interpreted. Monthly and seasonal frequency of occurrence of each pattern and relationship of spatial distribution of moisture flux at 850-hPa were investigated.

Based on the results obtained from principal component analysis, the first 9 components were selected, which explained 95.93% of the total variance of the data. According to the monthly and seasonal distribution of each pattern, it was observed that cp1, cp7 and cp18 circulation patterns occur in summer and early autumn and cp13 occur in late spring, summer and early autumn. Cp2, cp4, cp6, cp8, cp11 and cp16 occur in late autumn to early spring of the following year and cp3, cp5, cp9, cp10, cp12, cp14, cp15 and cp17 circulation patterns occur in late autumn to spring of the following year. These situations correspond to the synoptic conditions of each pattern at sea level and the level of 500-hPa.In the circulation patterns of the warm period of the year, thermal low pressure of sea level pressure map below 35° N is associated with subtropical high at mid-level of atmosphere. The maximum moisture flux in 850-hPa located in the eastern parts of Iran and the maritime borders of Yemen and Oman which is compatible with southern and southeastern flow. In cp2, cp4, cp6, cp8, cp11 and cp16 circulation patterns that occur in the cold period of the year; south and southwest flow of low pressure in the south of the Red Sea are associated with mid-level of atmosphere. There is convergence of moisture flux in the south of the Red Sea and in parts of the south and southwest of Iran, east of the Black Sea, northeast of the Mediterranean Sea and south of Turkey. In other patterns that occur in the cold period of the year as well as in spring, low pressure in south and center of the Red Sea is associated with trough in the mid-level of the atmosphere. Convergence of moisture flux was in the west, southwest, south an northwest of Iran, northeast of the Mediterranean Sea, southwest of Turkey and south of the Red Sea.

In the present study, principal component analysis with S array and K-means clustering was used to classify circulation patterns in Middle East and Iran. The results show significant differences in the arrangement of patterns, the frequency of air types and their path to Iran.