عنوان مقاله [English]
Daily precipitation records of Mehrabad synoptic station based in Tehran, for the period 1951–2013 was used to identify moderate to heavy cold weather precipitation events in the mainly rainy season of Iran which starts in October and ends in May. Mehrabad is one of the oldest stations in the country that holds the longest and most complete precipitation records available in the country with very few missing values; thus being suitable for identifying the types of precipitation events for the region and the associated atmospheric circulations. Following the Iranian Meteorological Organization definition, we identified moderate and heavy precipitation events for Tehran Province as the events for which total daily precipitation ranges from 5 to 20 mm and from 20 to 50 mm, respectively; but being characterized with anomalous cold weather conditions. This screening approach has resulted in a set of 133 days of moderate to heavy precipitation events featured with cold weather conditions, which is adequate for implementing a multivariate analysis. The 500 hPa geopotential height and relative vorticity, sea level pressure (SLP), 850 hPa wind field and advection of specific humidity at 00 UTC over the time period considered (October–May), covering a large geographical domain centred on Iran (20°E–70°E, 20°N–55°N) with a 2.5° latitude × 2.5° longitude spatial resolution were retrieved from the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) reanalysis archive (Kalnay et al., 1996; Kistler et al., 2001).
In the present study the S- and T-mode Principal Component Analysis (PCA) were used for classifying the 500 hPa atmospheric circulations associated with the 133 precipitation events. The S-mode PCA with correlation as a similarity measure was used as a data reduction tool and pre-processor of K-means clustering method, while a T-mode PCA with correlation as a similarity measure was employed to classify 500 hPa atmospheric circulations independently. Based on the scree plot (Cattel, 1966) and the sampling errors of the eigenvalues (North et al., 1982) five and six PCs were retained for, respectively, for the S- and T-mode PCA applications. The retained PCs were orthogonally and obliquely rotated using varimax and promax criteria, respectively. For an S-mode PCA, we used varimax rotated PC scores as input for K-means clustering, resulting in 5 circulation types (CTs). But applying a T-mode PCA coupled with varimax (promax) rotation classified all the considered days into six CTs. The skills of K-means clustering and un-rotated, varimax and promax rotated T-mode PCA in classifying atmospheric circulations were examined using some indicators measuring the separability and equability of the identified groups of each classification method. The results suggest that the obliquely rotated T-mode PCA outperforms both K-means clustering and orthogonally rotated T-mode PCA in classifying atmospheric circulations.
Each of the six CTs identified are capable of producing significant precipitation in Tehran, but all cases of heavy daily precipitation above 40 mm belong to the CT1 and CT2. Although various forms of tilt in mid-tropospheric geopotential trough are observed among the CTs, but the dominant tilting is that of the northeast–southwest direction, indicating the anti-cyclonic wave breaking. Except CT5, the CTs are associated with a dipolar structure in surface temperature anomaly consisting of a pair of negative and positive anomalies to the west and east of the country, respectively.