Identification of homogeneous precipitation sub-regions for Iran using principal component analysis

Delineation of homogeneous precipitation sub-regions featured with different time variabilities is very important for large countries such as Iran, which are characterized by complex topography and different climates. Very rare efforts have been devoted to identify modes of monthly precipitation variability in Iran and delineating sub-regions having different temporal variabilities of precipitation. On the other hand, most studies of precipitation regionalization in Iran have used very limited and unevenly scattered stations across the country; thus making it necessary to identify the most realistic precipitation sub-regions for Iran using almost all available stations. As such, 155 synoptic stations with relatively regular distribution over Iran, mostly having full data records for the 25 years common period of 1990–2014, were used for identifying an updated precipitation regionalization of the country. The cubic root transformed monthly precipitation of the considered stations were used as input for an S-mode principal component analysis (PCA) applied to the inter-stations correlation matrix (300×155) that is composed of 155 stations and 300 cubic root transformed monthly precipitation. The computed Kaiser–Meyer–Olkin measure of sampling adequacy for the considered matrix with a value of 0.98 indicates that the considered matrix is marvelous for a PCA application. The first five leading significant PCs accounting for approximately 80% of total variance of the dataset were considered for further analysis based on the Scree plot and the sampling errors of the PCs (North et al., 1982). To better characterize the underlying spatial structure of the considered data matrix, the retained PCs were then rotated using varimax orthogonal criteria. The five leading varimax rotated loadings were mapped to present spatial modes of monthly precipitation variability across the country and precipitation sub-regions borders were delineated using the maximum loading value approach (Comrie and Glenn, 1998; Miller and Goodrich, 2007; Chen et al., 2009).
The maps of varimax rotated loadings well represent areas characterized by different modes of precipitation variability and regimes. The five precipitation sub-regions identified using maximum loading values of the varimax rotated components are the Caspian Sea region, the northwestern, the western, the central-eastern, and the central-northeastern of the country. The Caspian Sea region featured with maximum precipitation in autumn and relatively regular distribution of precipitation throughout the year includes the coastal areas of the Caspian Sea and the northern faces of the Alborz Mountain in northern Iran. The north-western sub-region is distinguished from the rest of the country for its identical precipitation regime characterized by maximum precipitation in spring and relatively uniform precipitation all over the year. The three remained precipitation sub-regions of Iran are characterized with a much shorter rainy season, which maximizes in the winter time. The western sub-region encompasses mountainous areas of western Iran as well as the lowlands of the southwestern country. The central-eastern sub-region differs from the western sub-region due to its shorter rainy season and much lower precipitation values in all of the months, but similarly, its maximum precipitation occurs in January. Finally, the central-northeastern precipitation sub-region receives its maximum precipitation in March as opposed to the two aforementioned sub-regions which peak in January. The independence of the identified precipitation sub-regions was examined by applying the Kolmogorov–Smirnov non-parametric test to the regional anomalies of annual precipitation series; the result proved that all the sub-regions are statistically different at 99% confidence level. The identified precipitation sub-regions can serve as a tool for a better water resources management in the country.