Use of the spectral clustering to determine coherent precipitation regions in Turkey for the period 1929–2007



In this study, we suggest the spectral clustering (SC), a hybrid clustering technique based on singular value decomposition (SVD) and K-means for grouping features of precipitation totals of 96 stations in Turkey. Clustering process establishes an exhaustive set of occupied regimes into distinct climatic zones. Results of the SC satisfactorily represent the influences of the synoptic-scale weather systems including such as the mid-latitude and Mediterranean frontal cyclones, and the mid-latitude travelling and eastern Europe high pressures in winter, sub-tropical Azores high pressure and monsoon low in summer. Results of the SC also well display the influences of local-scale atmospheric disturbances, and direct influences of physical geographical features of Turkey (i.e. exposure, topography, orography, land-sea distribution, continentality and the high Anatolian peninsula) on the geographical variability and coherent distribution of the annual precipitation totals over Turkey. Finally, based on the results of the SC method employed to annual precipitation totals of 96 stations in Turkey for the period of 1929–2007, eight clusters of precipitation coherent zones are determined, namely Black Sea, Northwest Turkey, southern Aegean and western Mediterranean, Mediterranean, West Continental Central Anatolia, East Continental Central Anatolia, Continental eastern and south-eastern Anatolia. Copyright © 2010 Royal Meteorological Society

1. Introduction

In most cases, for the climatological and meteorological applications, comprehensive information is required with a regionalized spatial variability of climate data. One of the regionalization methods is clustering technique. Cluster analysis is based on some form of similarity matrix for the clustering of climatic time-series attempts to find the groups of datasets having similar characteristics. These groups can then be further analysed in detail to gain insight from the common characteristics of the datasets in each group of climatic sub-regions. For that reason, cluster analysis in climatology/meteorology is to define classes of synoptic types or climate regimes, or stations and/or grid points to define climatic sub-regions (Lund, 1963; Anderberg, 1973; Richman and Lamb, 1985; El-Kadi and Smithson, 1992; Fovell and Fovell, 1993; Wilks, 1995; Bunkers et al., 1996; von Storch and Zwiers, 1999; Spellman, 2000; Ünal et al., 2003; Gordon et al., 2005; Donna et al., 2007; Bisgin and Dalfes, 2008; van Cooten et al., 2009; Kucken et al., 2009).

In many fields, spatial distribution of precipitation amounts is required. The spatial distribution of precipitation treats clustering data. For example, hydrologists and groundwater hydrologists ask precipitation values corresponding to their catchments areas of interest. The ecologists who are studying the dynamics and responses of forests and various species in the mountainous terrain need to know features of precipitation. It is evident to increase such examples.

The climate of Turkey and its nearby regions is characterized mainly with the Mediterranean macro climate governed by an unequivocal seasonal alteration between the mid-latitudes and sub-tropical/tropical pressure and wind systems (circulation patterns) from winter to summer (Erinç, 1969; Trigo et al., 2006; Türkeş, 1996, 1998; Tatli et al., 2004, 2005; Türkeş and Erlat, 2003, 2005). There is, therefore, a highly seasonal precipitation regime in the western and southern regions, continental semi-arid climate in the Central and eastern Anatolia regions, and humid-temperate, uniformly rainy climate in the northern Marmara and the Black Sea regions of Turkey (Erinç, 1969; Türkeş, 1996, 1998, 1999).

The precipitation climatology, drought events and their spatial distribution and long-term variability, trends and changes in precipitation series over Turkey have been comprehensively analysed previously, for example by Türkeş (1996, 1998, 1999, 2003), Türkeş et al. (2002), Tatli et al. (2004), Aksoy et al. (2007); Tatli and Türkeş (2008), Türkeş and Tatli (2009), Türkeş et al. (2009) and Saris et al. (2010). Many studies indicate that precipitation regimes and variability in Turkey with the various time-scales are related with the locations, variations and activity of the atmospheric systems over the Mediterranean and beyond, except in summer months (e.g. Türkeş, 1998; Kutiel et al., 2001; Türkeş et al., 2002; Türkeş and Erlat, 2003, 2005, 2006; Tatli et al., 2004; Kutiel and Türkeş, 2005; Tatli, 2006; Türkeş et al., 2009).

Then again, some recent studies revealed that the North Atlantic Oscillation (NAO) could be one of the major atmospheric controls for the spatial and temporal variability of precipitation series over Turkey including the significant wet periods and meteorological droughts (Türkeş and Erlat, 2003, 2006; Tatli, 2006; Trigo et al., 2006).

In a recent study, Türkeş et al. (2009) have investigated spatial relations among the precipitation variations of Turkish stations by principal component analysis (PCA), making use of the knowledge that explains influences of the major physical geographical control mechanisms of Turkey, which are consist of the regional-scale circulation, topography, land-sea distribution, exposition and continentality, on regional rainfall regime types as the basis for spatial assessments over Turkey.

The main results of Türkeş et al. (2009) are briefly given in the following in order to bridge them with this study. In the relevant study, they found that the eigenvalue of the first principal component (PC1) is the highest in winter, and generally explains the variations of spatial relationships in precipitation series that are closely influenced by large-scale weather systems in west and relatively smaller-scale systems in the east controlled by the regional physical geographical factors. They have also underlined a relatively weak correlation pattern seen over the middle-east and north-eastern Anatolia sub-regions could be a result of the influence of dominant weather systems associated with the large-scale atmospheric circulation at the west half of the country, and forcing effects of nearly west to east trending high ranges of the northern Anatolian Mountains to orographically lift of the maritime polar air masses over these mountain ranges. On the other hand, the second PC pattern of the winter precipitation characterized with the highest loadings over the Continental Mediterranean and Continental Eastern Anatolia rainfall regions could be explained by the influences of humid-temperate and thus conditionally unstable Mediterranean weather systems on the variability of spatial relations. Türkeş et al. (2009) have found that the PC3 pattern treats humid-temperate influences of the Mediterranean Sea on winter precipitation totals by making stronger direct influence of the mid-latitude cyclones over the Mediterranean region.

They have stated that the PC1 of summer explains direct influences of mid-latitude cyclones related with the large-scale atmospheric circulations reaching from eastern Europe to Turkey. Accordingly, the loadings of PC2 established over western Black Sea sub-region could explain orographic rainfalls well over the region caused by the forcing of north-westerly air flows carrying from the Atlantic sourced polar air to the north of Turkey. However, the loadings, mainly over the continental type rainfall regions of Turkey, very likely explain the influences of regional and/or local convective instability over these regions in summer. They have also revealed that the loadings of PC3 over the eastern Black Sea and north-eastern Anatolia sub-regions of Turkey, which are the rainiest districts in summer, satisfactorily explain the influences of coastal orographic and continental convective instability showers and thunderstorms, both of which are closely associated with northerly sector surface and upper-air flows and atmospheric disturbances; and they have concluded that the regional surface warming contributes to strengthen that mechanism in the summer months.

Even though this study involves an objective classification of Turkish precipitation series by using the spectral clustering (SC) method, the information and maps of precipitation climatology of Turkey such as the spatial variations of long-term precipitation averages, coefficients of variation, seasonality of precipitation totals, drought events and regions and drought probabilities are not given in this study, because the detailed assessments for seasonal and annual precipitation climatologies had been performed previously, and one may find in the works of Türkeş (1996, 1998, 1999, 2003), Tatli et al. (2004), Türkeş and Tatli (2009) and Türkeş et al. (2009).

This paper includes the following aspects: after Section 1, the major features of SC method and data used in the study are described in Section 2. Results of the study are presented in Section 3, and some comments and discussion are given in Section 4, and finally conclusions are given in Section 5.

2. Data and methodology

2.1. Precipitation data

The data set used in this study includes yearly precipitation totals (millimetres) recorded at the stations of the Turkish State Meteorological Service. The precipitation data set has been developed by Türkeş (1996, 1998) for 99 stations covering the period 1929–1993, and 96 of these stations were updated to 2007 for this study (Figure 1). This station-based precipitation data set has the most continuous/homogeneous and longest precipitation time-series of Turkey, and they mostly have 65 to 77-years' records and represent a very good spatial distribution over Turkey. The required information for the data set and homogeneity analyses applied to the long-term precipitation series can be found in Türkeş (1996, 1998, 1999).

Figure 1.

Spatial distribution of 96 stations used in the study (names of the meteorology stations are abbreviated) over the rainfall regime regions of Turkey. BLS: Black Sea; MRT: Marmara Transition; MED: Mediterranean; MEDT: Mediterranean Transition; CMED: Continental Mediterranean; CCAN: Continental Central Anatolia and CEAN: Continental Eastern Anatolia

2.2. Methodology

The employed technique of SC in this paper builds on observations presented in (Ng et al., 2002; Von Luxburg, 2007). Here, we briefly review the algorithm of the so-called Ng–Jordan–Weiss (NJW algorithm) as in the following:

Given a set of X = {x1, …, xm} that we want to classify them into k clusters then

  • 1.Form the affinity (similarity) matrix Cm×m to be defined by
    equation image(1)
    where the scaling parameter s given in Equation (1) controls how rapidly the affinity matrix C falls off with the distance between x1 and x2. Since we have standardized the precipitation data to have zero mean and variance one, we have selected s = 1.
  • 2.Define E being the diagonal matrix whose (j,j)-element is the sum of C's j-th rows defined by
    equation image(2)
  • 3.Construct matrix of L as
    equation image(3)
  • 4.Find u1, u2, …, uk, the first k largest eigenvectors by singular value decomposition (SVD) of the matrix L (k < m), and assume to form the matrix Um×k = [u1u2uk] by ordering the eigenvectors in columns.
  • 5.Form the matrix P from U by renormalization of each row of U's to have unit length as
    equation image(4)
  • 6.Treating each row of P as a point in k-dimensional space, and cluster them into k clusters by using the traditional clustering techniques of K-means (MacQueen, 1967).
  • 7.Finally, assign the original point xi to cluster j if and only if row i of the matrix P was assigned to cluster j.

Technically, there are some methods for deciding on the number of clusters. Moreover, nearly all techniques for visualizing multivariate data can also be used for cluster visualization. Among the separation measures, validity index (VI) and cophenetic correlation (C) are well-known measures (Sokal and Rohlf, 1962; Dunn, 1974; Davies and Bouldin, 1979; Bezdek and Pal, 1998; Ray and Turi, 1999). Besides, Volkovich et al. (2008, 2010) have suggested another external cluster validation index with the misclassified quantities obtained in the process of repeated clustering based on the measuring of cluster stability.

For instance, in the K-means problem, assume that a set of X = {x1, …, xm} in m-dimensions is given. The goal is to arrange these points into K clusters, with each cluster having a representative point Z = {z1, …, zk}, usually chosen as the centroid of the points in the cluster. The energy of each cluster (Ek) is then obtained as

equation image(5)

For a given set of clusters, the total energy is then simply the sum of the cluster energies Eks. The goal is to choose the clusters in such a way that the total energy is minimized. Usually, a point xi goes into the cluster with the closest representative point zk. Hence, to define the clusters, it is enough simply to specify the locations of the cluster representatives. On the other hand, similar to the measure of cluster energies, the energy within clusters (Pjs) could be defined, where the clusters being as far as possible away from each other as

equation image(6)

Ray and Turi (1999) suggest a measure of the so-called validity index (VI) based on cluster energies:

equation image(7)

where Ray and Turi (1999) have named the term of equation image as intra and min (Pk) as inter, respectively. That is, by the notations of Ray and Turi (1999), VI is defined by

equation image(8)

where equation image indicates the metric of norm. Obviously, VI defined above is going to be minimized where we will have minimized cluster energies (i.e. Ek's) and maximized within cluster energies (i.e. Pk's) values. Figure 2 shows a comparison of the performances of the SC with K-means. In this figure, the original pattern is formed from nested two circles, while SC is able to extract these two circles perfectly; however, the K-means classifies the pattern into two half circles, on the other hand.

Figure 2.

Clustering example of two circles, with clusters indicated with different symbols. (a) Spectral clustering, (b) K-means cluster

3. Results

The names and spatial distribution patterns of major rainfall regime regions of Turkey developed by Türkeş (1996, 1998) were used in spatially assessing results of the SC applied to annual precipitation total series measured at 96 stations over Turkey (Figure 1). This previous well-known classification of Turkish rainfall regime regions was realized with respect to the seasonality and physical geographical characteristics of monthly and seasonal precipitation totals. In this study, our aim is to extract the major spatial dependency of precipitation by SC in the following analyses. We have applied the VI measure for deciding the number of precipitation clusters. After employing the VI measure for clustering precipitation data, we have had eight clusters, but here the results of clusters are interpreted separately from 2 to 8 in order to bridge the rainfall regimes with the obtained clusters under view of not only statistical measure but also using common climatological and physical geographical features. The abbreviations of cluster-2, cluster-3,… and cluster-8 used in the following text indicate clustering the precipitation data into 2, 3, …, 8 clusters, respectively.

3.1. Cluster-2

According to the results of the regionalization of Turkish annual precipitation totals by the SC method, the territory of Turkey first separates into two large rainfall regions with the exception of a small area over the Iskenderun Bay and the Amanos Mountain of the eastern Mediterranean sub-region of the Turkish Mediterranean coastal region (Figure 3). The resulting main western part of Turkey with the number one cluster region includes western Black Sea sub-region, both western sub-regions of the Continental Central Anatolia and the Continental Mediterranean rainfall regions along with the Marmara Transition, the true Mediterranean and the Mediterranean Transition rainfall regime regions (Figure 3). The spatially coherent western precipitation region with the first cluster of the cluster-2 is mainly explained with precipitation occurrences and variations controlled closely by the Northeast Atlantic and the Mediterranean originated large-scale pressure and wind circulation patterns. These patterns are connected to those of synoptic-scale weather systems mostly including mid-latitude cyclones and anticyclones, sub-tropical anticyclones and circulation-based influence of tropical weather systems such as north-western extension of the southern Asiatic monsoon low associated with the northward seasonal migration of the inter-tropical convergence zone (ITCZ) during the summer months, and so on. Consequently, this precipitation region is closely characterized mostly by the mid-latitude and the Mediterranean cyclones and associated frontal rainfall events over Turkey (Türkeş, 1996, 1998; Kutiel et al., 2001; Türkeş et al., 2002, 2009; Türkeş and Erlat, 2003, 2005, 2006; Tatli et al., 2004; Kutiel and Türkeş, 2005; Tatli, 2006, and so on). Furthermore, orographic lifting of the air masses over mountainous areas of the western Black Sea sub-region and the Mediterranean region have a great importance for precipitation occurrence, while precipitation occurrence and variations over the continental inner regions are also dominated mainly by the local and/or regional convective instability rainfall events in spring and early summer months of the year.

Figure 3.

Geographical distribution patterns of the cluster-2 results extracted from the spectral clustering approach applied to annual precipitation total series measured at 96 stations over Turkey. This figure is available in colour online at

On the other hand, the ensuing main eastern part of Turkey with the second cluster of the cluster-2 is determined strongly by influences of the weather systems originated from the further northern and the eastern European regions along with the western Siberia region of the great Asia continent. The rainfall events over that eastern region of Turkey are developed by controls of the mid-latitude cyclones, particularly their cold fronts extending towards further southern regions of Turkey, northerly circulation types and the convective instability events over the continental inner regions of the high Anatolian peninsula (Türkeş, 1996, 1998, 1999; Kutiel et al., 2001; Tatli et al., 2004; Türkeş and Erlat, 2005, 2006; Tatli, 2006; Türkeş et al., 2009).

The precipitation occurrence and variability over the humid-temperate Black Sea region are significantly contributed by orographically induced rainfall events related with the northerly circulations mostly ignoring whether they are cyclonic or anticyclonic flows in nature. This region, at the same time, except the Mediterranean Taurus Mountains at the west, is the highest elevation area of Turkey's territory. With the exception of the Black Sea rainfall region of eastern Turkey region, the mountainously higher elevation Northeast Anatolia sub- region of the region receives most of the precipitation amount in the late spring and summer months. The primary reason of this clear situation is of the higher territory, late warming of the surface soils of the area due to the continentality and higher elevation of the area with snow occurrences heavily in winter, and convective instability events having controlled by the lateness activity of upper-air atmospheric disturbances (e.g. troughs or lows). This is the primary and leading atmospheric factor.

3.2. Cluster-3

In the spatial pattern of cluster-3, the previous regionalization pattern derived from the cluster-2 increases to three rainfall regions including mainly the continental inner region of Turkey (Figure 4). With respect to this spatial pattern, the region of the first cluster is taken as a region of reflecting influences of the large-scale weather systems particularly related to those of the Northeast Atlantic, the Mediterranean and the northern and eastern European originated systems. Second cluster precipitation region is influenced by both of the eastern Mediterranean cyclones and the eastern European–western Siberia high pressures and their associated circulation types in winter, and the monsoon low-pressure related tropical circulation via dry and hot Mesopotamia region in the summer months. The third cluster region of the cluster-3 is mainly the result of spring and early summer time rain showers and thundery showers (precipitation producing thunderstorms) controlled by the regional scale convective instability events over that region with the exception of the Black Sea rainfall region.

Figure 4.

As in Figure 3, but for the cluster-3 results of annual precipitation totals. This figure is available in colour online at

3.3. Cluster-4

The regional pattern of cluster-4 is the one that is spatially divided into two parts of the previously extracted western and eastern regions of Turkey in the clustering level-two of regionalization (Figure 5). In other words, this pattern gives a facility to make a detailed examination of the pattern of these two rainfall regions. Consequently, number four of the SC rainfall region that corresponds to Northwest Turkey basically reflects the influences of the Northeast Atlantic and partially Mediterranean originated weather systems. Most of the precipitation amounts over that region are mostly produced under the control of the mid-latitude cyclonic variability. Effect of the orography is evident on the western Black Sea sub-region. The first cluster rainfall region of the cluster-4, which can be called as the south and south-west Turkey, generally corresponds to the true Mediterranean rainfall regime region with a hot and dry long summer and rainy and temperate winter along with a high climatic seasonality as a whole. Precipitation occurrences and variability over this region are closely controlled by the eastern Mediterranean basin originated frontal low-pressure systems and the Azores high-pressure systems. Some of the precipitation occurrences and variability are associated with frontal lifting, while some particularly over the Mediterranean Taurus Mountains and coastal belt of the Anatolian peninsula are explained with cloud and precipitation occurrences related to orographic lifting. Meteorologically, precipitation occurrences are mostly characterized by the frontal showers and thundery showers associated relatively large-scale frontal thunderstorms over the region.

Figure 5.

As in Figure 3, but for the cluster-4 results of annual precipitation totals. This figure is available in colour online at

The inner-southwest area at north of the Antalya Bay and the southern-west of the Continental Central Anatolia rainfall regime region have the precipitation occurrence and variability that is also significantly supported by the rainfall showers due to the convective instability over that region. The third cluster of precipitation region, which covers the western and eastern Black Sea sub-region, eastern and northern parts of the Continental Central Anatolia, and so on can be explained by both the northern and eastern European originated weather systems at the north (i.e. the Black Sea Region) and orography and convective instability events in the Central Anatolia region. These activities are also facilitated by northerly upper-air troughs and low centres. Precipitation events are mostly in the forms of rain showers and thundery showers with the thunderstorms at the inner regions, and in the forms of strong and long-lasting rainfall and rain showers in the Black Sea region. Snowfalls are of great importance over the high elevation continental regions and the North Anatolia Mountains.

The second cluster of precipitation region of the cluster-4 pattern, which is called as eastern and south-eastern Turkey, mainly reflects the influences of the weather systems that originated from the eastern European—western Siberia—Caspian Basin in winter, and of the weather systems from the south-western Asia and the monsoonal Asia. This region, with the exception of low plateaus and the Tigris River plains staying at the southern part of the south-eastern Taurus Mountains, is of the highest and the most mountainous area of Turkey. In the northern part of this region, snowfalls and snowy conditions in winter and the rain showers and thundery showers in summer take an important aspect of this precipitation region.

3.4. Cluster-5

The most attractive aspect of the results of spectral cluster-5 (Figure 6) of the precipitation regionalization of Turkey is that the number one Mediterranean precipitation region extracted from the cluster-4 level is divided into two parts with a small change occurs in the Izmir district: the western and the eastern Mediterranean regions. Western Mediterranean precipitation region is closely characterized by the westerly and south-westerly weather systems, while the eastern Mediterranean region reflects the influences of southerly and south-easterly weather systems. The eastern Mediterranean region has the stronger summer dryness (long-lasting climatological drought) with the higher maximum temperatures than those in the western Mediterranean region.

Figure 6.

As in Figure 3, but for the cluster-5 results of annual precipitation totals. This figure is available in colour online at

3.5. Cluster-6

The resultant pattern of the spectral cluster-6 (Figure 7) appears as the fractions of the number four region (i.e. Northwest Turkey) of the cluster-4 and the re-occurrence of the Black Sea precipitation region except its western part. This distribution pattern may be accepted to re-appear at the Mediterranean to Black Sea transition from the Marmara and the northern Aegean regions.

Figure 7.

As in Figure 3, but for the cluster-6 results of annual precipitation totals. This figure is available in colour online at

3.6. Cluster-7

Regionalization results of the spectral cluster-7 (Figure 8) give a precipitation region pattern of Turkey that has the clear similarity with both Turkey's conventional geographical regions and Türkeş's (1996, 1998) precipitation regime regions of Turkey. Consequently, in addition to the influences of the synoptic- or regional-scale weather systems, this pattern from cluster-7 analysis of Turkish annual precipitation data is closely associated with the direct influences of physical geographical features of Turkey (i.e. exposure, topography, orography, land-sea distribution, continentality and the higher peninsula). On the basis of the results of cluster-7 analysis, and the names of Turkey's seven geographical and rainfall regime regions and their spatial patterns, cluster-7 precipitation regions are given in the following names.

  • 1.Black Sea precipitation region (the 6th cluster of cluster-7);
  • 2.Northwest Anatolia—Thrace precipitation region (western Black Sea—Marmara—northern Aegean) (the 4th and 7th clusters of cluster-7);
  • 3.Southern Aegean and western Mediterranean precipitation region (the 1st cluster of cluster-7);
  • 4.Mediterranean precipitation region (the 5th cluster of cluster-7);
  • 5.Continental eastern and south-eastern Anatolia precipitation region (the 2nd cluster of cluster-7).
Figure 8.

As in Figure 3, but for the cluster-7 results of annual precipitation totals. This figure is available in colour online at

3.7. Cluster-8

This cluster-8 precipitation pattern (Figure 9) is very similar with the cluster-7 pattern of Turkey's precipitation totals with some differences. In the cluster-8 pattern, the Continental Central Anatolia precipitation region is divided into its western and eastern parts, and it is interesting to see that the Black Sea region is delimited mainly with the coastal region as in Türkeş (1996, 1998). Meanwhile, it is also observed that the north-western Anatolia—Thrace (shortly Northwest Turkey) precipitation region has now much more particles. The western Continental Central Anatolia precipitation region is mostly characterized by the influences of variability of the westerly and south-westerly weather systems from the Europe, Balkans and the Mediterranean regions, whereas the Eastern Continental Central Anatolia precipitation region is mostly explained by the influences of variability of the northerly weather systems and further convective instability events and snowfalls over the region.

Figure 9.

As in Figure 3, but for the cluster-8 results of annual precipitation totals. This figure is available in colour online at

Finally, in this study, following precipitation regions of Turkey are extracted from the SC method applied to the annual precipitation totals of 96 stations of Turkey for the period of 1929–2007 (Figures 9 and 10):

  • 1.Black Sea (BLS) precipitation region (the 6th cluster of cluster-8);
  • 2.Northwest Turkey (NWTR) precipitation region (the 7th and 8th clusters of cluster-8);
  • 3.Southern Aegean and western Mediterranean (SAEG-WMED) precipitation region (the 1st cluster of cluster-8);
  • 4.Mediterranean (MED) precipitation region (the 5th cluster of cluster-8);
  • 5.West Continental Central Anatolia (WCCAN) precipitation region (the 4th cluster of cluster-8);
  • 6.East Continental Central Anatolia (ECCAN) precipitation region (the 3rd cluster of cluster-8);
  • 7.Continental eastern and south-eastern Anatolia (CEAN-CSEAN) precipitation region (the 2nd of cluster-8).
Figure 10.

Resultant precipitation regions of Turkey re-drawn by simplifying spatial variations of the cluster-8 results of the spectral clustering analysis of annual precipitation totals measured at 96 stations in Turkey: BLS: Black Sea; NWTR: Northwest Turkey; SAEG-WMED: southern Aegean and western Mediterranean; MED: Mediterranean; WCCAN; West Continental Central Anatolia; ECCAN: East Continental Central Anatolia and CEAN-CSEAN: Continental Eastern and South-eastern Anatolia are the special regional names in Turkey. This figure is available in colour online at

By evidently regarding the similarity of its spatial distribution pattern to the precipitation regime regions of Turkey, represented mainly by the seasonal precipitation variability and physical geographical features of Turkey, these newly extracted regions of Turkey (Figure 10) may also be considered as the precipitation regime regions of Turkey produced by the SC method.

4. Discussion

Among others, precipitation is the main driver of the variations in the water balance in space and time, and changes in precipitation amounts and variability have very important implications for hydrology and water resources (Compagnucci et al., 2001). Hydro-climatological variability over time in a catchment or a geographical region is mainly influenced by daily, seasonal, annual and decadal time-scale variations in precipitation amounts and intensity. Furthermore, river flood frequency is influenced by changes in inter-annual variability in precipitation and by changes in short-term rainfall properties (such as storm rainfall intensity). The frequency of low or high streamflows is also affected mainly by the changes in seasonal distribution patterns and year-to-year variability of precipitation totals, and occurrence of severe prolonged drought events.

The impacts of climate change on hydrology and water resources are usually estimated by defining scenarios for changes in climatic inputs to a hydrological model from the output of general circulation models (GCMs) (Compagnucci et al., 2001; Tatli et al., 2004). Accordingly, hydrological impact assessments, developing and using realistic hydrological models, and understanding better the relations and feedback mechanisms between the climate and hydrological systems are the key developments for constructing scenarios. Understanding of the contemporary hydro-climatological systems and their variability in space and time is also of great importance for climatic model projections, their meaningful interpretation and assessments, and adaptation measures to impacts of the climate change on hydrological systems and water resources.

On the other hand, many downscaling techniques have been developed (e.g. Wilby and Wigley, 1997; Wilby et al., 1998, 1999; Tatli et al., 2004, 2005;, and so on) and widely used in climatological and hydrological studies. These techniques range from simple interpolation of climate model output (e.g. Felzer and Heard, 1999; Şen, 2009), through the use of empirical/statistical relationships between the catchment and regional climate (e.g. Crane and Hewitson, 1998; Wilby et al., 1998, 1999), to the use of nested regional climate models (e.g. Christensen and Christensen, 1998; Compagnucci et al., 2001; Önol and Semazzi, 2009).

Increased awareness of influences of the climatic variability on hydrology and water resources and related socioeconomic sectors and services including forestry, agriculture, energy and drinking water has increased for about 20 years (Kundzewicz et al., 2007). Various studies have investigated the linkages between recognizable oscillation and teleconnection patterns of large-scale climatic variability including particularly the El Niño—Southern Oscillation (ENSO), North Atlantic Oscillation (NAO), Arctic Oscillation (AO), North Sea-Caspian Pattern (NCP) and variations of the hydrological variables in time. These studies, for example McCabe (1996), Piechota et al. (1997), Vogel et al. (1997), Olsen et al. (1999) for North America, Marengo (1995), Marengo et al. (1998) and Compagnucci and Vargas (1998) for South America, Chiew et al. (1998) and Piechota et al. (1998) for Australasia, Shorthouse and Arnell (1997), Rodo et al. (1997), Cullen et al. (2002), Xoplaki et al. (2004), Mariotti et al. (2002), Tatli (2007) and Beranová and Huth (2008) for the Europe and Mediterranean region, Serreze et al. (1997) and Wang et al. (2005) for the Northern Hemisphere, Cullen et al. (2002) for the Middle East, Schulze (1997) for southern Africa and for Turkey, for example Tatli et al. (2004), Türkeş and Erlat (2003, 2005, 2006) revealed types of variability from inter-seasonal and year-to-year variability to decadal fluctuations. It is evidently seen that these observed patterns of variability also vary noticeably from region to region. Such linkage studies are appreciably valuable because they help in understanding atmospheric causes of observed hydrological and hydro-climatological changes and variability not only over time but also in space.

Water management including drought management plans is mainly dealing with the minimization of risk and adaptation to changing circumstances arising from both increased and/or altered demands in the water sector and spatially coherent significant precipitation and groundwater changes over time. Various adaptation techniques has been developed and applied in the water sector for decades (Compagnucci et al., 2001; Adger et al., 2007). One widely used classification differentiates between increasing capacity, changing operating rules for existing structures and systems, managing demand and changing institutional practices. The first two often are termed supply-side strategies, whereas the latter two are demand-side (World Bank, 1993; Kindler, 2000) strategies. Well-organized adaptation to climate change in the water sector requires endeavouring the following five main areas (Compagnucci et al., 2001):

  • Data for monitoring: Adaptive water management requires reliable data on which to make decisions, calibrate models and develop projections for the future.
  • Understanding patterns of variability: An understanding of the patterns of variability is important for medium-term water management, because, even in the absence of climate change, the recent past may not be a reliable guide to the hydrological resource base of the near future.
  • Analytical tools: Effective water management requires various analytical tools to assess options and the future, which should consist of the scenario and risk analyses.
  • Decision tools: Scenario and risk analyses provide information on the characteristics of possible futures and their consequences for policy makers and all related end users.
  • Management techniques: These are the techniques that are actually implemented to meet management objectives.

Nevertheless, such assessments generally do not include the understanding of contemporary hydroclimatological (i.e. existing and natural) characteristics of the forests, agricultural and hydrological systems, energy production, water resources and drought events with respect to the efficient and functional adaptation strategies to climate change. As we have already discussed previously, hydro-climatological aspects of the hydrology, groundwater hydrology and water resources in a region or in a country, particularly in the drought-prone regions faced water-scarcity under the present climatic conditions, such as arid and semi-arid African countries, and the semi-arid and dry sub-humid Mediterranean countries, predominantly characterized with high seasonal and year-to-year precipitation variability should be considered. One of these highlighted hydrological and hydro-climatological aspects is of precipitation regime types of a region or a country. These should clearly show the meaningful spatial distribution patterns of precipitation regime regions arranged by mainly physical geographical, seasonal and/or estimated inter-annual variations of precipitation amounts for a sufficient period of observing time. Therefore, as we have proposed in this paper, clustering of observed precipitation time series, for producing well-defined distribution maps of the precipitation seasonality (or precipitation regime) regions; in terms of the year-to-year variability of precipitation amounts is also of great importance for adapting to the impacts of climate change. This should be devoted particularly to mitigate the negative impacts of global warming that is one of the significant large-scale results of the increased human-induced green-house effect.

5. Conclusions

In this paper, the SC approach is used to determine the coherent precipitation regime regions in Turkey. Clustering techniques have been frequently applied in various disciplines. Generally, hierarchical clustering algorithm is used in climatology (e.g. Lund, 1963; El-Kadi and Smithson, 1992), but in this study we have applied the SC technique for sorting different meteorological stations into groups in a way that the degree of association between their precipitation values is maximal if they belong to the same group (or cluster) and minimal otherwise. The SC is a hybrid method having advantages over other clustering techniques that the example of the two circles given in Section 2 indicates its ability while determining the correct coherent structures.

In its traditional applications, cluster analysis is simply used to discover structures in data without explaining why they exist; however, in this study, we have given some comments in association with the atmospheric and physical geographical controls according to mostly the previous studies (e.g. Türkeş, 1996, 1998; Kutiel et al., 2001; Türkeş et al., 2002; Türkeş and Erlat, 2003, 2005 and 2006; Tatli et al., 2004; Türkeş et al., 2009). From 2- to 8-clusters of the precipitation coherent regions over Turkey were bridged with upper-air circulations and the well-known precipitation climatological aspects being frequently observable, though the cluster-8 is enough to interpret the precipitation clusters according to the VI given in Section 2.

On the other hand, the obtained spatial patterns of the precipitation regions by using the SC algorithm give a new hydro-climatological insight. As a result, not only in Turkey but also in other countries that have similar hydro-climatological conditions and socioeconomic features, policy makers and/or policy-making process of the various socioeconomic sectors, and experts on water management and other fields of expertise in agriculture and forestry, natural resources, decision making in all areas of private and public life related with the weather and climate and hydrology and water resources should make use of this kind of precipitation regime regions by considering the scientific frame of our discussion. For that reason, the main implications of our study might give them a strong facility and a unique chance for understanding and planning their hydrological and hydro-climatological systems and water resources including drinking waters, groundwater, rivers, hydroelectric production, forestry, agricultural and irrigation activities and others. Such hydro-climatological studies, drought and precipitation severity indices could also be taken a significant part in a region's or a country's medium and long-term range water resources and drought management plans along with other social and economic policies (Türkeş and Tatli, 2009, 2010; Türkeş, 2010).

Consequently, for future studies, besides the precipitation values, for example other atmospheric surface fluxes and upper circulation variables, together of three-dimensional atmospheric structure, could also be used to cluster at the same time in order to extract an evident picture of the micro-climate regions. This kind of studies also increases our understanding of climate change and variability, and adaptation capacities for projected future climate changes, particularly the spatial and temporal patterns of precipitation variability varying from small scale to large scale, and from a seasonal and year-to-year to centennial precipitation over time.


We would like to thank the anonymous reviewers who offered valuable suggestions and constructive comments. We are also grateful to the research assistant M. Zeynel ÖZTÜRK in our Department for his help in preparing a generalized map for the precipitation regions of Turkey.