Delineation of homogenous precipitation regions can be rather challenging when considerable spatial variability and seasonality of precipitation exist within a large geographic domain. This study aims to investigate and redefine the spatio-temporal variability of precipitation clusters of Turkey by K-means clustering at different time scales. The monthly precipitation of 234 stations for the 1980–2020 period was included. Five precipitation clusters are identified for the 1980–2020 period, while the clusters varied between four and eight and four and six for the 5- and 10-year sub-periods, respectively. The clustering structures exhibited an indication of certain links to the prominent teleconnection patterns. The monthly precipitation correlated more significantly with Arctic Oscillation (AO) in the Aegean and Mediterranean regions, while North Atlantic Oscillation correlated better with the Eastern and Southeastern Anatolian stations. The variability of the cluster structures is described in association with the coefficient of variation (COV) spatial patterns using the observed station data. At 5- and 10-year subsequent periods, no significant variability in the cluster patterns was observed in the lower COV areas where the monthly precipitation was more uniform throughout the year, while more inter-decadal variability was prominent in the higher COV areas where seasonality in precipitation was more pronounced. Overall, the precipitation regions were not spatially coherent over time, and considerable variability was revealed between different regions of the country.

  • Using the K-means machine learning technique for clustering homogenous precipitation regions and assessing their temporal evolution.

  • Associating changes in the spatial structure of the precipitation regions to climate variability, especially to the prominent teleconnection patterns.

  • Finding a link between the coefficient of variation (COV) and the stability of the precipitation cluster structures.

Graphical Abstract

Graphical Abstract
Graphical Abstract
ACRONYM

DEFINITION

AO

Arctic Oscillation

AWOS

Automated weather observing stations

HBSC

Humid Black Sea Coastal

CA

Cluster analysis

COV

Coefficient of variation

EA

Eastern Anatolian

ENSO

El Niño–Southern Oscillation

MK

Mann–Kendall

MO

Mediterranean Oscillation

NAO

North Atlantic Oscillation

NCP

North Sea Caspian Sea Pattern

NWA

Northwestern Anatolia

PCA

Principal component analysis

SC

Spectral clustering

SCCA

Semi-Humid Continental Central Anatolian

SCM

Semi-arid Continental Mediterranean Zone

SNHT

Standard Normal Homogeneity Test

SOI

Southern Oscillation Index

SQMK

Sequential Mann–Kendall

SS/HCM

Subtropical Semi-Humid/Humid Coastal Mediterranean

TSMS

Turkish State Meteorological Service

Understanding the variable character of climate today is essential and imperative for many economic activities and sectors, especially energy, water resources, and agriculture. Ascertaining how precipitation patterns change spatially and temporally can provide crucial information for planning and management activities in various water-dependent sectors. Moreover, an insight into the statistical characteristics of precipitation variability is crucial not only for water-dependent sectors but also for mitigating risks of hydrological extremes such as floods and droughts. Clustering based on the properties of climatological data facilitates in determining the regions reflecting similar climate characteristics and provides a scientific basis for the planning and management of climate data-reliant sectors. Furthermore, clustering large spatial domains into regions of comparable climatic characteristics can provide vital information for climate change and climate variability studies (Machiwal et al. 2019). Nevertheless, delineating homogeneous precipitation regions is a challenging and complicated task owing to the inherent temporal and spatial variability of precipitation. The identification of homogeneous rainfall regions can become even more troublesome with sparse data and data with reduced temporal resolution (Sankranti et al. 2013; Irwin et al. 2017). In a country like Turkey, where physiographic features, such as elevation and orography, affect climate significantly and different weather systems prevail throughout the year, identifying distinct climate regions becomes even more challenging and complex. Many techniques were developed and implemented over the years with the availability of more reliable historical data and improved computational capabilities, enabling the use of different approaches for clustering climate parameters (Ünal et al. 2003; Netzel & Stepinski 2016). Tuel & Martius (2021) argued that using re-analysis dataset is becoming common for cluster analysis (CA) due to shortcomings of observational data. They further claimed that the re-analysis data also allow for a better assessment of climate change in clustering of the climate parameters. Several statistical methods are available for spatio-temporal characterization of precipitation patterns and identifying their groupings, including L-moments, harmonic analysis, multivariate regression, spatial correlation functions, spatial interpolation, empirical orthogonal functions, and regional frequency analysis (Guttman 1993; Ünal et al. 2012; Sankranti et al. 2013; Wazneh et al. 2015; Zaifoglu et al. 2018). Various spatial pattern recognition techniques such as CA and principal component analysis (PCA) are also other favoured approaches applied by researchers to determine homogenous rainfall regions (Darand et al. 2014; Marston & Ellis 2021). Among those, CA is one of the most preferred data analysis techniques to investigate similarities and dissimilarities in data and categorize the data as groups or clusters (Gore 2000; Jajuga et al. 2002). A cluster is a group of data points aggregated together and is established based on the similarities of the data points using statistical criteria, such as average difference or correlation. The purpose of the clustering procedure is to classify or group a specific set of objects into separate individual groups. Similar objects are targeted to be placed in the same cluster while different objects are positioned in the other clusters based on the degree of similarity (or dissimilarity) between the individual objects or patterns (Rokach & Maimon 2005; Berkhin 2006). Hierarchical and non-hierarchical clustering methods are common approaches employed in clustering studies. K-means is one of the fundamental non-hierarchical clustering techniques which acquired widespread use over the years for the classification of climate data as the method is easy to implement, computationally faster than hierarchical clustering, and suitable for large datasets (Berkhin 2002; Wilks 2011). It became one of the popular unsupervised machine learning algorithms favoured by researchers in recent years with the availability of various computing tools (Rodriguez et al. 2019). Further details of the K-means method are discussed in Section 2.2 of this study.

The spatial distribution of Turkish precipitation is very irregular with high year-to-year variability, which makes it challenging to determine precipitation patterns accurately. Dynamic features of Turkish precipitation are heavily influenced by seasonal shifts of different air masses and changes in the atmospheric circulation systems over Eurasia. The polar air masses (cP) in the winter and tropical (marine) air masses (mP) in the summer are the two main air masses that affect Turkey's annual precipitation and its spatial and temporal distribution seasonally (Sarış et al. 2010). The marine-polar (mP) air masses originating from the Atlantic Ocean and moving towards Southern Europe bring considerable precipitation to the western and southern parts of the country during the winter and early spring months (Tatli et al. 2004). The prevailing westerlies in winter are the dominant circulation patterns that affect the advection of the moist and unstable mP air masses towards Turkey. Seasonal shifts and strengths of various pressure systems have been influential in the occurrence of wet and dry conditions across the country (Türkeş et al. 2008). While the Azores' high is more effective in summer, the Siberian anticyclone becomes more dominant during winter. Both pressure systems usually cause below-normal precipitation over the country during the periods when they are active. The circulation-based effect of the Asian summer monsoon causes very dry conditions in Turkey with the dominance of northerly winds. Some non-meteorological factors including continentality, land−sea contrast, diverse topographical features, and particularly high terrain further complicate the spatial variability of precipitation across the country. Turkey is a predominantly mountainous country with lowlands confined mostly to the deltas, coastal plains, and some interior high plains across the country (Atalay 2016). Nearly one-quarter of the country has an elevation of above 1200 m. Two mountain ranges increasing in elevation towards the east extend in the W−E direction in both the northern and southern parts of the country, causing great contrast in precipitation conditions between the interior and coastal areas. Annual precipitation portrays highly variable behaviour, ranging from 300 mm in internal parts to over 2200 mm in the eastern Black Sea coastal area (Figure 1). The precipitation in the western and southern parts of the country is exposed to more seasonality, while precipitation is uniformly distributed throughout the year in the northern parts.
Figure 1

Spatial distribution of annual average precipitation over Turkey based on the 1991–2020 climatological normal.

Figure 1

Spatial distribution of annual average precipitation over Turkey based on the 1991–2020 climatological normal.

Close modal

The teleconnection patterns also affect precipitation variability over the Mediterranean region, including Turkey. It is shown that the North Atlantic Oscillation (NAO) and Arctic Oscillation (AO) influence the spatial and temporal variability of precipitation over Turkey, causing wetter and drier periods inter-annually (Türkeş & Erlat 2003, 2006; Sezen & Partal 2019). The relationship between the teleconnection patterns and Turkish precipitation is discussed more in detail in Section 5 of this study. Several studies exist in the literature discussing the clustering of Turkish climate parameters. Six precipitation regime clusters were identified by Sarış et al. (2010) by employing the hierarchical, agglomerative CA using Ward's method based on monthly precipitation totals from 107 stations. In another study by Sönmez & Komuscu (2011), six main precipitation clusters were determined as optimum classification using the K-means methodology, and the resulting clustering was discussed in connection to the NAO pattern. They used monthly precipitation data from 148 stations covering the 1977–2006 period. In another study, Ünal et al. (2003) used hierarchical CA to redefine the climate zones of Turkey. They examined monthly precipitation and temperature data from 113 climate stations for the period of 1951–1998 and found that the Ward method yielded more accurate results as compared to the other methods. In the same study, Ünal et al. (2003) identified six different precipitation regions but found considerable differences in the boundaries as in many other similar clustering studies. Türkeş & Tatlı (2011) applied the spectral clustering (SC) technique involving singular value decomposition (SVD) and the K-means method to precipitation totals of 96 stations across Turkey for the 1929–2007 period. Their research concluded a total of eight clusters of coherent precipitation zones. In another study, Iyigun et al. (2013) used the Ward method hierarchical clustering technique for grouping common features of different climate parameters covering the 1970–2010 period at 244 stations in Turkey and determined 14 clusters of climate zones.

Although Turkish precipitation has been classified using a number of different methods over the last three decades, none of these studies covered the period after 2015 when the impact of climate change and variability gained considerable interest and visibility. In addition, most of those studies lacked using a sufficient number of stations and time period length to represent precipitation variability over the country, and more importantly, they did not investigate the dynamic behaviour of the precipitation clusters in both time and space. Most previous studies have identified different results of climate zones in Turkey using a varying number of stations and time periods spanning several decades. The clustering techniques used in those studies were typically only used within time or space domains. These results and the diverse nature of the precipitation distribution over Turkey necessitate a better understanding of the classification groupings for various societal needs and to further recognize the variability as time progresses. This study, therefore, aims to explore the spatial dynamics of precipitation regions across Turkey by clustering precipitation station data at varying time scales (5- and 10-year time scales) to capture the decadal and inter-decadal variability. It is our view that studies only analysing the long-term precipitation data will not reveal the exact nature of dynamics of the precipitation regions over time. In other words, we seek to simultaneously cluster homogenous precipitation regions of Turkey spatially over time, producing sub-clusters for each period selected at 5- and 10-year time scales along with the time period of 1980–2020. Such findings can provide valuable information to assess the implications of climate variability for the country's water resources, energy, and agricultural sectors considering the variability of Turkish precipitation even at a 5-year time scale. The variable behaviour of the precipitation regions at short temporal scales can cause further challenges, particularly for water-dependent sectors such as agriculture and hydropower generation in addition to the irregularities and seasonality of the monthly precipitation of Turkey. Therefore, this study makes significant progress over previous studies by encompassing a wider temporal and spatial coverage with additional stations, including the recent decade, and avoiding the assumptions made with the typical application of the K-means method. Moreover, connections between monthly precipitation totals and prominent teleconnection patterns, including AO, NAO, and El Niño–Southern Oscillation (ENSO), are investigated to explore whether any significant correlation exists between them. To relate and compare the spatial variability of the clusters with the relative variability of the station-observed precipitation data, the coefficient of variation (COV) analysis is also performed for the whole period as well as 5- and 10-year sub-periods. Here, we aimed to demonstrate how the clustering structures relate to the COV patterns in order to describe their spatio-temporal variability at 5- and 10-year time scales. The study also aimed to underline the importance of avoiding assumptions about choosing a single distance measure for the K-means clustering approach and argued that choosing an appropriate distance measure should be decided based on the silhouette analysis.

Data

This study uses monthly precipitation data obtained from 234 automated weather observing stations (AWOS), operated by the Turkish State Meteorological Service (TSMS). Figure 2 illustrates the spatial distribution of the stations included in the analysis. The data cover the 1980–2020 period and were subjected to initial quality control (QC) by TSMS Data Processing Center prior to our analysis. The QC system at TSMS is performed monthly by employing automated QC tests and manual QC checks. The automated QC tests include range, step, persistence, like-instrument, and spatial test. These tests are applied to determine the confidence level of the observation by using ‘good’, ‘suspicious’, or ‘bad’ flag types. Further details of the QC tests performed by the TSMS can be found in Sönmez (2013). In addition to the QC tests performed by the TSMS, we applied two robust statistical methods, namely Alexanderson's standard normal homogeneity (SNHT) test (Alexanderson 1986) and Pettitt's test (Pettitt 1979) to detect inhomogeneity in the precipitation data at the 95% confidence level. The stations were excluded if they failed either test. These tests determine non-climatic influences due to changes in instrumentation, problems with instrumentation, changes in surrounding environmental characteristics, and station relocations. In order to detect a change in the precipitation series due to climate change, a separate analysis was performed using linear regression, the Mann–Kendall (MK) trend test for the direction of the trends, and Sen's slope test for the magnitude of the trends. Moreover, the sequential Mann–Kendall (SQMK) test was used to identify trend change points and the beginning year of the precipitation time series trend. The test results concluded that except for the last decade, the trends identified in the precipitation data were associated with climate variability at the inter-decadal and decadal scales but not due to climate change.
Figure 2

Geographical distributions of the stations used in the K-means analysis.

Figure 2

Geographical distributions of the stations used in the K-means analysis.

Close modal

The monthly time series of the teleconnection indexes used in the analysis were obtained from the National Oceanic and Atmospheric Administration's (NOAA's) Climate Prediction Center (https://www.cpc.ncep.noaa.gov/data) and the University of East Anglia, Climate Research Unit (https://www.uea.ac.uk/web/groups-and-centres/climatic-research-unit/data).

Method

A common requirement of the clustering technique in meteorological applications is the necessity to group measurements into meaningful categories based on observations. In this study, the K-means algorithm is used to partition the monthly precipitation data representing 234 stations across Turkey into a set of spatial clusters. The aim here is to group the monthly precipitation data into clusters based on similarities between the data points and explore underlying spatial patterns. K-means clustering generates a single level of clusters rather than a multilevel hierarchy of clusters. Therefore, it is considered to be more suitable than hierarchical clustering for large amounts of data (Bock 2007). Despite its easy implementation and suitability for datasets with a large number of variables, the K-means method has its own drawbacks. One of the main drawbacks of the algorithm is the uncertainty with the choice of an initial point, which considerably influences the final results (Nisbet et al. 2009; Rodriguez et al. 2019). Different selections of the initial point directly affect the accuracy of aggregation. Another drawback of the method is its sensitivity to outliers.

With the implementation of the K-means method, observations are treated as objects having locations and distances from each other. The objects are assigned to mutually distinctive clusters, and each cluster is designated based on its centroid or centre point. As the distances used in clustering often do not represent spatial distances, a silhouette analysis is performed to decide the partitioning distance between the clusters. A silhouette plot specifies a measure of how close each object is to its assigned cluster in comparison to other clusters, and in this manner, the plot provides a way to assess the appropriate number of clusters and cluster quality visually (Rousseeuw 1987; Kaufman & Rousseeuw 1990). The thickness of the silhouette plot with similar sizes, the maximum silhouette value, and the lowest negative silhouette number are the main criteria to determine an optimum cluster number.

Two necessary input parameters are the number of clusters (k) and their initial centroids, which can be either selected by the user or acquired through pre-processing. The objects for (k) different clusters are configured in such a manner that they are more similar to one another within each cluster while dissimilar to the objects in other clusters (Bock 2007). The cohesion and separation of the data points are decided based on a silhouette value. An estimation of the silhouette values for each of the objects helps to assess the success of assigning an object to a cluster (Kaufman & Rousseeuw 1990). The silhouette value is estimated by the following equation:
formula
(1)
where a(i) is the average similarity between the ith object and the other members of the same cluster and b(i,k) is the average similarity between the ith object and the members of the kth cluster. Silhouette values (S(i)) range between −1 and +1, where +1 indicates that the ith object is retained in the most appropriate cluster and separated from the dissimilar objects in other clusters (Shahapure & Nicholas 2020). A value of zero specifies that the object does not fit specifically to one cluster or another, while −1 states that the object is assigned to the incorrect cluster. The average S(i) indicates how successful the partitioning is with k clusters for the N objects. Overall, less variation within clusters is aimed, while data points within the same cluster are expected to be similar. A detailed discussion about the silhouette analysis can be found in Rousseeuw (1987) and Kaufman & Rousseeuw (1990).
To start the K-means algorithm, a group of randomly selected centroids is assigned, and then the positions of the centroids are optimized by iterative calculations until no change in their values is obtained (Figure 3). This occurs when either the clustering is successful or the pre-defined number of iterations is achieved. A distance measure plays a crucial role in identifying similar data points and forming respective clusters as it defines how the similarity of two elements (x, y) is calculated and will influence the shape of the clusters (Thakare & Bagal 2015). The main purpose of distance measure calculation is to obtain an appropriate similarity (distance) function. The K-means algorithm available with the MATLAB computing tool computes the centroid of clusters for a variety of supported distance measures, including Squared Euclidean, cityblock, cosine, and correlation. The Squared Euclidean distance is the default metric, which is calculated as the square root of the sum of the squared differences between two data points. In this method, the distance between any two objects is not affected by the addition of new objects, such as outliers, to the analysis. The cityblock distance metric refers to the sum of absolute differences among the points after normalizing those points to unit Euclidean length, which refers to the root of square difference between coordinates of pair of objects (Singh et al. 2013). Absolute value distance provides a more robust result, whereas the Euclidean length distance method is influenced by outliers. Lastly, the correlation-based distance considers two objects to be similar if their features are highly correlated, even though the geometrical distance of the observed values may be far apart from each other. This method is computed based on one minus the sample correlation between points, treated as sequences of values. Each of these distance metrics supports some types of cluster and disallows other ones, and thus selecting the proper distance measure is not straightforward and rather depends on different factors of the observed dataset (Everitt et al. 2011). These distance metrics mentioned briefly above are described extensively by Sreevalsan-Nair et al. (2022) (refer to the MATLAB and Statistics Toolbox Release (2012b) for further details on each distance measure). In this study, no initial selection was made to begin the K-means algorithm, and using the default distance metric was avoided. Instead, an optimum distance method was determined based on the optimum cluster number, considering the minimum negative values and the maximum average silhouette value for the respective sub-periods.
Figure 3

Flowchart of the K-means algorithm employed in the study.

Figure 3

Flowchart of the K-means algorithm employed in the study.

Close modal

Based on the proposed methodology above, the precipitation data were processed for the entire 1980–2020 period and 5- and 10-year sub-periods as well to capture the decadal and inter-decadal structure of the precipitation clusters. In the next phase of the study, a correlation analysis was performed in order to establish links between monthly precipitation totals at 234 stations and prominent teleconnection patterns which have a distinctive influence on Turkish precipitation. To quantify the relationship between the teleconnection indices and precipitation totals, we calculated linear correlations between the monthly teleconnection indices for the three teleconnection patterns, namely the AO, NAO, and ENSO, and the precipitation time series at each station. The resulting connections were analysed to gain some insights into the strength of the linear relationship between the station precipitation and the teleconnection indices. The spatial distribution of the stations that correlated significantly with the single and/or multiple teleconnection indices and the percentage of the stations in each region that correlated with each teleconnection pattern is also examined.

Lastly, in order to relate the spatial variations observed with the K-means clusters to the variability of the station-observed precipitation data at respective sub-periods, the COV analysis was carried out. The resulting COV patterns were visually compared to the spatial shifts in the clusters for the 1980–2020 period and 5- and 10-year sub-periods. The aim here was to identify if any coherency exists between the COV patterns and the clustering configurations.

Silhouette analysis based on different distance metrics

Analysis for the 1980–2020 period

Silhouette analysis was performed for the 1980–2020 period encompassing 41 years of data, indicating that the cosine distance method yielded the optimum result for determining the cluster members (Table 1). The analysis shows that when the n_clusters are equal to 3 and 4, all the plots portray similar thickness, and hence, similar sizes as also confirmed from the labelled scatter plot (Figure 4). Other silhouettes showed more ambiguous and misleading results in deciding the proper number of clusters. With the cosine distance method, the average silhouette score and the number of negative values are 0.287 and 3, respectively, for n-clusters which equal 5. In other words, the optimum cluster number is defined based on the minimum negative values and the maximum average silhouette value.
Table 1

The silhouette analysis for the 1980–2020 period with the resulting optimum method and number of clusters (highlighted in bold and italics)

Cluster periodDistance methodSilhoutte informationCluster members
345678
1980–2020 Squared Euclidean Average silhouette value 0.412 0.387 0.398 0.402 0.392 0.316 
Number of negative value 56 44 32 25 33 30 
Cosine Average silhouette value 0.260 0.277 0.287 0.282 0.268 0.255 
Number of negative value 3 
Cityblock Average silhouette value 0.243 0.233 0.249 0.135 0.184 0.160 
Number of negative value 36 36 29 39 32 24 
Correlation Average silhouette value 0.228 0.266 0.280 0.253 0.249 0.247 
Number of negative value 17 16 14 13 
Cluster periodDistance methodSilhoutte informationCluster members
345678
1980–2020 Squared Euclidean Average silhouette value 0.412 0.387 0.398 0.402 0.392 0.316 
Number of negative value 56 44 32 25 33 30 
Cosine Average silhouette value 0.260 0.277 0.287 0.282 0.268 0.255 
Number of negative value 3 
Cityblock Average silhouette value 0.243 0.233 0.249 0.135 0.184 0.160 
Number of negative value 36 36 29 39 32 24 
Correlation Average silhouette value 0.228 0.266 0.280 0.253 0.249 0.247 
Number of negative value 17 16 14 13 
Figure 4

Silhouette plots based on cosine distance metric (left) and the visualization of clustered data for the period of 1980–2020 with n-clusters =5 (right).

Figure 4

Silhouette plots based on cosine distance metric (left) and the visualization of clustered data for the period of 1980–2020 with n-clusters =5 (right).

Close modal

Considering the thickness of the silhouette plot, the maximum silhouette value, and the lowest negative silhouette number, the silhouette analysis yielded a total of five clusters as the optimum number of precipitation clusters for the 1980–2020 period. In Figure 4, the thickness of the silhouette plot (cluster size) is illustrated. Here, n_clusters equal to 5 yield the best average silhouette score of 0.287 along with all the clusters above the average, indicating that the resulting five clusters are reasonable. In Figure 3, the right panel indicates whether the clusters are located in the lower variance region or the higher variance region. More importantly, it indicates if the boundary among the clusters is well-defined or not. This description applies to all the silhouette plots given in the analysis.

Analysis for 10-year sub-periods

Similarly, a silhouette analysis was performed for the four 10-year sub-periods in order to determine the number of K-means clustering for each sub-period. In doing so, all the silhouettes were visually examined based on different distance metrics as previously described in the analysis of the full-time period (1980–2020). Our analysis indicates that the cosine distance method gave the best result for the three sub-periods, including 1980–1989, 1990–1999, and 2010–2020. Only for the 2000–2009 period does the correlation distance method give the optimum number of clusters (Table 2).

Table 2

The silhouette analysis for the four 10-year sub-periods with the resulting optimum distance method (highlighted in bold and italics)

10-year sub-periodsDistance methodSilhoutte informationCluster members
345678
1980–1989 Cosine Average silhouette value 0.279 0.310 0.308 0.274 0.261 0.254 
Number of negative value 13 3 
1990–1999 Cosine Average silhouette value 0.281 0.284 0.286 0.294 0.277 0.274 
Number of negative value 14 4 11 
2000–2009 Correlation Average silhouette value 0.298 0.265 0.306 0.293 0.242 0.248 
Number of negative value 21 3 
2010–2020 Cosine Average silhouette value 0.262 0.264 0.276 0.279 0.279 0.243 
Number of negative value 18 10 22 17 7 18 
10-year sub-periodsDistance methodSilhoutte informationCluster members
345678
1980–1989 Cosine Average silhouette value 0.279 0.310 0.308 0.274 0.261 0.254 
Number of negative value 13 3 
1990–1999 Cosine Average silhouette value 0.281 0.284 0.286 0.294 0.277 0.274 
Number of negative value 14 4 11 
2000–2009 Correlation Average silhouette value 0.298 0.265 0.306 0.293 0.242 0.248 
Number of negative value 21 3 
2010–2020 Cosine Average silhouette value 0.262 0.264 0.276 0.279 0.279 0.243 
Number of negative value 18 10 22 17 7 18 

K-means clustering of the sub-period monthly total precipitation data is shown in the following figures that display silhouette plots for the clusters based on different distance metrics and the visualization of clustered data (Figure 5).
Figure 5

Silhouette plots based on cosine distance metric (left) and the visualization of clustered data for the periods of (a)1980–1989 with n-clusters = 4; (b)1990–1999 with n-clusters = 6; (c) 2000–2009 with n-clusters = 6; and (d) 2010–2020 with n-clusters = 7.

Figure 5

Silhouette plots based on cosine distance metric (left) and the visualization of clustered data for the periods of (a)1980–1989 with n-clusters = 4; (b)1990–1999 with n-clusters = 6; (c) 2000–2009 with n-clusters = 6; and (d) 2010–2020 with n-clusters = 7.

Close modal

Analysis for 5-year sub-periods

The silhouette analysis was again performed to obtain an accurate number of K-means clustering for the eight 5-year sub-periods in a similar way used for the 10-year sub-periods and the full-time period. Our analysis indicates that the cosine distance method gave the best result for the six 5-year sub-periods, including 1980–1984, 1985–1989,1990–1994, 1995–1999, 2000–2004, and 2015–2020 (Table 3). For the 2005–2009 and 2010–2014 sub-periods, however, the correlation distance method identified the cluster numbers more accurately.

Table 3

The silhouette analysis for the eight 5-year sub-periods with the resulting optimum distance method (highlighted in bold and italics)

5-year sub-periodsDistance methodSilhoutte informationCluster members
345678
1980–1984 Cosine Average silhouette value 0.342 0.298 0.295 0.271 0.273 0.255 
Number of negative value 5 10 14 13 
1985–1989 Cosine Average silhouette value 0.290 0.288 0.312 0.266 0.246 0.252 
Number of negative value 11 1 
1990–1994 Cosine Average silhouette value 0.276 0.295 0.315 0.325 0.299 0.307 
Number of negative value 10 11 10 11 5 
1995–1999 Cosine Average silhouette value 0.285 0.274 0.284 0.266 0.258 0.269 
Number of negative value 11 11 7 10 
2000–2004 Correlation Average silhouette value 0.228 0.281 0.321 0.285 0.312 0.263 
Number of negative value 26 23 12 15 8 
2005–2009 Correlation Average silhouette value 0.214 0.258 0.282 0.274 0.242 0.254 
Number of negative value 23 14 8 14 
2010–2014 Cosine Average silhouette value 0.247 0.255 0.265 0.268 0.258 0.241 
Number of negative value 15 10 8 11 13 
2015–2020 Cosine Average silhouette value 0.290 0.300 0.287 0.307 0.310 0.265 
Number of negative value 12 7 12 10 11 19 
5-year sub-periodsDistance methodSilhoutte informationCluster members
345678
1980–1984 Cosine Average silhouette value 0.342 0.298 0.295 0.271 0.273 0.255 
Number of negative value 5 10 14 13 
1985–1989 Cosine Average silhouette value 0.290 0.288 0.312 0.266 0.246 0.252 
Number of negative value 11 1 
1990–1994 Cosine Average silhouette value 0.276 0.295 0.315 0.325 0.299 0.307 
Number of negative value 10 11 10 11 5 
1995–1999 Cosine Average silhouette value 0.285 0.274 0.284 0.266 0.258 0.269 
Number of negative value 11 11 7 10 
2000–2004 Correlation Average silhouette value 0.228 0.281 0.321 0.285 0.312 0.263 
Number of negative value 26 23 12 15 8 
2005–2009 Correlation Average silhouette value 0.214 0.258 0.282 0.274 0.242 0.254 
Number of negative value 23 14 8 14 
2010–2014 Cosine Average silhouette value 0.247 0.255 0.265 0.268 0.258 0.241 
Number of negative value 15 10 8 11 13 
2015–2020 Cosine Average silhouette value 0.290 0.300 0.287 0.307 0.310 0.265 
Number of negative value 12 7 12 10 11 19 

K-means clustering of the monthly total precipitation data selected from Table 4 is shown in the following figures that display silhouette plots for the clusters based on different distance metrics and also the visualization of clustered data (Figure 6).
Table 4

Number of clusters identified for each 10-year sub-period

No.PeriodNumber of clustersDistance method
1980–1989 Cosine distance 
1990–1999 Cosine distance 
2000–2009 Correlation distance 
2010–2020 Cosine distance 
No.PeriodNumber of clustersDistance method
1980–1989 Cosine distance 
1990–1999 Cosine distance 
2000–2009 Correlation distance 
2010–2020 Cosine distance 
Figure 6

Silhouette plots based on cosine and correlation distance metrics (left) and visualization of clustered data for the period of (a) 1980–1984 with n-clusters = 6; (b)1985 = 1989 with n-clusters = 5; (c)1990 = 1994 with n-clusters = 4; (d)1995 = 1999 with n-clusters = 5; (e) 2000–2004 with n-clusters = 5; (f) 2005–2009 with n-clusters = 5; (g) 2010–2014 with n-clusters = 5; and (h) 2015–2020 with n-clusters = 4.

Figure 6

Silhouette plots based on cosine and correlation distance metrics (left) and visualization of clustered data for the period of (a) 1980–1984 with n-clusters = 6; (b)1985 = 1989 with n-clusters = 5; (c)1990 = 1994 with n-clusters = 4; (d)1995 = 1999 with n-clusters = 5; (e) 2000–2004 with n-clusters = 5; (f) 2005–2009 with n-clusters = 5; (g) 2010–2014 with n-clusters = 5; and (h) 2015–2020 with n-clusters = 4.

Close modal

When we refer to the silhouette plots of the clusters at the respective sub-periods, it is concluded that the ‘cosine’ distance measure provided a better interpretation of the clustered data for the most respective sub-periods, except for a very few cases when the ‘correlation’ distance method was the resulting distant metric.

K-Means clustering for the 1980–2020 period

For the 1980–2020 period, the K-means clustering analysis indicated five distinct precipitation clusters based on silhouette values and different distance methods (Figure 7). The first cluster (Cluster label 1 in Figure 7) includes the entire Southeastern Anatolian region and southern parts of the Eastern Anatolian (EA) area, extending towards the Mediterranean region in the west. This cluster can be called the semi-arid continental Mediterranean (SCM) cluster, which represents a drier form of the typical Mediterranean precipitation region with higher summer temperatures. The second cluster (Cluster label 2 in Figure 7) encompasses the eastern Anatolian parts of Turkey and can be called the EA precipitation cluster. That is a unique cluster to Eastern Anatolia due to its high altitude and being far from ocean influences. Continental climate dominates this cluster with long winters and short summers. The third cluster (Cluster label 3 in Figure 7) includes the entire Aegean region, western parts of the Marmara region, and the Mediterranean Sea coast extending to the east of Antalya Bay. This cluster can be called the subtropical semi-humid/humid coastal Mediterranean (SS/HCM) cluster, with some regional variations in precipitation. It is characterized by semi-humid conditions along the Aegean coasts and greater humidity over the Mediterranean coasts with dry summers. The wetter Mediterranean precipitation patterns penetrate to the east towards the inner parts of Turkey and the Marmara region, which are normally considered the transitional area between the Black Sea and Mediterranean climatic regions. The fourth cluster (Cluster label 4 in Figure 7) lies over a narrow coastal zone of the Black Sea region with a distinct geographic pattern. It also extends slightly towards the eastern Marmara region with southward penetration. This cluster is called the Humid Black Sea Coastal (HBSC) cluster and reflects the physiography of the region where the mountains lie from west to east near the coast. The Marmara region can be considered a sub-regional cluster in the fourth cluster with a more continental character. It has a transitional climate between the Mediterranean climate and the HBSC cluster. Precipitation distribution over the HBSC is usually more uniform throughout the year, although year-to-year variability can occur in eastern parts of the region where the highest amount of annual precipitation is recorded in Turkey. Finally, the fifth precipitation cluster (Cluster label 5 in Figure 7) covers central Anatolian parts. The spatial domain of this cluster closely resembles the Central Anatolian geographic region as displayed in Figure 8. This cluster also penetrates the Black Sea region, especially in the western parts, and can be called a semi-humid continental central Anatolian (SCCA) precipitation cluster. It represents the semi-arid environment of Turkey with the least amount of precipitation observed but gains a semi-humid character towards Eastern Anatolia. The semi-arid precipitation cluster extends towards both the east and the north, making up the largest precipitation cluster. This cluster highlights the amplification of the semi-arid character of the Turkish climate towards the northern and eastern parts of the country, which is known to be wetter than the central Anatolian areas.
Figure 7

Main precipitation clusters delineated for the 1980–2020 period.

Figure 7

Main precipitation clusters delineated for the 1980–2020 period.

Close modal
Figure 8

Geographical regions of Turkey (adapted from www.mapsofworld.com).

Figure 8

Geographical regions of Turkey (adapted from www.mapsofworld.com).

Close modal

While the clustering structure exhibited some influence from the physiographic features of Turkey, the structure of the precipitation clusters does not exactly fit those features and exhibits some significant shifts (Figure 7). These physiographic regions (also called typical geographical regions of Turkey) are displayed in Figure 8. The regions are referred to in the study describing shifts in the precipitation clusters.

K-means clustering for the 10-year sub-periods (decadal scale)

When the 10-year subsequent periods are taken into account, four clusters are obtained for the 1980–1989 period, six clusters for the 1990–1999 period, six clusters for the 2000–2009 period, and seven clusters for the 2010–2020 period, based on the silhouette values and the cosine distance method (Table 4). Overall, the 10-year cluster structures differ from the 41-year cluster pattern described in the previous section considering the entire 1981–2020 period. Both the number of clusters and their spatial coverage change slightly to moderately in the sub-periods. It should be noted that the last sub-period (2010–2020) consists of 11 years just to update the period until 2020 instead of the typical 10-year period

The clustering structure of the first 10-year period (1980–1989) reveals four separate clusters. Figure 9 shows that Marmara, Aegean, and Mediterranean regions formed one unique cluster (Cluster label 4). It is similar to the clustering structure observed for the entire period (1981–2020) with a slight difference in the clustering points in the Marmara and Mediterranean regions exhibiting a larger spatial extent in the 1980–1989 period as compared to the entire period. Another striking feature observed is that the northeastern Anatolia and Eastern Black Sea zone formed one unique single cluster (Cluster label 3). The coastal Black Sea cluster did not retain its spatial pattern as was observed in the full-time period (1980–2020) clustering structure and became part of the SCCA precipitation cluster in the northern and western parts of the Black Sea region. During the 1990–1999 sub-period, the cluster structures changed slightly with the most significant change being the reappearance of the coastal Black Sea cluster (Cluster label 6). Similarly, a transition cluster between the coastal Black Sea cluster and the Southeastern Anatolian cluster developed, which is characterized as the EA cluster (Cluster label 2). During the third 10-year period covering the years 2000–2009, the SCCA precipitation cluster shrank considerably, allowing the formation of a small cluster towards eastern Anatolia (Cluster label 3). The EA precipitation cluster also dwindled in its spatial extent considerably (Cluster label 5). The last period in 2010–2020 displays a completely different precipitation clustering structure as compared to the previous three structures. First, there is more diversity in the cluster structures with a total of seven clusters. A noticeable difference is the presence of two separate clusters, namely the Marmara region cluster (Cluster label 2) and the Eastern Mediterranean cluster (Cluster label 5). Cluster 2, which developed during this period, closely resembles the Marmara region geographic domain as illustrated in Figure 8. The Eastern Mediterranean region is characterized as a separate cluster, while the Aegean and Marmara regions form two different clusters. The coastal Black Sea reappeared as a separate cluster (Cluster label 3), while the EA cluster (Cluster label 7) expanded southward and covered a larger spatial domain as compared to the two previous 10-year sub-periods. A total of seven precipitation regions formed during this period, showing a highly fragmented structure. The climate conditions that prevailed during the last 10-year sub-period contained more clusters with definite boundaries. Considering the warmest years in the Turkish temperature records are observed during the last 10-year sub-period, it is inevitable to argue that climate change is impacting the precipitation patterns.
Figure 9

Precipitation clusters for 10-year subsequent periods between 1980 and 2020.

Figure 9

Precipitation clusters for 10-year subsequent periods between 1980 and 2020.

Close modal

K-means clustering for the 5-year sub-periods (inter-decadal scale)

The overall number of the clusters for the 5-year sub-periods varies between three and eight with the cosine distance method resulting as the optimum for five sub-periods and the correlation distance method optimum for three sub-periods (Table 5).

Table 5

Number of clusters identified for each 5-year sub-period

No.PeriodNumber of clustersDistance method
1980–1984 Cosine distance 
1985–1989 Correlation distance 
1990–1994 Cosine distance 
1995–1999 Cosine distance 
2000–2004 Cosine distance 
2005–2009 Cosine distance 
2010–2014 Correlation distance 
2015–2020 Correlation distance 
No.PeriodNumber of clustersDistance method
1980–1984 Cosine distance 
1985–1989 Correlation distance 
1990–1994 Cosine distance 
1995–1999 Cosine distance 
2000–2004 Cosine distance 
2005–2009 Cosine distance 
2010–2014 Correlation distance 
2015–2020 Correlation distance 

The clusters identified in the 5-year sub-periods portray more variable patterns than those in the 10-year sub-periods (Figure 10). In the first 5-year period of 1980–1984, the coastal Black Sea appears as a unique cluster (Cluster label 1), while Central Anatolia and Eastern Anatolia emerge as one single cluster (Cluster label 2). Marmara, Aegean, Mediterranean, and Southeastern Anatolian regions make up a single cluster (Cluster label 3) representing different sub-types of the Mediterranean climate. The next 5-year period from 1985 to 1989 portrays interesting clustering features. The Southeastern Anatolia breaks off the main Mediterranean cluster as observed in the previous sub-period and emerges as one single cluster (Cluster label 1) in this period. Similarly, a small cluster (Cluster label 5) develops over Eastern Anatolia and breaks off the Central Anatolian cluster (Cluster label 2 in the 1980–1984 sub-period) formed in the previous 5-year sub-period. The 1990–1994 period is characterized by a highly fragmented pattern with eight clusters. The clustering structure in this period closely resembles the physiographic regions of Turkey, with an exception of the formation of a cluster (Cluster label 8) between the Aegean and Central Anatolian regions that developed at the expense of the Central Anatolian cluster. A total of five clusters are identified during the 1995–1999 sub-period. The coastal Black Sea cluster expands further west, occupying the eastern and northern parts of the Marmara region (Cluster label 3). The Central Anatolian cluster enlarges towards the east, covering a greater spatial extent than its original size while causing the EA cluster to dwindle (Cluster label 1). Interestingly, the northwestern parts of Central Anatolia and northern parts of Eastern Anatolia presented the same clustering pattern (Cluster label 2), although they were located far from each other. Lastly, during this sub-period, the Southeastern Anatolian cluster (Cluster label 5) expands westward into the Mediterranean cluster, forming one large cluster as a true representative of the Mediterranean climate. The 2000–2004 period resembles the overall climatic pattern identified for the 1980–2020 period with well-defined boundaries, except for the Mediterranean climate zone (Cluster label 5) extending further north into the Marmara region.
Figure 10

Precipitation clusters for 5-year subsequent periods between 1980 and 2020.

Figure 10

Precipitation clusters for 5-year subsequent periods between 1980 and 2020.

Close modal

The most noticeable change in the cluster structure during the 2005–2009 period is an enlargement of the Southeastern Anatolian cluster (Cluster label 3) extending further north and west, causing a reduction in the EA cluster size. During this period, Marmara, Aegean, and Mediterranean regions formed one single cluster (Cluster label 5). The coastal Black Sea appears as a unique cluster (Cluster label 4) and the central Anatolian cluster (Cluster label 2) nearly resembles the physical geographic boundaries of the region, except for a northward enlargement observed towards the western Black Sea region.

The 2010–2014 period is characterized by five clusters, and two of them indicate spatial enlargement beyond their typical geographic boundaries. One of them is the coastal Black Sea cluster (Cluster label 5) which extends further west, occupying large parts of the Marmara region. Similarly, the Southeastern Anatolian cluster (Cluster label 1) tends to expand further west occupying the coastal Mediterranean area. The last period encompassing the 2015–2020 sub-period portrays striking clustering features. This last sub-period consists of six years just to update the period until 2020 instead of the typical 5-year period. We should keep in mind that Turkey observed its highest annual mean temperature anomalies and decreased rainfall trends during this last 6-year period. The semi-arid central Anatolian cluster (Cluster label 2) expanded its boundaries southward, reflecting the warming temperatures and below-normal precipitation of the last decade. The moist north Black Sea cluster (Cluster label 3) stretched further westward, causing the retreat of the less moist Aegean climate zone (Cluster label 4) back to its typical geographic boundaries. Moreover, the continental form of the Mediterranean climate covering Southeastern Anatolia also occupied the coastal Mediterranean zone as was the case in the previous 5-year sub-period (Cluster label 1 in both sub-periods). Overall, the clustering structure during the last 6-year period changed remarkably with the impact of the warming and less precipitation than the normal observed.

Teleconnections are recurring and persistent disturbances in the atmosphere that cause weather and climate anomalies over many areas of the globe on both inter-annual and decadal timescales (Nigam & Baxter 2015; Feldstein & Franzke 2017). The AO, the NAO, and the ENSO are three main teleconnection patterns that have proven to influence the larger Mediterranean region, including Turkey. The Mediterranean Oscillation (MO) and the North Sea Caspian Sea Pattern (NCP) are also recognized to influence the Turkish climate, but they were not taken into consideration in this study as their data series were of insufficient length to be correlated with the precipitation data.

The teleconnection patterns affect temperature and precipitation over the Mediterranean region, especially by regulating the moisture transport from the oceans into the Mediterranean area (Mariotti et al. 2002; Lionello et al. 2006). Changes in storm tracks associated with teleconnection patterns can result in wet and dry conditions across Eurasia. Several studies argued that NAO profoundly influences the Mediterranean basin precipitation patterns (Adjez 2000; Jacobeit et al. 2001; Dünkeloh & Jacobeit 2003). During a positive phase of the NAO, Turkey is influenced by a cooler and drier climate (Cullen & deMenocal 2000). Another study also highlighted that drier than normal conditions prevail during a positive phase of NAO over parts of southern Europe, including the northern Mediterranean countries (Marshall et al. 2001). The link between the atmospheric teleconnections and the climate of Turkey has also been thoroughly investigated by several studies (Kahya & Karabörk 2001; Türkeş & Erlat 2003; Karabörk et al. 2005; Karabörk & Kahya 2009). Türkeş & Erlat (2009) questioned the link between the NAO and winter temperatures and found significant negative correlations. The AO is another teleconnection pattern that impacts the mid-latitude climates. A strongly negative phase of the AO is associated with warm weather in high latitudes while bringing cold and stormy weather to the more temperate regions (Thompson & Wallace 2001). During the positive AO phase, a surge of arctic cold air and enhancement of the polar front jet (PFJ) bring colder temperatures to Eurasia and North America, while northern Europe gets wetter and the Mediterranean area drier (Hurrell 1995; Thompson & Wallace 2000). A negative AO is distinguished by the southward displacement of arctic cold air and the southward shift of PFJ, causing low temperature and high precipitation in mid-latitude regions. The displacements of both are considered to be associated with the weakening of the polar vortex over the Arctic region (Kömüşcü & Oğuz 2021). The Mediterranean region receives more moisture during the negative AO phase with storms originating in the central Atlantic Ocean. Some studies identified links between AO and Turkish precipitation (Gong et al. 2001; Türkeş & Erlat 2008; Choi & Byun 2010). In a more recent study, Sezen & Partal (2019) demonstrated the presence of a significant correlation between AO and winter precipitation in western parts of Turkey and claimed that the Atlantic depressions advect more moisture to the Mediterranean region during the negative AO phase. Of the third teleconnection pattern examined, ENSO is believed to have some impact on precipitation anomalies over the Euro-Mediterranean region. The oscillating warming and cooling patterns of ENSO influence not only the rainfall distribution in the tropics but also the weather patterns across many parts of the world (Halpert & Ropelewski 1992; Ropelewski & Halpert 1996; Davey et al. 2014; Lin & Qian 2019). Several studies highlighted the ENSO-associated significant changes in the frequency and intensity of rainstorms and rainfall anomalies in the Euro-Mediterranean region (McPhaden et al. 2006; Brönnimann 2007; López-Parages & Rodriguez-Fonseca 2012). Kadıoğlu et al. (1999) studied the potential effects of the ENSO warm patterns on the Turkish monthly precipitation totals and argued that the high ENSO index has distinct impacts on the seasonal variability of Turkish precipitation. Kiladis & Diaz (1989) related the wetter or drier than normal conditions observed in the eastern and northeastern parts of Turkey to the different phases of El Niño. They noted that wetter than normal conditions over northeastern Turkey coincided with the El Niño warm phase, while drier than normal conditions occurred over eastern Turkey the year following an El Niño year. Lastly, Kahya & Karabörk (2001) found coherent and significant streamflow responses to the El Niño and La Niña events in Northwestern Anatolia (NWA) and EA.

Although the terrain features, topography, and latitude have a direct influence in determining boundaries between the identified clusters and the existence of spatial differences between the clusters, other factors play a major role to explain the spatial variability of the clusters at decadal and inter-decadal temporal scales. Our correlation analyses between the monthly precipitation time series (full period of 1980–2020) of the 234 stations and the prominent teleconnection patterns that affect the Turkish climate indicate that the monthly precipitation totals correlate more significantly with AO in the Marmara, Aegean, Mediterranean, and Central Anatolian stations, while the NAO correlates better with the Eastern and Southeastern Anatolian stations (Figure 11). In other words, monthly precipitation in coastal regions of Turkey, except for the Black Sea region, correlates more significantly with AO. Southeastern Anatolian stations exhibited mixed-signal correlations with AO and NAO. Furthermore, for the stations which exhibited a significant correlation with AO, the direction of the relationship was negative. However, the direction of the relationship was positive for the stations correlated with NAO. Only a few stations in the Black Sea and Southeastern Anatolia regions exhibited a notable correlation with ENSO.
Figure 11

Geographical distribution of stations that correlated significantly with the single and/or multiple teleconnection indices.

Figure 11

Geographical distribution of stations that correlated significantly with the single and/or multiple teleconnection indices.

Close modal
Another interesting feature observed with the correlation analysis is the percentage of the stations in each region correlated with each teleconnection pattern (Figure 12). In Southeastern Anatolia, all the stations were correlated with NAO, while the percentage of correlated stations decreased by less than 50% in the other four regions which exhibited some connection to the NAO. Among the regions correlated with AO, the Marmara region exhibited the highest percentage, namely 87%, followed by the Mediterranean region with 68%.
Figure 12

Percentage of stations correlated with each teleconnection pattern.

Figure 12

Percentage of stations correlated with each teleconnection pattern.

Close modal

Overall, we argue that in addition to the physiographic features, climate variability resulting from teleconnection patterns, such as AO and NAO, influences the precipitation patterns over Turkey. The change in the spatial patterns of the precipitation regions can be associated with different regional precipitation mechanisms that are controlled by the varying atmospheric circulation patterns, including teleconnection patterns operating at an inter-annual scale. Some previous studies indicate that NAO usually has ∼6–8 years of variability with the highest power for a periodicity of 7.7 years (Zhang et al. 2011; Deser et al. 2017; Rust et al. 2022). In an earlier study, Jevrejeva et al. (2003) showed that NAO and AO presented high coherency in the 2–3.5, 5.2–7.8, and 12–20 year bands. Zhang et al. (2011) concluded that the spectrum of the NAO presented different frequency domains seasonally, did not reveal the classical peaks at about 2.5 and 6–10 years in winter, and represented the enhanced power at periods 6–10 years. The same study also concluded that ENSO (Niño 3.4 index) represents statistically significant power for a band of ∼2–7 years, indicating quasi-periodic behaviour. Moreover, Pinault (2020) found that the Southern Oscillation Index (SOI) signal has an average period of about four years and a broad peak prolonging beyond 6.5 years based on the Fourier spectrum analysis. The variability of the teleconnection patterns mentioned here is thought to influence change in the spatial patterns of the precipitation clusters, which needs to be further detailed.

In order to relate the spatial variability of the precipitation clusters to the dispersion of the observed station data, the COV is calculated. Then, the spatial distribution of the calculated COVs for the stations is mapped and compared to spatial shifts in the clusters for each period visually. The COV is expressed as the ratio of the standard deviation to the mean and indicates the extent of variability in relation to the mean (Brown 1998). As the COV gets higher, the level of dispersion around the mean becomes greater. The COV is simply calculated as:
formula
(2)
where σ and μ represent standard deviation and mean, respectively.

The COV analysis aimed to gain insights into how the identified cluster patterns relate to the dispersion in the observed station data in order to describe their spatio-temporal variability at varying time scales. We especially looked at shifts in the cluster patterns at subsequent time periods in relation to the variability of the COV patterns. It is our view that identifying such a relationship between the two will provide some understanding of changes in the precipitation regime at decadal and inter-decadal scales.

Relationship between the COV patterns and cluster structures at respective periods/sub-periods

Figure 13 displays the spatial variation of the COV for the time period of 1980–2020. As illustrated in Figure 13, the COV values increase from north to south, with longitudinal bands of similar COV values extending west to east, and narrowing towards the south as the precipitation becomes more variable and the highest COV values occur. The COV pattern exhibited in Figure 13 clearly illustrates the precipitation distribution of Turkey with more variability in the south and a steadier pattern in the northern parts of the country. Figure 14 shows the uniform and non-uniform behaviour of the monthly precipitation distribution for the different geographic regions. The coastal Black Sea region has the lowest COV values ranging from 0.56 to 0.73 receiving a more uniform distribution of monthly precipitation throughout the year, corresponding with Cluster 4 in Figure 7. This region also receives the highest amount of precipitation in the country with more than 2000 mm annually. The main factors controlling the precipitation over the area are the northerly surface and upper air flow advecting Atlantic humid, cold air masses (maritime polar), and the added orographic influence. The Aegean and Mediterranean coastal areas are characterized by high COV values (greater than 1.08) with a distinct precipitation decrease from the winter months to the summer. These two regions clearly define Cluster 3 identified when the whole study period is taken into account (Figure 7). Both Aegean and Mediterranean regions receive a large portion of their precipitation in the winter months in association with the eastward propagating mid-latitude cyclones and Mediterranean depressions (Türkeş 1996). The Southeastern Anatolia region is also depicted by higher COV values with a very distinctive Mediterranean climate of wet winters and a very dry summer season, corresponding to Cluster 1 in Figure 7. The COV values representing the Central and EA regions vary between 0.73 and 0.91, which are lower than the southern and western coastal parts of Turkey but slightly higher than the coastal Black Sea region. These areas correspond to Clusters 5 and 2, respectively, covering the largest cluster combination in the country with latitudinal banding and increasing southward. Precipitation features in those regions reflect continentality, land–sea contrast, and proximity to sea effects. The COV values tend to be lower where precipitation is relatively uniform across the year, while they tend to be higher in regions where a marked seasonal regime occurs and peak winter precipitation is dominant.
Figure 13

Spatial variation of the COV for the 1980–2020 time period.

Figure 13

Spatial variation of the COV for the 1980–2020 time period.

Close modal
Figure 14

Annual mean precipitation cycles for Turkey and its geographic regions.

Figure 14

Annual mean precipitation cycles for Turkey and its geographic regions.

Close modal
The spatial variability of the COV for each 10-year sub-period portrays the east–west banding apparent in the general pattern described in the previous paragraph, although differences are apparent during the sub-periods (Figure 15). The most noticeable change occurs in Southeastern Anatolia during the last 10-year sub-period (2010–2020) with much greater COV variability, nearly ranging from 1.0 to 1.58. The Southeastern Anatolian region exhibited less variability during the first 10-year sub-period (1980–1989), but in the subsequent decades, the COV values increased considerably showing a similar COV pattern with that observed in the Mediterranean coastal zone. That is also reflected in the clustering structures of the region, which were highly variable. The enhancement of the COV band corresponded well with the variabilities of Clusters 5 and 6 observed in the 2000–2009 and 2010–2020 sub-periods, respectively, as illustrated in Figure 10. Similarly, Central Anatolia is remarked with an enhanced variability of COV values that enlarge spatially forming a band extending eastward in the last two 10-year sub-periods, corresponding well with the Semi-humid Continental Central Anatolia (SCCA) precipitation cluster. The moist humid Eastern Black Sea zone is characterized by the lowest COV values for the first three sub-periods and then suddenly shrinks in the last (2010–2020) sub-period with the expansion of the SCCA northward. Cluster 7 in Figure 10 represents the expansion and the northward shift of the SCCA cluster.
Figure 15

Spatial variation of the COV for the 10-year sub-periods.

Figure 15

Spatial variation of the COV for the 10-year sub-periods.

Close modal
A common feature as found in the full period of the study and the 10-year sub-periods is also observed in all the 5-year sub-periods with an increase in COV values from north to south (Figure 16). At 5-year sub-periods, almost all the precipitation regions indicate the expansion or contraction of the COV bands. For example, the SCCA cluster portrays northerly expansion and occupies the Black Sea coastal region during 1985–1989, 1990–1994, and 2010–2014 sub-periods as compared to the spatial coverage of the clusters as given in Figure 7. The expansion and contraction exhibited in the COV bands match the spatial shifts of the clusters identified for most of the 5-year sub-periods (Figure 10). For instance, the clustering patterns observed during the 2005–2009 sub-period (Figure 10) closely resemble the spatial extent of the COV patterns as exhibited in Figure 15. Cluster 4 representing the HBSC zone was limited to a narrow band along the Black Sea coast due to the northward expansion of Clusters 1 and 2 as observed with the COV patterns representing the same clusters. One of the most noticeable shifts in the cluster domains is observed in the 2005–2009 period when the entire Black Sea, northern Eastern Anatolia, and Eastern Central Anatolia were classified within the same low COV variability, suggesting that the year-to-year precipitation during this 5-year sub- period is a more stable precipitation regime over a large area (Figure 16). During the 2010–2014 sub-period, stable precipitation areas are confined to a narrow band in the HBSC zone, while the SCCA precipitation cluster shifts northward. One of the most striking features observed with the COV values occurs during the 2015–2020 sub-period when larger COV values shift southward in western Turkey while shifting northward in eastern Turkey. This suggests that precipitation became more uniform in most areas of Turkey, except for a narrow zone along the Mediterranean coast characterized by the largest COV values.
Figure 16

Spatial variation of the COV for the 5-year sub-periods.

Figure 16

Spatial variation of the COV for the 5-year sub-periods.

Close modal

The 5-year sub-period analysis indicates that the precipitation over the Black Sea coastal areas is becoming more uniform throughout the year and lacking significant year-to-year variability, while the coastal Mediterranean region has more irregular precipitation during the year as it receives a great portion of its precipitation in the winter months. Overall, we can argue that no significant spatial shifts in the cluster patterns are observed in the lower COV areas (i.e. Black Sea region) where the monthly precipitation is more uniform throughout the year while larger inter-decadal variability is observed in the higher COV areas (i.e. Mediterranean region) where seasonality in precipitation is more pronounced. We find that the precipitation regime of a region with more irregular behaviour and a dominant seasonal component is more vulnerable to spatial shifts than a region that has a more uniform monthly precipitation distribution throughout the year.

In this study, the K-means clustering is applied to define precipitation regions of Turkey and identify variations in the clusters' spatial extensions in subsequent 5- and 10-year periods because of climate variability. Our results found that the number and the spatial domains of the precipitation regions changed considerably throughout the study period reflecting regional climate variability impacts. As the domain of the clusters expanded or contracted, some stations did not retain their original position within a cluster but rather exhibited spatially variable behaviour as they were classified under another cluster when the sub-periods changed. The main conclusion of the study is that the precipitation regions are not spatially coherent over time and considerable variability can be manifested even between different regions of the country at an inter-decadal time scale.

Five distinct precipitation clusters (Figure 7) are revealed when the entire 1980–2020 periods are taken into account, each reflecting different characteristics of Turkish precipitation. The identified clusters include:

  • 1.

    SCM cluster

  • 2.

    EA cluster

  • 3.

    Subtropical semi-humid/humid coastal Mediterranean (SS/HCM) cluster

  • 4.

    HBSC cluster and

  • 5.

    SCCA cluster

The SCM cluster appeared to be a more continental form of a typical Mediterranean precipitation region with high summer temperatures. The SS/HCM cluster differed from the SCM exhibiting wetter conditions and extended north to cover the western Marmara region. The semi-arid central Anatolian precipitation cluster encompasses the largest precipitation cluster. The wetter Mediterranean precipitation region consisted of one single region with an extensive spatial coverage. The resulting cluster structure reveals the semi-arid character of the Turkish climate extending towards the north and the east which are more humid than the central Anatolian parts of the country.

Clustering of the 10-year subsequent periods resulted in the number of clusters varying between four and eight indicating considerable decadal variability. During the first 10-year period (1980–1989), Marmara, Aegean, and Mediterranean regions appeared in a cluster matching the general pattern except with less spatial extent. The last period including the 2010–2020 years displayed more diverse cluster structures with a total of seven clusters. During these 10 years, the Eastern Mediterranean region is characterized as a separate cluster, whereas the Aegean and Marmara regions formed two different clusters during the 2010–2020 period.

The number of clusters for the 5-year sub-periods varied between three and eight, portraying a more variable pattern than the 10-year sub-period patterns. The main clustering feature observed during the 5-year sub-periods was the entire Aegean, Marmara, and most parts of the Mediterranean regions that were classified as one single region in the first two sub-periods. Except for the first 5-year sub-period, the continental form of the Mediterranean rainfall regime dominates Southeastern Anatolia, indicating less variability in its spatial extension as compared to the other clusters. Occasionally the Humid Coastal Black Sea cluster, which usually has a well-defined coastal extension, expanded further west into the Marmara region forming one single cluster in some 5-year sub-periods. The 2015–2020 period portrayed the most striking cluster features of all 5-year sub-periods that are characterized by semi-arid features in Central Anatolia that expanded its boundaries reflecting warming temperatures and reduced precipitation. Moreover, the moist north Black Sea zone extended further west during the last 5-year sub-period occupying the Marmara region, causing the retreat of the less moist Aegean climate zone.

The variable precipitation patterns observed in the decadal and inter-decadal periods highlight the dynamic behaviour of Turkish monthly precipitation. Not only do the number of precipitation regions change, but also their spatial coverage enlarges or shrinks even during 5-year sub-periods. The variability of the clusters may also affect precipitation regimes in the future. For example, if the Humid Coastal Black Sea cluster keeps expanding westward in future periods, it can modify the precipitation regime of the Marmara region and cause more uniform precipitation distribution and less inter-annual variability. Similarly, penetration of the SCCA cluster northward may bring the more semi-humid and semi-arid character of the SCCA precipitation zone into the Black Sea region and alter its uniform precipitation distribution pattern. Such variability can have significant implications for water availability and water allocation for various water-reliant sectors in the country as the precipitation regime will change during inter-decadal and decadal periods. Future climate change may exert additional pressure on the country's water resources, and demand for water by different sectors might increase during the next 2–3 decades. Therefore, analysis of the regional precipitation variability using the observed station data can shed a light on future studies to project such variabilities with more accuracy and assess its implications on the country's water resources more efficiently.

It is our view that although the physiographic factors, such as topography and elevation, influence the general pattern of the precipitation regions, climate variability resulting from teleconnection patterns, such as AO and NAO, impacts the dynamics of the precipitation patterns over Turkey. The monthly precipitation totals were correlated with AO more significantly in the coastal regions except for the Black Sea region, while the NAO yielded a better correlation at Eastern and Southeastern Anatolian stations. All the stations in Southeastern Anatolia were significantly correlated with NAO, while the Marmara region exhibited the highest percentage of correlation with AO, namely 87%.

In order to investigate and compare the coherency of the observed station precipitation variability with the spatial changes of the clusters for the respective sub-periods, an analysis of the COV is performed. We attained two main conclusions from this analysis. First, the cluster structures tend to retain their spatial domain in the lower COV areas and do not exhibit a significant shift in subsequent sub-periods. Second, the areas characterized by the lower COV have a more uniform precipitation regime, while areas of high COV have more irregular precipitation during the year and receive a great portion of their precipitation in the winter months. In other words, COV values tend to be higher in regions (such as the Mediterranean region) where the marked seasonal regime and winter precipitation are dominant, while they decrease in areas (such as the Black Sea region) where precipitation is relatively uniform throughout the year. An important conclusion reached based on the COV values is that the cluster structures are more stable in areas where the lower COV values dominate, such as in the HBSC cluster, while the variability of the clusters in the subsequent sub-periods increases in the areas where the COV values are higher, such as in the Southeastern Anatolian and Mediterranean region clusters, identified as the SCM cluster. In summary, we find that the areas with highly variable precipitation regimes tend to shift more spatially. The analysis of the spatial distribution of the COV values can also be crucial for understanding the risk of extreme events. The regions identified with higher inter-decadal and decadal variability in precipitation may be more susceptible to drought events.

Another important conclusion reached in the study is the selection of different distance methods for determining the optimal number of classes for silhouette analysis using the K-means clustering technique. Our study found that distance measures may vary among the different sub-periods due to the inherent nature of temporal variability of precipitation, and therefore any assumption of choosing a single distance measure by default should be avoided and an appropriate distance measure should be searched. In this study, primarily the cosine method appeared to be an optimum distance method except for very few cases when the correlation distance method was the resulting distant metric. The optimum distance metrics were determined based on the optimum cluster number, considering the minimum negative values and the maximum average silhouette value for the respective sub-periods.

In order to thoroughly investigate the factors affecting the spatial clustering of the precipitation regions and, more specifically, how the clusters are structured at decadal and inter-decadal time scales, additional research is needed, especially concerning climate variability. The authors plan to focus future work on a more detailed investigation of the role of the teleconnection patterns and large-scale atmospheric patterns on the variability of precipitation clusters, given the results found in this study. Moreover, it is unclear how future changes in climate may affect the current clustering structure. Climate projections indicate decreasing rainfall throughout Turkey, particularly in the southern and inner parts of the country, and irregularities in rainfall regimes have already been noticed (Demircan et al. 2017). If such changes impact the amount of local and regional precipitation, the defined clusters may eventually shift and reorient. Therefore, further attention needs to be focused on how the projected climate change may redistribute the existing precipitation regions.

The spatial patterns in precipitation regions and their variability as defined in this study have significant implications for the water-related sectors in Turkey considering the irregularities and seasonality of the monthly precipitation. Hence, it is our view that understanding temporal and spatial changes in precipitation patterns and their groupings is a critically important aspect of planning and management of water resources including planning for drinking water supplies, groundwater, rivers, hydroelectric production, agricultural activities, and other water-related sectors in the country in future. Such information also provides invaluable input to authorities in taking adaptation and mitigation measures to minimize the impacts of extreme climatic events such as floods and droughts common in recent years. We expect that the results of this study will provide essential scientific settings when preparing the country's medium and long-term water resources, and agricultural and drought management plans. Last but not least, knowledge of the homogenous precipitation regions’ identification and their inter-decadal variability consideration can pave the way for a better scientific basis to understand the role of climate variability on the precipitation regime in different temporal scales over the country.

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

All authors contributed to the study's conception and design. Material preparation, data collection, and analysis were performed by A.Ü.K. and E.T. The first draft of the manuscript was written by A.Ü.K., and all authors commented on previous versions of the manuscript. T.D. edited the final version of the manuscript. All authors read and approved the final manuscript.

MATLAB codes for the K-means analysis are available upon request.

Data cannot be made publicly available; Data is available upon request of the authors.

The authors declare there is no conflict.

Adjez
A.
2000
Changes in the rainfall regime in the North-African region: links with large-scale oscillation phenomena
. In:
Conference Abstracts ‘Detection and Modeling of Recent Climate Change and its Effects on a Regional Scale
,
Tarragona
. Vol.
24
.
Alexandersson
H.
1986
A homogeneity test applied to precipitation data
.
Journal of Climatology
6
,
661
675
.
https://doi.org/10.1002/joc.3370060607
.
Atalay
I.
2016
A new approach to the land capability classification: a case study of Turkey
.
Procedia Environmental Sciences
32
,
264
274
.
doi:10.1016/j.proenv.2016.03.031
.
Berkhin
P.
2002
Survey of Clustering Data Mining Techniques. Technical Report
.
Accrue Software
,
San Jose, CA
.
Berkhin
P.
,
2006
A survey of clustering data mining techniques
. In:
Grouping Multidimensional Data
(
Kogan
J.
,
Nicholas
C.
&
Teboulle
M.
, eds).
Springer
,
Berlin, Heidelberg
.
doi:10.1007/3-540-28349-8_2
.
Bock
H. H.
,
2007
Clustering methods: a history of k-means algorithms
. In:
Selected Contributions in Data Analysis and Classification. Studies in Classification, Data Analysis, and Knowledge Organization
(
Brito
P.
,
Cucumel
G.
,
Bertrand
P.
&
de Carvalho
F.
, eds).
Springer
,
Berlin, Heidelberg
.
Brönnimann
S.
2007
Impact of El Niño-Southern Oscillation on European climate
.
Reviews of Geophysics
5
.
doi:10.1029/2006RG000199
.
Brown
C. E.
1998
Coefficient of variation
. In:
Applied Multivariate Statistics in Geohydrology and Related Sciences
.
Springer
,
Berlin, Heidelberg
.
doi:10.1007/978-3-642-80328-4_13
.
Choi
J. W.
&
Byun
H. R.
2010
Possible relationship between western North Pacific tropical cyclone activity and Arctic oscillation
.
Theoretical and Applied Climatology
100
,
261
274
.
doi:10.1007/s00704-009-0187-9
.
Cullen
H. M.
&
deMenocal
P.
2000
North Atlantic influence on Tigris-Euphrates streamflow
.
International Journal of Climate
20
,
853
863
.
Darand
M.
&
Mansouri Daneshvar
M. R.
2014
Regionalization of precipitation regimes in Iran using principal component analysis and hierarchical clustering analysis
.
Environmental Processes
1
,
517
532
.
https://doi.org/10.1007/s40710-014-0039-1
.
Davey
M. K.
,
Brookshaw
A.
&
Ineson
S.
2014
The probability of the impact of ENSO on precipitation and near-surface temperature
.
Climate Risk Management
(
1
),
5
24
.
doi:10.1016/j.crm.2013.12.002
.
Demircan
M.
,
Gürkan
H.
,
Eskioğlu
O.
,
Arabacı
H.
&
Coşkun
M.
2017
Climate change projections for Turkey: three models and two scenarios
.
Turkish Journal of Water Science & Management
1
,
22
43
.
doi:10.31807/tjwsm.297183
.
Deser
C.
,
Hurrell
J. W.
&
Phillips
A. S.
2017
The role of the North Atlantic Oscillation in European climate projections
.
Climate Dynamics
49
,
3141
3157
.
doi:10.1007/s00382-016-3502-z
.
Dünkeloh
A.
&
Jacobeit
J.
2003
Circulation dynamics of Mediterranean precipitation variability 1948–98
.
International Journal of Climatology
23
,
1843
1866
.
doi:10.1002/joc.973
.
Everitt
B.
,
Landau
S.
,
Leese
M.
&
Stahl
D.
2011
Cluster Analysis
, 5th edn.
Wiley
,
Chichester
.
http://dx.doi.org/10.1002/9780470977811
.
Feldstein
S.
,
Franzke
C.
,
2017
Atmospheric teleconnection patterns
. In:
Nonlinear and Stochastic Climate Dynamics
(
Franzke
C.
&
O'Kane
T.
, eds).
Cambridge University Press
,
Cambridge
, pp.
54
104
.
doi:10.1017/9781316339251.004
.
Gong
D. Y.
,
Wang
S. W.
&
Zhu
J. H.
2001
East Asian winter monsoon and Arctic Oscillation
.
Geophysical Research Letters
28
(
10
),
2073
2076
.
Gore
P. A.
2000
Cluster analysis
. In:
Handbook of Applied Multivariate Statistics and Mathematical Modeling
.
Academic Press
, pp
297
321
.
doi:10.1016/B978-012691360-6/50012-4
.
Halpert
M. S.
&
Ropelewski
C. F.
1992
Temperature patterns associated with the Southern Oscillation
.
Journal of Climate
5
,
577
593
.
Hurrell
J. W.
1995
Decadal trends in the North Atlantic Oscillation
.
Science
269
,
676
679
.
doi:10.1126/science.269.5224.676
.
Irwin
S.
,
Simonovic
S. P.
&
Burn
D. H.
2017
Delineation of precipitation regions in two Canadian study areas: the role of the temporal resolution of the precipitation data
.
Hydrological Sciences Journal
62
(
13
),
2061
2071
.
doi:10.1080/02626667.2017.1353694
.
Iyigun
C.
,
Türkeş
M.
,
Batmaz
İ.
,
Yozgatligil
C.
,
Purutçuoğlu
V.
,
Kartal Koç
E.
&
Öztürk
M. Z.
2013
Clustering current climate regions of Turkey by using a multivariate statistical method
.
Theoretical and Applied Climatology
114
,
95
106
.
doi:10.1007/s00704-012-0823-7
.
Jacobeit
J.
,
Jones
P.
,
Davies
T.
,
Beck
C.
,
2001
Circulation changes in Europe since the 1780s
. In:
History and Climate
(
Jones
P. D.
,
Ogilvie
A. E. J.
,
Davies
T. D.
&
Briffa
K. R.
, eds).
Springer
,
Boston, MA
.
https://doi.org/10.1007/978-1-4757-3365-5_5
.
Jajuga
K.
,
Sokolowski
A.
&
Bock
H. H.
2002
Classification, Clustering and Data Analysis
.
Springer-Verlag
,
Berlin
,
Heidelberg
,
New York
.
Jevrejeva
S.
,
Moore
J.
&
Grinsted
A.
2003
Influence of the Arctic Oscillation and El Niño-Southern Oscillation (ENSO) on ice conditions in the Baltic Sea: the wavelet approach
.
Journal of Geophysical Research
108
.
doi:10.1029/2003JD003417
.
Kadioğlu
M.
,
Tulunay
Y.
&
Borhan
Y.
1999
Variability of Turkish precipitation compared to El Niño events
.
Geophysical Research Letters
26
(
11
),
1597
1600
.
doi:10.1029/1999GL900305
.
Kahya
E.
&
Karabörk
M.
2001
The analysis of El Niño and La Nina signals in streamflows of Turkey
.
International Journal of Climatology
21
,
1231
1250
.
doi:10.1002/joc.663
.
Karabörk
M.
,
Kahya
E.
&
Karaca
M.
2005
The influences of the Southern and North Atlantic Oscillations on climatic surface variables in Turkey
.
Hydrological Processes
19
,
1185
1211
.
doi:10.1002/hyp.5560
.
Kaufman
L.
&
Rousseeuw
P. J.
1990
Finding Groups in Data: An Introduction to Cluster Analysis
.
Wiley
,
New York, NY
,
USA
.
Kiladis
G. N.
&
Diaz
H. F.
1989
Global climatic anomalies associated with extremes in the Southern Oscillation
.
Journal of Climate
2
,
1069
1090
.
https://doi.org/10.1175/1520-0442(1989)002<1069:gcaawe>2.0.co;2
.
Kömüşcü
A. Ü.
&
Oğuz
K.
2021
Analysis of cold anomalies observed over Turkey during the 2018/2019 winter in relation to polar vortex and other atmospheric patterns
.
Meteorology and Atmospheric Physics
133
,
1327
1354
.
doi:10.1007/s00703-021-00806-0
.
Lin
J.
&
Qian
T. A.
2019
New picture of the global impacts of El Niño-Southern Oscillation
.
Science Report
9
,
17543
.
doi:10.1038/s41598-019-54090-5
.
Lionello
P.
,
Malanotte-Rizzoli
P.
,
Boscolo
R.
,
Alpert
P.
,
Artale
V.
,
Li
L. Z.
,
Luterbacher
J.
,
May
W.
,
Trigo
R. M.
,
Tsimplis
M.
,
Ulbrich
U.
&
Xoplaki
E.
2006
The Mediterranean climate: An overview of the main characteristics and issues. 1 (1-26), Developments in Earth and Environmental Sciences., 1st edn. Elsevier Science. eBook ISBN: 9780080460796
.
López-Parages
J.
&
Rodriguez-Fonseca
M.
2012
Multidecadal modulation of El Niño influence on the Euro-Mediterranean rainfall
.
Geophysical Research Letters
39
(
2
),
533
.
doi:10.1029/2011GL050049
.
Machiwal
D.
,
Kumar
S.
,
Meena
H. M.
,
Santra
P.
,
Singh
R.
&
Singh
D.
2019
Clustering of rainfall stations and distinguishing influential factors using PCA and HCA techniques over western dry region of India
.
Meteorological Applications
26
(
2
),
300
311
.
doi:10.1002/met.1763
.
Mariotti
A.
,
Zeng
N.
&
Lau
W.
2002
Euro-Mediterranean rainfall and ENSO − a seasonally varying relationship
.
Geophysical Research Letters
29
.
doi:10.1029/2001GL014248
.
Marshall
J.
,
Kushnir
Y.
,
Battisti
D.
,
Chang
P.
,
Chaja
A.
,
Dickson
R.
,
Hurrell
J.
,
McCartney
M.
,
Saravanan
R.
&
Visbeck
M.
2001
North Atlantic climate variability: phenomena, impacts and mechanisms
.
International Journal of Climatology
21
,
1863
1898
.
Marston
M. L.
&
Ellis
A. W.
2021
Delineating precipitation regions of the contiguous United States from cluster analyzed gridded data
.
Annals of the American Association of Geographers
111
(
6
),
1721
1739
.
doi:10.1080/24694452.2020.1828803
.
MATLAB and Statistics Toolbox Release
2012b
The MathWorks, Inc., Natick, MA, USA
.
McPhaden
M.
,
Zebiak
S.
&
Glantz
M.
2006
ENSO as an integrating concept in earth science
.
Science
314
,
1740
1745
.
doi:10.1126/science.1132588
.
Netzel
P.
&
Stepinski
T.
2016
On using a clustering approach for global climate classification
.
Journal of Climate
29
(
9
),
3387
3401
.
Nigam
S.
,
Baxter
S.
,
2015
General circulation of the atmosphere | teleconnections
. In:
Encyclopedia of Atmospheric Sciences (Second Edition)
(
North
G. R.
,
Pyle
J.
&
Zhang
F.
, eds).
Academic Press
, pp.
90
109
.
ISBN 9780123822253
.
doi:10.1016/B978-0-12-382225-3.00400-X
.
Nisbet
R.
,
Elder
J.
&
Miner
G.
2009
The Handbook of Statistical Analysis and Data Mining Applications
.
Academic Press
, p.
864
.
doi:10.1016/B978-0-12-374765-5.X0001-0
.
Pettitt
A. N.
1979
A non-parametric approach to the change-point problem
.
Journal of the Royal Statistical Society: Series C (Applied Statistics)
28
,
126
135
.
https://doi.org/10.2307/2346729
.
Rodriguez
M. Z.
,
Comin
C. H.
,
Casanova
D.
,
Bruno
O. M.
,
Amancio
D. R.
,
Costa
L.
,
Costa
L.
&
Rodrigues
F. A.
2019
Clustering algorithms: a comparative approach
.
PLoS ONE
14
(
1
),
e0210236
.
doi:10.1371/journal.pone.0210236
.
Rokach
L.
&
Maimon
O.
2005
Clustering methods
. In:
Data Mining and Knowledge Discovery Handbook
.
Springer
,
Boston, MA
.
https://doi.org/10.1007/0-387-25465-X_15
.
Ropelewski
C. F.
&
Halpert
M. S.
1996
Quantifying Southern oscillation − precipitation relationships
.
Journal of Climate
9
,
1043
1059
.
Rousseeuw
P. J.
1987
Silhouettes: a graphical aid to the interpretation and validation of cluster analysis
.
Journal of Computational and Applied Mathematics
20
,
53
65
.
doi:10.1016/0377-0427(87)90125-7
.
Rust
W.
,
Bloomfield
J.
,
Cuthbert
M.
,
Corstanje
R.
&
Holman
I.
2022
The importance of non-stationary multiannual periodicities in the North Atlantic Oscillation index for forecasting water resource drought
.
Hydrology and Earth System Sciences
26
(
9
),
2449
2467
.
Sankranti
K.
,
Rao
T. N.
,
Rajeevan
M.
&
Bhaskara Rao
S. V.
2013
Identification and validation of homogeneous rainfall zones in India using correlation analysis
.
Journal of Hydrometeorology
14
(
1
),
304
317
.
doi:10.1175/JHM-D-12-071.1
.
Sarış
F.
,
Hannah
D. M.
&
Eastwood
W. J.
2010
Spatial variability of precipitation regimes over Turkey
.
Hydrological Sciences Journal–Journal des Sciences Hydrologiques
55
(
2
),
234
249
.
doi.10.1080/02626660903546142
.
Sezen
C.
&
Partal
T.
2019
The impacts of Arctic oscillation and the North Sea Caspian pattern on the temperature and precipitation regime in Turkey
.
Meteorology and Atmospheric Physics
131
,
1677
1696
.
doi:10.1007/s00703-019-00665-w
.
Shahapure
K. R.
&
Nicholas
C.
2020
Cluster quality analysis using silhouette score, 2020
. In:
IEEE 7th International Conference on Data Science and Advanced Analytics (DSAA)
, pp.
747
748
.
doi: 10.1109/DSAA49011.2020.00096
.
Singh
A.
,
Yadav
A.
&
Rana
A.
2013
K-means with three different distance metrics
.
International Journal of Computer Applications
67
(
10
),
13
17
.
doi:10.5120/11430-6785
.
Sönmez
İ
.
2013
Quality control tests for western Turkey Mesonet
.
Meteorological Applications
20
(
3
),
330
337
.
doi:10.1002/met.1286
.
Sreevalsan-Nair
J.
,
McKinley
J.
,
Agterberg
F.
,
2022
K-means clustering
. In:
Encyclopedia of Mathematical Geosciences. Encyclopedia of Earth Sciences Series
(
Daya Sagar
B.
&
Cheng
Q.
, eds).
Springer
,
Cham
.
doi:10.1007/978-3-030-26050-7_171
.
Tatli
H.
,
Dalfes
N. H.
&
Menteş
S.
2004
A statistical downscaling method for monthly total precipitation over Turkey
.
International Journal of Climatology
24
,
161
180
.
Thakare
Y. S.
&
Bagal
S. B.
2015
Performance evaluation of K-means clustering algorithm with various distance metrics
.
International Journal of Computer Applications
110
,
12
16
.
Thompson
D. W. J.
&
Wallace
J. M.
2000
Annular modes in the extratropical circulation, part I: month-to-month variation
.
Journal of Climate
13
,
1000
1016
.
doi:10.1175/1520-0442
.
Tuel
A.
&
Martius
O.
2021
A global perspective on the sub-seasonal clustering of precipitation extremes
.
Weather and Climate Extremes
33
,
100348
.
doi:10.1016/j.wace.2021.100348
.
Türkeş
M.
1996
Spatial and temporal analysis of annual rainfall variations in Turkey
.
International Journal of Climatology
16
,
1057
1076
.
Türkeş
M.
&
Erlat
E.
2006
Influences of the North Atlantic Oscillation on precipitation variability and changes in Turkey
.
Nuovo Cimento Della Societa Italiana Di Fisica C– Geophysics and Space Physics
29
,
117
135
.
Türkeş
M.
&
Erlat
E.
2008
Influence of the Arctic oscillation on variability of winter mean temperatures in Turkey
.
Theoretical and Applied Climatology
92
,
75
85
.
Türkeş
M.
&
Erlat
E.
2009
Winter mean temperature variability in Turkey associated with the North Atlantic Oscillation
.
Meteorology Atmospheric Physics
105
,
211
225
.
https://doi.org/10.1007/s00703-009-0046-3.
Türkeş
M.
&
Tatlı
H.
2011
Use of the spectral clustering to determine coherent precipitation regions in Turkey for the period 1929–2007
.
International Journal of Climatology
31
,
2055
2067
.
doi:10.1002/joc.2212
.
Türkeş
M.
,
Koç
T.
&
Sarış
F.
2008
Spatiotemporal variability of precipitation total series over Turkey
.
International Journal of Climatology
.
doi:10.1002/joc.1768
.
Ünal
Y.
,
Kindap
T.
&
Karaca
M.
2003
Redefining the climate zones of Turkey using cluster analysis
.
International Journal of Climatology
23
,
1045
1055
.
Ünal
Y.
,
Deniz
A.
,
Toros
H.
&
Incecik
S.
2012
Temporal and spatial patterns of precipitation variability for annual, wet, and dry seasons in Turkey
.
International Journal of Climatology
32
(
3
),
392
405
.
Wazneh
H.
,
Chebana
F.
&
Ouarda
T. B. M. J.
2015
Delineation of homogeneous regions for regional frequency analysis using statistical depth function
.
Journal of Hydrology
521
,
232
244
.
Wilks
D. S.
2011
Cluster Analysis, Statistical Methods in the Atmospheric Sciences, International Geophysics
, Vol.
100
.
Academic Press
, pp.
603
616
.
doi.10.1016/B978
.
Zaifoglu
H.
,
Akintug
B.
&
Yanmaz
A.
2018
Regional frequency analysis of precipitation using time series clustering approaches
.
Journal of Hydrologic Engineering
23
(
6
),
1
13
.
https://doi:10.1061/(ASCE)HE.1943-5584.0001659
.
Zhang
X.
,
Jin
L.
,
Chen
C.
,
Guan
D.
&
Li
M.
2011
Interannual and interdecadal variations in the North Atlantic Oscillation spatial shift
.
Chinese Science Bulletin
56
,
2621
2627
.
doi.10.1007/s11434-011-4607-8
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY-NC-ND 4.0), which permits copying and redistribution for non-commercial purposes with no derivatives, provided the original work is properly cited (http://creativecommons.org/licenses/by-nc-nd/4.0/).