This study aimed to identify primary modes of annual and seasonal precipitation in Turkey using a rotated Empirical Orthogonal Function (EOF) method, and the resulting patterns were described concerning the prevalent atmospheric circulations, orography, and continentality. The varimax rotation of the EOF determines modes that are more localized in space than the conventional EOF modes The first three EOFs accounted for approximately 67% and 62% of the total variance in the annual and wet season precipitation series, respectively, whereas only 50% of the variance was captured by the first three EOFs for dry season precipitation. Spatially different atmospheric circulation mechanisms are major drivers of variability in Turkish precipitation on annual and seasonal timescales. The spatial coherence of the highest negative and positive EOF1 loadings of the annual data was observed in the western and southern regions where westerly and northwesterly circulations prevailed during the wet season. A lower spatial coherency was observed with the dry season precipitation. The contribution of atmospheric moisture advection to precipitation variability diminishes in summer, whereas that of local land surface processes increases. Some regional teleconnection patterns, such as the Arctic Oscillation (AO) and North Atlantic Oscillation (NAO), also contributed to the annual variability in precipitation.

  • Long-term spatial and temporal changes in precipitation variability in Turkey.

  • Rotated empirical orthogonal function (REOF) approach for retrieving the spatial patterns and identifying probable driving mechanisms.

  • Large-scale and subsynoptic atmospheric circulation patterns in shaping the precipitation patterns.

  • Role of the teleconnection patterns such as NAO and AO in spatial variability of precipitation.

An improved depiction of the spatial variability of precipitation and understanding its underlying causes are not only essential but also crucial for providing a sound scientific basis for planning and project development in water-reliant sectors and managing water allocation among different sectoral users of water. A good perception of precipitation variability can also provide decision-makers with fundamental information for hazard management of hydrometeorological extremes, such as floods and droughts (Bruni et al. 2015; Cristiano et al. 2017). A well-defined representation of the spatial patterns of precipitation is desirable for more efficient design and operation of engineering structures for flood control, water conservation, and agricultural irrigation. Moreover, insights into precipitation variability are critical in the decision-making process for climate change-induced impacts on water resources, agriculture, and energy sectors as well as on environmental planning (Longobardi & Boulariah 2022). Thorough knowledge of precipitation variability is particularly crucial in Turkey because its highly seasonal and erratic precipitation patterns have far-reaching effects on several critical sectors of the economy, including water resources, agriculture, and hydropower (Unal et al. 2012; Kömüşcü & Aksoy 2023). Any change in precipitation patterns can significantly affect the water allocation in sectors that compete for water at the desired time and quantity. Furthermore, characterizing the variable behaviour of precipitation in Turkey, which has recently been suffering from hydrological extremes more frequently, will provide decision-makers with invaluable inputs for more efficient management of meteorological and hydrological hazards. Because Turkey is located in a vulnerable geographic region concerning climate change impacts, an improved understanding of its precipitation variability and identifying the driving factors can provide invaluable scientific input for managing the adverse impacts of climate-induced changes on water resources, agricultural production, and hydropower generation, which are critical sectors of the Turkish economy (Sariş et al. 2010; Raja et al. 2017; Kömüşcü et al. 2022). The potential impacts of climate change on precipitation patterns may also pose a challenge for the effective management of the above-mentioned water-based sectors.

The spatial and temporal features of Turkish precipitation have been analysed by various scientists and researchers over the last few decades from different perspectives. Most of these studies covered the temporal or spatiotemporal aspects of precipitation data, including trends and periodicities (for example, Toros 1993; Türkeş 1996; Partal & Kahya 2006; Şen 2012, 2014; Unal et al. 2012; Yavuz & Erdoğan 2012; Ay & Kisi 2015; Çiçek & Duman 2015; Hadi & Tombul 2018; Partal 2018; Sezen & Partal 2020a; Topuz et al. 2021; Ay 2022; Kömüşcü et al. 2022). Kömüşcü & Aksoy (2023) examined the trends and cyclic features of the long-term monthly and seasonal precipitation data of Turkey and demonstrated that Turkey's monthly precipitation presented a declining trend during the 1975–2021 period while decreasing and consistently increasing trends were observed in the wet and dry seasons, respectively. However, they did not examine the variability of the spatial precipitation patterns in detail. In recent years, very few studies (e.g., Yılmaz et al. 2021; Acar & Gönençgil 2022) have attempted to explore the spatiotemporal dynamics of Turkish precipitation and the factors and/or processes controlling or governing its dynamics in the last decade, during which climate change arguments have dominated climate studies. Acar & Gönençgil (2022) studied the spatiotemporal characteristics of Turkish precipitation indices at 142 stations and found that NAO was the most governing index for variability in Turkey's precipitation indices. Yılmaz et al. (2021) examined monthly precipitation data from 74 stations in Turkey to reveal the seasonal and annual cycles with the data and associated the 3-year cycles with the NAO impact. However, both studies focused on the temporal aspects of precipitation using a limited number of stations. Kömüşcü et al. (2022) applied the K-means clustering technique to identify the main precipitation regions of Turkey and explore their dynamic behaviour over the decadal sub-periods. Their study showed that the precipitation subregions were characterized by considerable variability between different geographic regions of the country. Studies using empirical orthogonal function (EOF) analysis to identify spatial patterns of Turkish precipitation are limited and do not reflect the precipitation variability features peculiar to the last decade (for example, Kadıoğlu 2000; Türkeş et al. 2009; Unal et al. 2012). To fill the gaps of the previous studies and complement them, this study aims to add further insight into the dynamic nature of Turkish precipitation and probable driving factors/mechanisms behind its variable nature in connection with teleconnection patterns as well as large-scale and subsynoptic atmospheric circulation patterns.

Considering the essential role of water availability in water-reliant sectors, it is crucial to obtain a broad understanding of the monthly and seasonal precipitation variability over the country and relevant atmospheric and non-atmospheric processes and/or physical mechanisms, as well as their association with large-scale and regional circulation patterns. Therefore, this study attempts to provide a comprehensive insight into the spatial features of precipitation variability in Turkey by employing the rotated empirical orthogonal function (REOF)-based method and inquiring into the underlying and/or overriding processes that contributed to the formation of the identified spatial patterns. An EOF analysis using varimax rotation on the monthly annual, wet season, and dry season precipitation was performed to extract significant precipitation patterns at 213 stations across Turkey for the 1975–2021 period. Using these extracted patterns, we aimed to reveal the primary spatial structures of annual and seasonal precipitation variability, while their corresponding time series were expected to yield dominant temporal variability in the precipitation data. The underlying processes of the variability of the identified spatial patterns were then described concerning the prevalent large-scale and subsynoptic atmospheric circulation patterns in the region and non-meteorological factors (e.g., topography and continentality). Thus, the main objectives of this study are to

  • re-evaluate the spatial variability of the monthly based annual and seasonal (wet and dry) precipitation of Turkey by the use of EOF analysis to retrieve the leading spatial modes of the variability and discern coherent/non-coherent regions of the precipitation variations,

  • identify and describe the probable mechanisms/factors behind the retrieved spatial modes of the variability concerning the large-scale and regional circulation patterns, and other non-meteorological factors, including topography, continentality, land-sea proximity, and

  • to fill gaps in research on the spatial variability of Turkish precipitation using longer periods of data length and more data points to achieve improved and more representative spatial patterns of annual and seasonal precipitation.

The study area includes the geographic domain of Turkey, which occupies approximately 784,000 km2 between South-eastern Europe and Western Asia (Figure 1). Turkey's physiography is characterized by high and complex topographic features, including high plateaus, mountain ranges, and coastal fringes. Its mean elevation exceeds 1,130 m, and nearly one-fourth of the country's terrain settles at an elevation above 1,200 m and even exceeds 1,500 m in the Eastern Anatolia region (Atalay & Mortan 2008; Atalay 2016). Turkey's climate primarily reflects the features of the Mediterranean macroclimatic subregion of the subtropical zone. The climate of the three coastal regions, including the Marmara, Aegean, Mediterranean regions, and South-eastern Anatolia region, is classified as a dry summer subtropical Mediterranean climate (Csa) based on the Köppen–Geiger climate classification (Öztürk et al. 2017; Beck et al. 2018). Along with the influence of different seasonally shifting atmospheric pressure systems and air masses, the diverse topographic features, the presence of mountain chains, continentality, and land-sea contrast cause highly variable climatic conditions, resulting in several climatically diverse subregions across Turkey. The North Anatolian and the Taurus Mountains strongly influence the local climate, producing contrasting temperature and precipitation conditions between the interior parts and the coastal lowlands.
Figure 1

Geographic location of the study area with the elevation information (Adapted from https://cgiarcsi.community/data/srtm-90m-digital-elevation-database-v4-1).

Figure 1

Geographic location of the study area with the elevation information (Adapted from https://cgiarcsi.community/data/srtm-90m-digital-elevation-database-v4-1).

Close modal

Several studies have extensively discussed Turkey's climate from various perspectives (Çiçek 1996; Türkeş 1996, 2020; Tayanç et al. 1997; Kadıoğlu 2000; Karaca et al. 2000; Komuscu 2001; Fahmi et al. 2011; Deniz et al. 2011; Yılmaz & Çicek 2016; Gönençgil 2020). Çiçek (1996) used the Thornthwaite method to classify Turkey's subclimates based on precipitation, temperature, and potential evapotranspiration (PET) data from 202 stations. Yılmaz & Çiçek (2016) improved this study using the same methodology but with gridded data at a resolution of 1 km in four different categories and produced integrated climate class maps, indicating that nearly 34% of the country was classified in the dry-semi-humid category. A more innovative methodology called principal component analysis (PCA) was first used by Kadıoğlu (2000) to study meteorological and non-meteorological factors that affect seasonal precipitation variability in Turkey. The study attributed the winter precipitation variability to general atmospheric circulation patterns, while continentality and maritime influence were more dominant during the transitional months. Gönençgil (2020) introduced a clustering analysis approach to classify the diverse climatic features of Turkey by integrating different climate indices. Finally, Türkeş (2020) brought a new perspective to the climate classification of Turkey by integrating meteorological, hydrometeorological, and physical geography approaches and linked the resulting climate regimes to drought probabilities and risks.

The spatial distribution of Turkey's annual precipitation representing the 1991–2020 climatological normal period is 574 mm, approximately 70% of which falls during the wet period from October to May. The annual precipitation shows diverse geographic patterns and varies sharply between coastal areas and interior plains, ranging from approximately 300 mm in central Anatolia to more than 2,000 mm in the coastal areas of the eastern Black Sea region. Topographic influences resulting from diverse terrain features, high elevation, and proximity to the sea cause significant variability in precipitation at the local scale. Precipitation in the western and southern parts of the country, where the Mediterranean climate dominates, exhibits more seasonality, whereas it is more evenly distributed throughout the year in the northern parts (Kömüşcü et al. 2022).

The seasonal movement of different air masses of polar and tropical origin and changes in the strength of the regional circulation systems determine the dynamic aspects of Turkish precipitation. Subtropical highs and prevailing westerlies are two prominent atmospheric circulation patterns that affect the seasonal precipitation conditions in the country (Türkeş & Erlat 2008). Winter season precipitation in the western and southern parts of Turkey is primarily a product of marine-polar (mP) air masses originating from the Atlantic Ocean. The mP air masses reaching the Mediterranean Basin are enhanced by relief-induced changes and form moist and unstable Mediterranean air masses after warming by the Mediterranean Sea. Karaca et al. (2000) argue that the majority of the cyclones that affect Turkey originate from the Genoa Gulf, and then travel eastward to reach the inner parts of the country after passing over the Aegean Sea. Overall, the Mediterranean Sea significantly influences Turkey's weather patterns as it provides moisture for cyclone development (Trigo et al. 1999). In the October-April period, cyclones formed over the Mediterranean Sea with frontal characteristics preceding the development of strong rainfall in Turkey's coastal Mediterranean regions (Ceylan & Kömüşcü 2008). Furthermore, Turkey's climate is strongly altered by various atmospheric pressure systems, including Iceland's Low, Azores High, Siberian High, and Monsoon Low (Akçar et al. 2007). Their seasonal shifts and strengths govern temperature and precipitation conditions in most parts of the country. When the Azores High is strengthened, drier-than-normal conditions occur as the storm tracks shift northward, causing fewer intense storms to reach and affect the country (Yılmaz et al. 2021). In May–September, Turkey receives relatively lower precipitation, except for the Black Sea coastal region, where the Atlantic Ocean-sourced north-westerly flows and the mP air masses provide the moisture needed for orographically induced precipitation.

This study used monthly precipitation data obtained from 213 automated observing stations (AWOS) dispersed across Turkey during the 1975–2021 period (Figure 2). The data were downloaded from https://mevbis.mgm.gov.tr/mevbis/ui/index.html#/Workspace. The highest possible number of stations with good quality data and a sufficiently long data period was chosen to reflect the dynamic nature of Turkish precipitation variability. The selected stations are part of a meteorological network operated by the Turkish State Meteorological Service (TSMS). Both automated and manual quality controls (QC) were performed by the TSMS on a monthly basis before the data were made publicly available. The specifications of the QC tests implemented by the TSMS have been broadly described by Sönmez (2013). The QC tests that Sönmez (2013) used included automated quality control tests such as range, step, persistence, like-instrument and spatial (site-to-site), and temporal (month-to-month) at varying thresholds. The confidence level of the data quality was classified by one of the flag types, including ‘good, suspicious, or bad’ concerning the test applied. The station data were subjected to additional QC checks for temporal consistency (e.g., missing observations) and homogeneity tests before use in the EOF analysis. While 243 stations were initially included in the study, this number dropped to 213 based on the results of QC and homogeneity tests (Kömüşcü & Aksoy 2023). Thirty stations were eliminated because their data series were either inhomogeneous or included a significant number of incomplete observations during the selected period. The Alexanderson standard normal homogeneity (SNHT) test (Alexandersson 1986) and Pettit's test (Pettitt 1979) were applied to the precipitation series to detect inhomogeneity at the 95% confidence level; stations that exhibited inhomogeneity in at least one test were excluded. Moreover, any station with more than 5% of missing observations was omitted from the analysis. An arithmetic mean procedure was applied to fill the gaps in the data series at only four stations with less than 5% of the observations missing. The geographical distribution of the stations is shown in Figure 2, which illustrates that the stations were either included or excluded after the QC and homogeneity tests were applied.
Figure 2

Geographic distribution of the stations used in the study based on the geographic regions of Turkey along with topographic information. The numbers I, II, III, IV, V, VI, and VII designate geographic regions of Aegean, Black Sea, Central Anatolian, Eastern Anatolian, Marmara, Mediterranean, and South-eastern Anatolian regions, respectively.

Figure 2

Geographic distribution of the stations used in the study based on the geographic regions of Turkey along with topographic information. The numbers I, II, III, IV, V, VI, and VII designate geographic regions of Aegean, Black Sea, Central Anatolian, Eastern Anatolian, Marmara, Mediterranean, and South-eastern Anatolian regions, respectively.

Close modal

Regional and national data series were obtained using the arithmetic average method based on station data. The total precipitation values were converted to precipitation anomalies by using the 1991–2020 climatological normal for each station. It should also be noted that the ‘annual’ and ‘monthly’ series concepts are used interchangeably in this study. Annual data refer to the precipitation series of monthly data. The data for the monthly time series of each teleconnection index were accessed from the National Oceanic and Atmospheric Administration (NOAA) at https://www.cpc.ncep.noaa.gov/data and http://www.esrl.noaa.gov/psd/data/timeseries), and the University of East Anglia's Climate Research Unit at https://www.uea.ac.uk/web/groups-and-centres/climatic-research-unit/data.

The methodology used in this study is implemented in several steps. After filtering the precipitation input data, as described in the previous section, an EOF-based analysis was performed to identify prominent spatial patterns in the data via modes of variability known as EOFs (Figure 3). The EOF methodology is described in more detail in Section 4.1. The rotational EOF (REOF) approach was used in the analysis of the precipitation data; however, to avoid confusion regarding the terminology, we will use ‘EOF’ in the rest of the manuscript in its literal meaning.
Figure 3

Major steps pursued in the implementation of the EOF methodology.

Figure 3

Major steps pursued in the implementation of the EOF methodology.

Close modal

Empirical orthogonal function

EOF is a robust technique frequently employed in the analysis of time series to depict prevailing spatial patterns and the corresponding temporal variability of multivariate datasets in atmospheric studies (for example, Joyce 2002; Hannachi et al. 2007; Wilks 2011; Roundy 2015; Hannachi et al. 2023). This method has been used extensively in meteorology and hydrology since its introduction by Lorenz (1955) in the mid-1950s (Hannachi 2004; Hannachi et al. 2007). Navarra & Simoncini (2010) provided a practical guide for demonstrating the application of a traditional EOF method to climate data. EOF examines possible spatial modes, reducing them to spatial patterns explained by a few modes of variability known as EOFs. A typical approach to obtain EOFs is to compute the eigenvalues, and the derived eigenvalues represent a measure of the percent variance explained by each mode.

The EOF analysis decomposes a time-varying field T(x,y,t) into a set of mutually orthogonal spatial base functions Rn(x,y) and time series Pn(t).
formula

In the above equation, Rn (x,y) is called the nth EOF mode and Pn (t) is the principal component (PC). In climate data, the first few EOF modes typically explain the majority of total variance in the domain of analysis. As most of the total variance is accounted for by a few leading modes, EOF analysis permits one to choose the leading and sometimes physically meaningful modes of variability while significantly reducing the data space (Lian & Chen 2012). These modes are designated by orthogonal spatial patterns (eigenvectors) and corresponding time series (PCs). The main characteristics of temporal variability are represented by the coefficient time series (PCs) associated with the most important EOFs, the most relevant being PC1 associated with EOF1 (Hannachi 2004). The first EOF pattern (EOF1) is associated with the largest percentage of variance in the data, while the second EOF pattern (EOF2) explains the second largest percentage of the remaining variance. Identifying the spatial structures represented by the most variance provides valuable information in guiding the development of physical perception. The physical interpretation of EOFs, however, can sometimes be challenging owing to constraints resulting from orthogonality, and the derived patterns may be domain-dependent (Hannachi et al. 2006). Dommenget & Latif (2002) argue that extracted eigenvalues and identified spatial patterns in an EOF analysis can be responsive to the preference of the spatial domain and period selected, which may then deter providing proper physical meanings.

The rotated EOF (REOF) approach is preferred to alleviate some of the constraints of the classical EOF method (Unal et al. 2012). This approach attempts to retain the orthogonality of the modes, but the PCs remain orthogonal in time (NCAR 2013). In practice, the rotating EOFs approach helps to simplify the patterns obtained in EOFs and make them more interpretable. Moreover, the spatial modes would become more stable and robust concerning the unrotated EOFs. In this study, we employed the varimax rotation approach for orthogonal rotation to maximize the correlation between the obtained PCs and the original variables, and to ensure enhanced interpretability. As Turkish precipitation tends to exhibit localized spatial patterns owing to its high variability, it was thought that the varimax REOF approach would be a better choice for detecting localized patterns with more accuracy, thus facilitating their physical interpretation. One of the most well-known and used rotation algorithms is the varimax criterion suggested by Kaiser (1958). It is an orthogonal rotation based on the below criterion:
formula
where U= (uij), and m is the number of EOFs chosen for rotation. The quantity inside the square brackets in (2) is proportional to (spatial) variance of the square of the rotated vector uk= (u1k, . . ., upk)T. Therefore the varimax rotation aims to simplify the structure of the spatial modes by tending the loadings towards zero, or ±1.

The R software packages and libraries were used in the study for computing EOFs with varimax rotation and related quantities as well as producing graphics (R Core Team 2019). Further description of the R software packages and related libraries used for the EOF computations can be accessed via https://search.r-project.org/CRAN/refmans/wql/html/eof.html. An extensive discussion on the computational aspects of the REOF method and its application for climate data is available in Martinson (2018) and Navarra & Simoncini (2010), respectively.

Spatial analysis of monthly and seasonal precipitation

In this study, REOF analysis was employed to determine the leading modes for explaining the most variance in the monthly precipitation series at 213 locations in Turkey. The resulting spatial patterns were visualizations of the scores of the main components of each mode. A plotting order rank technique of ±2 was used to represent wet or dry anomalies in the precipitation series. The specifics of this procedure have been broadly discussed by Makkonen (2006). It is suggested that the concept of distribution-specific plotting formulas in analysing return periods should be discarded and the Weibull plotting formula P = m/(N + 1) should be used instead regardless of the underlying distribution (Makkonen 2006).

The procedure begins with running a ‘scree test’ to determine the number of significant orthogonal functions to explain some of the variances in the precipitation dataset. The eigenvalues are plotted against the eigenvalue number, called the scree plot, and the cumulative variance as a percentage of the total is plotted over each eigenvalue. The approximate 0.95 confidence limits are depicted for each eigenvalue using the rule of thumb of North et al. (1982), which ignores any autocorrelation in the data. No universal rule exists for determining the number of EOFs that should be retained for rotation (Hannachi et al. 2007). In practice, the number is chosen by requiring a minimum cumulative variance to obtain a marked break in the spectrum. Figure 4 illustrates a scree plot of the explained variances for the acquired PCs for the monthly series and allows us to visualize the variance breakdown. The variances accounting for monthly precipitation variability for the entire study period were 51.19, 8, and 7.41%, respectively, based on the three leading EOF modes. At approximately the fifth EOF, the graph starts to level off, and all subsequent EOFs start to contribute nearly the same amount of variance.
Figure 4

Scree plot of the explained variances for the acquired principal components for the monthly precipitation series.

Figure 4

Scree plot of the explained variances for the acquired principal components for the monthly precipitation series.

Close modal

The first EOF mode (EOF1) explains the highest monthly precipitation variance (51.19%). The first three EOFs sufficiently explain the spatial pattern, as they are mostly associated with large-scale atmospheric circulation that influences Turkey's weather patterns. Türkeş et al. (2009) argue that the PC1 generally designates the climatology of the precipitation totals in Turkey that is largely governed by the large-scale and/or synoptic-scale atmospheric features, such as surface and upper-air pressure and wind systems. They further claimed that in winter, the greater PC1 loadings over the western and southwestern parts of Turkey indicate the influence of large-scale atmospheric circulation and associated weather patterns, forming Mediterranean rainfall regimes. In their study, they attributed smaller PC1 loadings over the north-eastern and eastern parts of Turkey to the influence of the northerly mid-latitude cyclones and the northerly and easterly circulations linked with the high pressures developed over Eastern Europe and Siberia. The remaining EOFs were mostly associated with local influences (such as continentality and terrain features) that govern precipitation variability. This argument is supported by Tatli et al. (2004), who argued that the influences of local physical and geographic conditions are prevalent during summertime precipitation, as the polar and tropical air systems that are influential on mid-latitude precipitation during winter are not very dynamic in summer. In another study, Kadıoğlu (2000) performed a PC analysis to identify the main spatial patterns of Turkish precipitation and described the seasonal precipitation variability using the first four PCs, the first PC explained the influence of cyclonic activities on winter and autumn precipitation, while the second and third PCs substantiated the continentality and sea effects, respectively.

We retained the first three EOFs that explained approximately 67% of the total variance. The resulting EOF loadings were mapped to characterize the spatial distribution of the EOFs. EOF1 exhibited positive anomalies over the majority of Turkey, with the strongest variation over the central and eastern parts, and relatively weaker variation over the Mediterranean and Aegean coastal regions (Figure 5(a)). In general, EOF loadings increased from west to east and from south to north. The strongest positive variation was observed in the northern part of Eastern Anatolia. However, areas influenced by the Mediterranean climate, including the southern parts of the Marmara region and nearly the entire Aegean and Mediterranean regions, show negative variations.
Figure 5

The first empirical orthogonal function mode of the monthly precipitation: (a) spatial variability and (b) its corresponding principal component (PC1) over Turkey.

Figure 5

The first empirical orthogonal function mode of the monthly precipitation: (a) spatial variability and (b) its corresponding principal component (PC1) over Turkey.

Close modal

The spatial distribution of EOF loadings represents a wide range of atmospheric and physical geography influences on precipitation variability across Turkey. Although the EOF sign is causal, its determination is required to determine the atmospheric configurations or mechanisms that explain each EOF. In this regard, the EOF loadings were discussed because while large-scale and subsynoptic atmospheric circulation patterns account for the majority of the annual and seasonal variations in Turkish precipitation, smaller-scale processes, such as land-sea effects, continentality, and orography, interact to influence the regional variability of precipitation. The southward movement of the frontal depressions of polar origin and the Siberian anticyclone are responsible for the westerly and easterly north-easterly air advection into Turkey, respectively, especially in winter (Türkeş & Erlat 2009). In contrast, south-westerly and southerly air flows mainly originate from the winter Mediterranean frontal depressions (Tatli et al. 2004). As Turkey receives a considerable part of its precipitation in winter, the annual pattern reflects the wet season precipitation characteristics, with a spatially coherent pattern over a large area covering the western and southern regions. However, larger EOF loadings do not reflect the wet season precipitation characteristics over Turkey's north-eastern and eastern parts because of the influence of mid-latitude cyclones in a more northerly direction. As mentioned earlier, the wet season in Turkey ranges from October to May based on the monthly cycle of annual precipitation (Kömüşcü & Aksoy 2023). The main influence reflected in the EOF1 loadings over the western and southwestern parts is large-scale atmospheric circulation and associated frontal depressions, eventually bringing rainstorms into the Mediterranean region (Türkeş et al. 2009). The Mediterranean depressions leading to cyclones favour moist westerly wind advection into the country. Overall, the spatial pattern of EOF1 appeared to be largely shaped by the advancement of moist-unstable air masses from the Atlantic Ocean to the Mediterranean Basin. The areas where precipitation primarily results from Mediterranean cyclones are mainly represented by the Mediterranean climate. The precipitation patterns observed in western and southern Turkey cannot be explained by large-scale atmospheric patterns alone. At the subsynoptic scale, the Aegean Sea also contributes to the spatial and temporal precipitation patterns over the Mediterranean coastal parts of Turkey as a major source of winter and spring cyclones. Previous studies underscored the importance of this factor. However, the EOF1 loadings over the north-eastern and eastern parts can be explained by other influences: first, the northerly and easterly circulations triggered by polar front depressions and the Siberian anticyclone and second, the continentality effect. The largest positive loadings were observed around the high elevations of Eastern Anatolia but declined towards the southwestern part of the region as the elevation decreased. Furthermore, the increase in the EOF1 loadings from west to east can be linked to the influences associated with the continentality effect mounting from the central Anatolian Plateau towards the high elevations of Eastern Anatolia. EOF1 may not be significant in the continental inland and eastern parts of Turkey because precipitation is usually linked to local weather events such as convective occurrences and high terrain effects (Türkeş et al. 2009). Moreover, major cyclogenesis that develops over the Eastern Black Sea throughout the year, becoming particularly effective in July and August, contributes considerably to the precipitation climatology of the region (see also Trigo et al. 1999).

The first three leading EOF modes and their related time-variant PCs for the monthly series are presented in Figures 57. The PC time series of EOF1 revealed four wet years (1981, 1987, 1998, and 2009) with an amplitude greater than two standard deviations, and four dry years (1989, 1990, 2008, and 2013) with an amplitude of less than two standard deviations (Figure 5(b)).
Figure 6

The second empirical orthogonal function mode of the monthly precipitation: (a) spatial variability and (b) its corresponding principal component (PC2) over Turkey.

Figure 6

The second empirical orthogonal function mode of the monthly precipitation: (a) spatial variability and (b) its corresponding principal component (PC2) over Turkey.

Close modal
Figure 7

The third empirical orthogonal function mode of the monthly precipitation: (a) spatial variability and (b) its corresponding principal component (PC3) over Turkey.

Figure 7

The third empirical orthogonal function mode of the monthly precipitation: (a) spatial variability and (b) its corresponding principal component (PC3) over Turkey.

Close modal

The second and third EOF (EOF2 and EOF3) modes accounted for approximately 16% of the overall variability in monthly precipitation. EOF2 accounted for 8% of the total variance. EOF2 (Figure 6(a)) displayed positive anomalies in the majority of the country, and the variance generally decreased from north to south. The Eastern Black Sea coastal area has the highest variance, while most parts of South-eastern Anatolia are characterized by the lowest variance. The EOF2 loadings are characterized by a nearly zonal scattering pattern, with the highest and positive values discerned, particularly along the Eastern Black Sea coastal zone. To a certain extent, the EOF2 loadings represent the effects of the orographic influence of the Black Sea Mountains and Taurus Mountains in the north and south, respectively. These mountain ranges provide the necessary environment for orographically induced precipitation by elevating unstable Mediterranean air masses. The EOF2 loadings also portray the Black Sea coastal region and North-eastern Anatolia as separate zones, which are the wettest parts of Turkey. These regions are influenced by coastal orographic and continental convective precipitation, particularly during the summer.

EOF3 explained the lowest variance in monthly precipitation (7.41%). It shows positive anomalies only over parts of the Marmara and the coastal parts of the Aegean and Mediterranean regions (Figure 7(a)). The positive loadings increased from north to south towards the Mediterranean coastal band. In particular, the narrow zone in the western Mediterranean region had the highest variance with positive anomalies. Nearly two-thirds of Turkey, covering the central and eastern regions, and parts of the Black Sea region, had negative anomalies. Negative values peaked along the Eastern Black Sea coastal zone. The negative EOF3 values observed in South-eastern Anatolia refer to precipitation variations resulting from local convective instability due to different surface heating. The EOF3 pattern of Turkey's annual precipitation anomalies reflects the wet season precipitation regime characteristics and can be associated with the influence of the moist, unstable Mediterranean Sea and mid-latitude depressions over the Mediterranean coasts, whereas the orographic influence on the precipitation anomalies is more visible in the Eastern Black Sea coasts.

EOF analysis of annual data, which includes monthly precipitation, incorporates the influence of a wide range of factors. To specify and determine the seasonal influences, we performed EOF analyses during the wet and dry seasons. Based on the annual cycle of the monthly average precipitation, October–May was classified as the wet season, whereas June–September was classified as the dry season. Table 1 illustrates the first five leading EOF modes on a monthly basis during the wet and dry seasons. While the first five EOF modes account for over 70% of the variations in monthly and wet season precipitation, this figure increases to nearly 58% for variations in dry season precipitation.

Table 1

EOF modes with their percentiles on a monthly basis and for wet and dry seasons

EOFsMonthly (%)Wet season (%)Dry season (%)
51.19 44.1 25.63 
8.0 9.36 15.78 
7.41 8.55 8.08 
4.86 5.78 4.53 
2.97 3.31 3.21 
Total 74.43 71.1 57.23 
EOFsMonthly (%)Wet season (%)Dry season (%)
51.19 44.1 25.63 
8.0 9.36 15.78 
7.41 8.55 8.08 
4.86 5.78 4.53 
2.97 3.31 3.21 
Total 74.43 71.1 57.23 

Figure 8 illustrates the scree plots of the explained variances for the acquired EOFs for the wet and dry seasons. Capturing the 62% variability with only three components, the wet season is characterized by a higher spatial coherence than the dry season.
Figure 8

Scree plot of the explained variances for the acquired principal components for the (left) wet and (right) dry seasons' precipitation series, respectively.

Figure 8

Scree plot of the explained variances for the acquired principal components for the (left) wet and (right) dry seasons' precipitation series, respectively.

Close modal
Figure 9 shows the maps resulting from the EOF analysis, which were used to investigate the prevailing modes of spatial distribution of seasonal precipitation. The left and right columns display the spatial distribution of EOF loadings for the wet and dry seasons, respectively.
Figure 9

Spatial variability of the first empirical orthogonal function mode for the (a) wet season and (b) dry season; spatial variability of the second empirical orthogonal function mode for the (c) wet season and (d) dry season; spatial variability of the third empirical orthogonal function mode for the (e) wet season and (f) dry season.

Figure 9

Spatial variability of the first empirical orthogonal function mode for the (a) wet season and (b) dry season; spatial variability of the second empirical orthogonal function mode for the (c) wet season and (d) dry season; spatial variability of the third empirical orthogonal function mode for the (e) wet season and (f) dry season.

Close modal

For the wet season, the EOF1 loadings accounted for 44% of the total variance. A comparison of Figures 6(a) and 9(a) shows that the annual and wet season EOF1 loadings exhibit a highly similar spatial pattern, indicating the dominance of the wet season precipitation variability over the annual precipitation variability in Turkey (see also Unal et al. 2012). The spatial pattern of the first two EOFs in the wet season, which represented nearly 55% of the variability, closely resembled that of the loadings of the annual EOFs. As the precipitation climatology of the wet season controls the annual pattern, the higher EOF1 loadings over the western and southwestern parts of Turkey indicate the influence of large-scale atmospheric circulations originating from the northeast Atlantic Ocean and cyclogenesis areas in the Mediterranean Sea on the weather patterns (i.e., the Gulf of Genoa). These parts of Turkey experience direct influences of westerly and north-westerly air flows advecting from the northeast Atlantic Ocean and the Mediterranean Sea during winter and early spring. The highest EOFs with negative signs were found along the western and southern coastal parts of the country and were mainly affected by the Mediterranean precipitation regimes (Figure 9(a)). Turkey receives the majority of its precipitation in winter; thus, the observed patterns of winter precipitation reveal spatial coherency over a large domain that occupies nearly two-thirds of the country, particularly covering the western, central, and eastern regions. However, similar to the annual pattern, the smaller EOF loadings over the north-eastern and eastern parts of Turkey may be linked to the influence of northerly advection associated with mid-latitude cyclones (frontal lows) steering from Eastern Europe and Siberia (see also Türkeş et al. 2009). EOF1 did not seem to be significant in the inner and eastern parts of the country because of local orographic- and convective-type precipitation occurrences.

The eigenvalue of EOF2 explained 9.36% of the variance in the wet season precipitation anomalies (Figure 9(c)). The EOF2 map shows the dominance of negative and positive values in western and eastern Turkey, respectively. Negative values peaked along the Aegean and Mediterranean coastal regions. The spatial pattern of EOF2 explains two separate influences on the wet season precipitation anomalies. The first is the influence of unstable and moist Mediterranean air masses advecting from the west, and the second is the increasing continentality effect prevailing from the interior parts of the country towards the high elevations of Eastern Anatolia. The eigenvalue of EOF3 explained only 6.2% of the variance in wet season precipitation. In EOF3, the presence of a narrow zone along the Black Sea coast is distinguished, indicating the influence of northerly air advection and orography on the wet season precipitation over the region (Figure 9(e)). The influence of the Mediterranean frontal systems over the coastal Aegean region and, in particular, a narrow zone covering the Antalya and Muğla districts, is also apparent in EOF3.

Regarding dry season precipitation anomalies, the eigenvalue of EOF1 accounted for 25.63% of the year-to-year variability, whereas the combination of EOF2 and EOF3 accounted for approximately 24% of the total dry season precipitation variance (Table 1). Thus, the first three EOFs accounted for approximately 50% of the total variance in dry season precipitation variations. It can be argued that the dry season precipitation variability is relatively less than the annual and wet season precipitation variability because the wintertime precipitation formation mechanisms are more dynamic and diverse. The main atmospheric circulation pattern during summer is governed by the Azores high and persistent Asiatic monsoon low, creating northerly and north-westerly flows. During summer, Turkey receives relatively lower precipitation and is drier than normal, except for the Black Sea coastal region. Maritime polar air masses still affect parts of Turkey during the summer months with less frequency, and humid air masses towards the Black Sea coast (Koçman 1993). In the EOF1 spatial distribution map, the coastal Black Sea region appeared as a distinct zone, reflecting its wetter nature, as precipitation in this region was almost evenly distributed throughout the year (Figure 9(b)). This explains the orographically induced precipitation over the region, caused by north-westerly flows emerging from the Atlantic Ocean and maritime polar air masses. Additionally, the negative signs dominating the majority of the Black Sea coast explain the effects of the regional or local convective instability of the weather in summer. Positive sign loadings in EOF1 that are evident in the South-eastern Anatolia region and the Mediterranean coasts explain the influence of the extension of the monsoon low during the dry season when drier-than-normal conditions prevail. The positive loadings in Eastern Anatolia may be related to the northerly upper-air flow causing the occurrence of orographic and convective rainfall over the high terrain of the region. The eigenvalue of EOF2 explained approximately 16% of the variance in dry season precipitation. The EOF2 loadings showed the highest negative values over northeast Turkey (Figure 9(d)). The dry season precipitation variance over this region highlights the complexity of precipitation regimes and their seasonality. Atlantic-sourced moist air masses are advected to the north of Turkey by north-westerly air flows, which enhance precipitation formation under the influence of topography. Turkey's central and north-western parts had high loadings for EOF2. EOF1 does not seem to account for regions that are predominantly influenced by the maritime climate. The negative signs that prevail in most parts of Central Anatolia reflect the impact of continentality during the dry season. The positive EOF2 loadings observed in the north-western parts of Turkey largely explain the influence of the frontal Mediterranean depressions, which are still effective in the later parts of the dry season, covering autumn precipitation. EOF3 accounted for 8% of the total variance in the dry season precipitation. The EOF3 loadings revealed a roughly zonal pattern, with the highest and most positive values in the North-eastern Anatolia region (Figure 9(f)). Nearly two-thirds of Turkey, except for the Marmara region and the eastern parts of the Black Sea region, is characterized by a coherent positive sign, indicating the influence of continental local convection precipitation closely related to northerly advection and upper atmospheric disturbances. Türkeş et al. (2009) argue that this mechanism is strengthened by the surface warming of the highlands of Eastern Anatolia during the summer season.

In addition to the spatial variability analysis, the first three EOF (PC) loadings for the annual and seasonal data were also subjected to a trend analysis using the Mann–Kendall (MK) test to explore whether they presented monotonic increases or decreases over time. MK is a commonly preferred statistical test for analysing the trend characteristics of climatological and hydrological time series (Sang et al. 2014; Helsel et al. 2020; Kömüşcü & Aksoy 2023). The resulting z-statistics for these trends are summarized in Table 2.

Table 2

Trend analysis of the PC loadings time variations using the Mann–Kendall test

PCAnnual
z
Wet season
z
Dry Season
z
PC1 −0.88 Insignificantly negative −1.34 Insignificantly negative 1.82 Insignificantly positive 
PC2 −1.14 Insignificantly negative 2.26 Significantly positive −0.86 Insignificantly negative 
PC3 1.45 Insignificantly positive 0.83 Insignificantly positive −0.59 Insignificantly negative 
PCAnnual
z
Wet season
z
Dry Season
z
PC1 −0.88 Insignificantly negative −1.34 Insignificantly negative 1.82 Insignificantly positive 
PC2 −1.14 Insignificantly negative 2.26 Significantly positive −0.86 Insignificantly negative 
PC3 1.45 Insignificantly positive 0.83 Insignificantly positive −0.59 Insignificantly negative 

The PC1 loadings, which represent the primary mode for the annual and wet seasons, indicated insignificant negative trends, while the PC1 loadings for the dry season data revealed insignificantly positive trends. In particular, an increase in the amplitude of the PC1 loadings after 2010 indicated increased variability in the westerly and north-westerly circulations and cyclonic activities that bring much of the precipitation to the western and southern parts of the country. The decreasing trends support the fact that there may be a decrease in mid-latitude westerly circulation and Mediterranean cyclonic activities, eventually causing drier winters. This may also be associated with the positive phases of the NAO, which usually cause drier than normal conditions in Turkey. The trends detected in the PC1 loadings were also in good agreement with the annual, wet, and dry season precipitation monotonically increasing and decreasing, respectively (Kömüşcü & Aksoy 2023).

The linkage between prominent teleconnection patterns and Turkish precipitation variability

Understanding the relationship between teleconnection patterns and regional climate variability is essential, as seasonal and longer-timescale climate anomalies resulting from atmospheric disturbances associated with teleconnection patterns can directly impact human life. Teleconnection patterns are persistent large-scale pressure and circulation anomalies that cause disturbances in the climate and weather patterns across vast geographical areas worldwide (Baxter & Nigam 2015; Feldstein & Franzke 2017).

The variability in teleconnection patterns affects precipitation conditions over the Mediterranean region in several ways. Their major impact is on the regulation of moisture advection from the Atlantic Ocean to the Mediterranean region (Mariotti et al. 2002; Lionello et al. 2006). A large part of Southern Europe and Northern Mediterranean countries experience meteorological drought during the positive NAO index during the winter season (Marshall et al. 2001). Similarly, the precipitation conditions over Turkey are linked to the seasonal variability of certain prominent teleconnection patterns such as the NAO and AO. NAO influences the spatial and temporal variability of Turkish precipitation to a certain extent, leading to significant wet and dry periods (Türkeş & Erlat 2003, 2006). The positive phases of the NAO are marked by considerably cooler and drier conditions when the North Atlantic westerlies shift northward, resulting in a decrease in precipitation in Southern Europe and the Mediterranean region (Hurrell 1995; Cullen & deMenocal 1999). The variability of winter precipitation, especially in the Mediterranean, Aegean, and Central Anatolia regions, is linked to the variability of the NAO (Türkeş & Erlat 2003, 2006). Kadioğlu et al. (1999) revealed distinct effects of a high El Niño-Southern Oscillation (ENSO) index on the month-to-month variability of Turkish precipitation. Regarding the seasonal impacts of El Niño variability, wetter-than-normal conditions prevail during the El Niño fall season over the eastern parts of Turkey, whereas drier-than-normal conditions are prevalent during the spring season in the following El Niño year (Kiladis & Diaz 1989). The link between AO Index (AOI) and Turkish precipitation has been revealed by several studies over the last two decades (Gong et al. 2001; Choi & Byun 2010). A significant correlation was found between the negative phase of the AO and above-normal precipitation, particularly winter precipitation in the western regions of Turkey (Sezen & Partal 2019). They argued that during the negative AO phase, depressions originating from the central Atlantic advect more moisture in the Mediterranean region. Sezen & Partal (2020b) found a negative correlation between the Mediterranean Oscillation Index (MOI) and summer precipitation across Turkey, particularly in the Marmara region, non-coastal parts of the Black Sea region, and Central Anatolia. In the literature, there are almost no studies that relate the Atlantic multidecadal oscillation (AMO) to Turkish precipitation variability. In this sense, this study is one of the earliest to explore the subject even at a fundamental level. None of the earlier studies tried to identify the impacts of the AMO on the geographic domain of Turkey's climate. The AMO is quantified as fluctuations in the North Atlantic sea surface temperature (SST) anomalies between 0° and 70 °N latitudes and has cyclic oscillations varying between positive and negative phases (Trenberth & Shea 2006). It is linked to climate variation across many parts of the globe. During a positive AMO, much of Asia and Northern Europe is likely to face wetter-than-normal conditions, and much of Southern Europe, especially the Iberian Peninsula, is likely to be drier than normal (O'Reilly et al. 2017; Börgel et al. 2020). However, parts of Central Europe and the Middle East were likely to be drier than normal during the negative phase of the AMO.

Analysis of the spatial distribution of the correlation between monthly precipitation anomalies and teleconnection indices indicated that while AO and NAO were negatively correlated in Turkey's western and central regions, the MOI was positively correlated along the Aegean coastal areas and parts of the Marmara region (Figure 10). The MOI also correlates negatively in the coastal parts of the Black Sea and Eastern Anatolia regions. For clarification, the MOI calculated using sea level pressure (SLP) between the regions of Gibraltar and Lod (Israel) was used as described by Palutikof (2003). Except for some parts of the South-eastern Anatolia region, the Southern Oscillation Index (SOI) does not exhibit a widespread correlation with annual precipitation. Interestingly, the AMO index has the largest spatial coverage of positive correlation with the monthly precipitation anomalies, ranging from 0.1 in Eastern and Central Anatolia to 0.5 in coastal parts.
Figure 10

Spatial distribution of Pearson's r values and significance of the relationship between the monthly precipitation anomalies and the teleconnection indices.

Figure 10

Spatial distribution of Pearson's r values and significance of the relationship between the monthly precipitation anomalies and the teleconnection indices.

Close modal

Concerning the significance of the relationship between the monthly precipitation anomalies and teleconnection indices, AO was significantly correlated with the monthly precipitation anomalies in the western parts of Turkey, especially in coastal areas (Figure 10). The MOI has a significantly positive correlation along a narrow coastal zone in the Aegean Sea region and a significantly negative correlation along the central and Eastern Black Sea coastal regions. NAO indicates a significantly negative correlation at a few stations in the transition zone between Central Anatolia and the Black Sea regions. Except for the Black Sea coastal region and south-eastern parts of the Eastern Anatolia region, NAO indicates a negative correlation with the monthly precipitation series and is not significant. The AMO index indicated the largest spatial coverage, with an insignificant positive correlation with monthly precipitation anomalies. Noticeably, the SOI did not have a significant correlation with any of the 213 stations.

The first three EOFs retrieved from the EOF analysis were correlated with each teleconnection pattern for the annual precipitation anomalies using the Pearson correlation coefficient (Table 3). EOF1 had the highest correlation (r = −0.53 and r = −0.55) with the annual precipitation anomalies for the AO and NAO indices, respectively (Table 3). For both the indices, the correlation was negative. The relatively high correlation between the NAO and AO indices and EOF1 suggests the likely influence of large-scale atmospheric circulation patterns on the annual climate in Turkey. Among the four indices examined in this study, the AO index appears to be the most correlated with precipitation fields over Turkey. In particular, the variability of the annual precipitation anomalies observed in the western parts of Turkey and the Mediterranean coast was linked to the variability of the AO index. The MOI does not correlate well with Turkish precipitation anomalies, unlike other parts of the Eastern Mediterranean. Törnros (2013) found a significant correlation between winter precipitation and positive MOI phases for the 1960–1993 period and argued that the influence of the MOI may be more limited to the Southern Levant part of the Eastern Mediterranean region.

Table 3

Pearson's correlation between the first three EOFs and the five main teleconnection indices

IndexAnnual
EOF1EOF2EOF3
AMO 0.31 −0.35 −0.1 
AOI −0.53 0.19 0.06 
MOI 0.11 0.03 0.29 
NAOI −0.36 0.29 0.12 
SOI −0.03 −0.24 −0.03 
IndexAnnual
EOF1EOF2EOF3
AMO 0.31 −0.35 −0.1 
AOI −0.53 0.19 0.06 
MOI 0.11 0.03 0.29 
NAOI −0.36 0.29 0.12 
SOI −0.03 −0.24 −0.03 

Finally, the time series of the first EOF was plotted against the annual NAO, AO, AMO, and MO indices, which had the most significant influence, to determine whether any temporal association existed between the first EOF series and each index (Figure 11(a)–11(d)). The positive phase of EOF1 corresponded well with the negative phase of the NAO and AO indices. However, the different phases of both the MOI and AMO do not present a consistent pattern with the EOF1 series. The negative phase of the AMO aligns well with the negative EOF1 in most years, except for the 2000–2012 period when the positive phase of the AMO corresponded to the negative EOF time series. For MO, the positive phase of the MO usually matched the positive EOF values until early 2000, and then the MOI turned into a negative phase, while the EOF series remained on the positive side in most of the years, with some exceptions.
Figure 11

Comparison of the first principal component (PC1) with the annual (a) NAO and (b) AO, (c) AMO, and (d) MO indices for the 1975–2021 period.

Figure 11

Comparison of the first principal component (PC1) with the annual (a) NAO and (b) AO, (c) AMO, and (d) MO indices for the 1975–2021 period.

Close modal

In general, a decreasing trend dominates both annual and wet season precipitation in Turkey during the 1975–2021 period (Kömüşcü & Aksoy 2023). The periods exhibiting a decrease in monthly and seasonal precipitation could have resulted from low-frequency fluctuations in large-scale circulations, particularly the NAO and AO. Amplification of the positive phase of NAO and AO and negative phases of AMO and MO, especially after 2005 (except for the 2009–2013 period), could be one of the causes of the below-normal precipitation conditions of the annual and wet season precipitation in Turkey.

This study re-evaluated the variability of annual and seasonal spatial patterns of Turkish precipitation using the REOF-based approach. The retrieved spatial patterns were described in association with possible large-scale, regional, and local governing factors including atmospheric circulation, teleconnection patterns, topography, and continentality. The key findings of this study are summarized as follows.

  • The first three EOFs accounted for approximately 67 and 62% of the total variance in the annual and wet season precipitation series, respectively. The first three EOFs of the dry season precipitation captured only 50% of the variance.

  • The spatial distribution of EOF loadings represents a diverse range of atmospheric and non-atmospheric influences on precipitation variability across Turkey. Spatially different atmospheric circulation mechanisms can be major drivers of variability in Turkish precipitation on annual and seasonal timescales.

  • The EOF1 loadings of the annual and wet season precipitation depicted identical spatial distribution patterns, indicating that the wet season precipitation variability exerts control over annual precipitation variability in Turkey. The wet season was characterized by a higher spatial coherence than the dry season. Thus, atmospheric circulation patterns that influence regional precipitation variations during the dry season require further attention to describe the lack of coherency in spatial precipitation fields.

  • For the annual series, EOF1 loadings generally exhibited an increasing pattern from west to east and from south to north. While areas characterized by the Mediterranean climate (e.g., western and southern coastal parts) exhibited negative variations, areas with more prevalent continental climates (e.g., central and eastern Anatolian parts) displayed positive variations. Westerly circulation, which advects moisture into the Mediterranean region from the Atlantic Ocean, is considered the major influencing factor reflected in the EOF1 pattern retrieved over the western and southwestern parts of Turkey.

  • Positive EOF2 loadings represented the orographic forcing effects of the Black Sea Mountains on the precipitation pattern, whereas the negative EOF3 values observed in South-eastern Anatolia implied the impact of local convective instability.

  • The spatial pattern of EOF2 for the wet season precipitation anomalies revealed the influence of Mediterranean air masses over the western and southern parts of Turkey and continentality in Eastern Anatolia. The eigenvalue of EOF3 is indicative of the presence of a coherent zone along the Black Sea coast, indicating the influence of northerly air advection and orography on wet season precipitation.

  • Dry season EOF loadings were more coherent in the coastal Black Sea region and appeared as separate zones, reflecting wetter features. Contradictorily, the South-eastern Anatolia region and the Mediterranean coasts of Turkey were influenced by the extension of the monsoon during the dry season, which caused drier-than-normal conditions. The highly negative values of EOF2 loadings over northeast Turkey reflect the complexity of the precipitation regime over the region where both convective and orography-induced frontal precipitation occur throughout the year.

  • The negative signs prevailing in most parts of the Central Anatolian region for EOF2 reflect the influence of regional and local convective instabilities during the dry season. The coherent positive sign of EOF3 dominated nearly two-thirds of the country, except for the Marmara and Eastern Black Sea regions, indicating the influence of continental local convective precipitation linked to northerly advection.

  • In addition to large-scale circulation patterns, non-seasonal teleconnection patterns can lead to variability in precipitation across different regions of the country. AO and NAO were negatively correlated with annual precipitation anomalies in the western and central parts of Turkey, whereas the MOI was positively correlated with the Aegean coastal areas and parts of the Marmara region. The SOI showed a very weak correlation with annual precipitation anomalies, except in parts of south-eastern Anatolia. Notably, SOI did not exhibit a significant correlation with any of the 213 stations. It is likely that the contribution of the SOI to the monthly-based annual precipitation variations is very weak or may be masked by cyclone activities. Nearly 90% of the stations revealed a positive correlation with the AMO index, but the correlation was insignificant. The MOI indicated a relatively high correlation, but in a negative direction, with monthly based annual precipitation in Eastern and South-eastern Anatolia.

  • It is our view that both the Atlantic Ocean and the Mediterranean Sea have a strong modulating influence on precipitation across Turkey. Therefore, the variability in precipitation in Turkey may be attributed to two main reasons. The first is the change in the frequency and number of Mediterranean cyclones that advect moist and unstable air into the Mediterranean region (Trigo 2006; Reale & Lionello 2013). The second reason may be associated with the variability in regional teleconnection patterns. The enhanced frequency of the positive/negative phase of the NAO may have caused drier/wetter conditions than normal, particularly over parts of the country where the Mediterranean climate is dominant.

  • Finally, it should be recognized that regional precipitation in Turkey is characterized by a complex inherent variability, and the presence of several modes of climate variability also have spatially varying influences across the country, making it difficult to recognize their influence over the whole domain very accurately.

The findings of our EOF analysis verified and supported the results of several previous studies on this subject, albeit with some differences resulting from the different numbers of stations, longer study periods, and different wet and dry season periods (Kadıoğlu 2000; Türkeş & Erlat 2008; Unal et al. 2012). According to Kadıoğlu (2000), EOF3 characterizes maritime influence in the early parts of the wet season, whereas the fourth component (EOF4) indicates only local variations such as the coastal configuration of the Black Sea. In contrast, we found that EOF3 mainly described local influences, topography, and continentality. Our findings are in close agreement with the interpretation of the EOF1 loadings made by Türkeş et al. (2009), who associated EOF1 with large-scale and/or synoptic-scale atmospheric features, especially over the western and southwestern parts of Turkey where large-scale atmospheric circulation and associated weather patterns are influential in winter. However, our findings do not support their argument regarding the decreasing trend in winter precipitation variability. Finally, our results agree with the findings of Unal et al. (2012), who conducted a rotated EOF analysis of wet/dry periods of Turkish precipitation and found that the annual precipitation variability was controlled mainly by the wet season because most of the significantly rotated EOFs for the annual and wet periods exhibited similar spatial patterns. Moreover, both our study and Unal et al. (2012) strongly agree that among the teleconnection indices included in the study, the AO index was the most correlated with precipitation patterns over Turkey.

Considering the irregularity and complex behaviour of Turkish precipitation, its spatiotemporal variability needs to be investigated, and the underlying causes of the shift in this variability should be explored. The EOF methodology presented in this study provides an objective approach for determining the dominant modes of precipitation variability in Turkey and for identifying the primary and secondary influences on these modes. We conclude that while large-scale and subsynoptic atmospheric circulation patterns account for the majority of the annual and seasonal variations in Turkish precipitation, smaller-scale processes, such as land-sea effects, continentality, and orography, interact to influence the regional variability of precipitation. This study shows that the spatial extent of precipitation variability changes seasonally, reflecting the meteorological processes that dominate its variability. However, it should be emphasized that a substantial part of the dry season precipitation variability over Turkey remains partially described as the first three EOF modes explain up to 50% of the total variability. We believe that understanding the mechanisms that govern the dry season precipitation variability is necessary for future studies. Therefore, the analysis of dry season precipitation variability at shorter timescales with higher temporal resolution data (e.g., reanalysis data) may be promising and potentially provide a new vision of the mechanisms driving regional precipitation variability across the country. Thus, the findings of this study imply that more attention should be paid to the further analysis of summertime precipitation variability in Turkey. As anomalously high precipitation events usually occur during the summer months in Turkey and substantially impact human activities, further studies on summertime precipitation variability are even more important.

Once dominant spatiotemporal modes of variability from the observed data are identified, it will be easier to characterize the behaviour of the relevant and persistent precipitation patterns for the future prediction of atmospheric influences, including global teleconnections, and facilitate their modelling. The rotational EOF-based analysis provided an objective approach for determining the dominant modes of precipitation for highly variable Turkish precipitation and may be useful for identifying future high- and low-frequency circulation impacts on these modes. Moreover, a better representation and understanding of coherent precipitation patterns can enhance the predictability of the future variability of precipitation over the country and ultimately increase the accuracy of seasonal precipitation forecasting in connection with regional and local circulation systems as well as teleconnection patterns. In this sense, the identified relationships between precipitation patterns and teleconnections can also help reduce uncertainties in the projections of future regional precipitation. Our findings on the seasonal spatial variability of precipitation patterns can provide essential information to decision-makers in their planning of water resources, agriculture, and hydropower production projects and activities. Moreover, identifying the dominant seasonal precipitation variability and associating them with the circulation patterns can also provide essential information to decision-makers for better management of hydrometeorological hazards, such as droughts and floods that the country frequently suffers from. Finally, the findings of this study can also be used for risk assessment of the sectoral implications of precipitation variability that might be induced in the future by climate change arguments. The high spatial variability of Turkish precipitation and the impacts of diverse sources of atmospheric and non-atmospheric factors that shape variability might pose specific challenges or uncertainties associated with projecting future precipitation variability in the context of climate change, particularly at regional scales, despite generally agreeing to the direction of such changes. Considering that projected changes in future precipitation extremes can exhibit substantial uncertainties among climate models and pose inevitable challenges to climate actions and adaptation planning, studies of this nature can help improve the prediction of regional precipitation variability at both annual and inter-annual scales by establishing linkages between the identified spatial patterns and driving processes.

Both authors have been personally and actively involved in substantial work leading to the paper and will take public responsibility for its content. Both authors contributed to the study's conception and design. Material preparation, data collection, and analysis were performed by M.A. and A. Ü. K. M. A. prepared the precipitation data, ran the QC tests, and developed the codes to run the REOF analysis. He also drafted most of the figures and tables. A. Ü. K. analysed and interpreted the outputs. The first draft of the manuscript was written by AÜK and both authors commented on previous versions of the manuscript. Both authors read and approved the final manuscript.

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

This material is the authors' own original work, which has not been previously published elsewhere. The paper reflects the authors' own research and analysis in a truthful and complete manner. The paper properly credits the meaningful contributions of co-authors.

Both authors give their consent for the publication of identifiable details, which can include photograph(s) and/or videos and/or case history and/or details within the text (figures and tables) to be published in the Journal of Water and Climate Change.

The precipitation datasets used during and/or analysed during the current study are not publicly available as they are the sole entity of the Turkish State Meteorological Service (TSMS). The data can only be used for research and other academic purposes and cannot be shared with third parties unless written permission is granted by the TSMS.

The R programming software package and libraries were used to develop the codes to analyse the precipitation data by the proposed EOF methods. Both authors confirm that the codes and algorithm used to generate results that are reported in the paper and central to its main claim can be available to editors and reviewers upon their request.

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.

Acar
Z.
&
Gönençgil
B.
2022
Investigation of extreme precipitation indices in Turkey
.
Theor. Appl. Climatol.
148
(
9
).
https://doi.org/10.1007/s00704-022-03971-3
.
Akçar
N.
,
Yavuz
V.
,
Ivy-Ochs
S.
,
Kubik
P. W.
,
Vardar
M.
&
Schlüchter
C.
2007
Paleoglacial records from Kavron Valley, NE Turkey: Field and cosmogenic exposure dating evidence
.
Quat. Int.
164
,
170
183
.
https://doi.org/10.1016/j.quaint.2006.12.020.
Alexandersson
H.
1986
A homogeneity test applied to precipitation data
.
J. Clim.
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
.
Proc. Environ. Sci.
32
,
264
274
.
https://doi.org/10.1016/j.proenv.2016.03.031.
Atalay
İ.
&
Mortan
K.
2008
Regional Geography of Turkey with Pictures and Maps
.
Inkılap Yayinevi
,
Istanbul
,
ISBN:975102644X
,
632
pp.
(in Turkish)
.
Ay
M.
2022
Trend of minimum monthly precipitation for the East Anatolia region in Turkey
.
Theor. Appl. Climatol.
148
,
603
615
.
https://doi.org/10.1007/s00704-022-03947-3.
Ay
M.
&
Kisi
O.
2015
Investigation of trend analysis of monthly total precipitation by an innovative method
.
Theor. Appl. Climatol.
120
,
617
629
.
https://doi.org/10.1007/s00704-014-1198-8.
Beck
H. E.
,
Zimmermann
N. E.
,
McVicar
T. R.
,
Vergopolan
N.
,
Berg
A.
&
Wood
E. F.
2018
Present and future Köppen-Geiger climate classification maps at 1-km resolution
.
Sci. Data
5
,
1
12
.
https://doi.org/10.1038/sdata.2018.214
.
Börgel
F.
,
Frauen
C.
,
Neumann
T.
&
Meier
H. M.
2020
The Atlantic multidecadal oscillation controls the impact of the north Atlantic oscillation on north European climate
.
Env. Res. Lett.
15
,
104025
.
https://doi.org/10.1088/1748-9326/aba925
.
Bruni
G.
,
Reinoso
R.
,
van de Giesen
N. C.
,
Clemens
F. H. L. R.
&
ten Veldhuis
J. A. E.
2015
On the sensitivity of urban hydrodynamic modelling to rainfall spatial and temporal resolution
.
Hydrol. Earth Syst. Sci.
19
,
691
709
.
https://doi.org/10.5194/hess-19-691-2015
.
Ceylan
A.
&
Kömüşcü
A. Ü.
2008
Meteorolojik karakterli doğal afetlerin uzun yıllar ve mevsimsel dağılımları. Long-term and seasonal distributions of meteorological disasters
.
İklim Değişikliği ve Çevre
1
,
1
10
(in Turkish)
.
Choi
K. S.
&
Byun
H. R.
2010
Possible relationship between western North Pacific tropical cyclone activity and Arctic oscillation
.
Theor. Appl. Climatol.
100261
100274
.
https://doi.org/10.1007/s00704-009-0187-9
.
Çiçek
İ.
1996
Thorthwaite metoduna göre Türkiye'de iklim tipleri. Climate types in Turkey according to Thornthwaite method
.
Coğrafya Araştırmaları Dergisi
12
,
33
71
(in Turkish)
.
Çiçek
İ.
&
Duman
N.
2015
Seasonal and annual precipitation trends in Turkey
.
Carpathian J. Earth Environ. Sci.
10
,
77
84
.
Cristiano
E.
,
ten Veldhuis
M. C.
&
van de Giesen
N.
2017
Spatial and temporal variability of rainfall and their effects on hydrological response in urban areas
.
Hydrol. Earth Syst. Sci.
21
,
3859
3878
.
https://doi.org/10.5194/hess-21-3859-2017
.
Cullen
H. M.
&
deMenocal
P. B.
1999
North Atlantic influence on Turkish climate and water supply
.
Int. J. Climatol.
20
,
853
863
.
Deniz
A.
,
Toros
H.
&
Incecik
S.
2011
Spatial variations of climate indices in Turkey
.
Int. J. Climatol.
31
,
394
403
.
https://doi.org/10.1002/joc.2081
.
Dommenget
D.
&
Latif
M.
2002
A cautionary note on the interpretation of EOFs
.
J. Clim.
15
(
2
),
216
225
.
Fahmi
F. M.
,
Kartal
E.
,
İyigun
C.
,
Yozgatlıgil
C.
,
Purutcuoğlu
V.
,
Batmaz
İ.
,
Türkeş
M.
,
Köksal
G.
,
2011
Determining the climate zones of Turkey by center-based clustering methods
. In:
Nonlinear and Complex Dynamics: Applications in Physical, Biological, and Financial Systems
(
Baleanu
D.
,
Tenreiro-Machado
J. A.
&
Luo
A. C.
, eds).
Springer New York
,
New York
, .pp.
171
177
.
ISBN: 978-1-4614-0231-2
.
Feldstein
S. B.
&
Franzke
C. L.
2017
Atmospheric teleconnection patterns
. In:
Nonlinear and Stochastic Climate Dynamics
(Franzke, C. L. E. & O'Kane, T. J. eds.). Cambridge University Press, Cambridge, pp.
54
104
.
https://doi.org/10.1017/9781316339251.004
.
Gönençgil
B.
2020
Evaluate Turkey's climate classification by clustering analysis method
. In:
Smart Geography
(Nedkov, S., ed.). Springer, Cham, pp.
41
53
.
https://doi.org/10.1007/978-3-030-28191-5_4
.
Gong
D. Y.
,
Wang
S. W.
&
Zhu
J. H.
2001
East Asian winter monsoon and Arctic oscillation
.
Geophys. Res. Lett.
8
,
2073
2076
.
https://doi.org/10.1029/2000gl012311
.
Hadi
S. J.
&
Tombul
M.
2018
Long-term spatiotemporal trend analysis of precipitation and temperature over Turkey
.
Meteorol. Appl.
25
,
445
455
.
https://doi.org/10.1002/met.1712
.
Hannachi
A.
2004
A Primer for EOF Analysis of Climate Data
.
Department of Meteorology, University of Reading
, Reading, pp.
1
29
.
Hannachi
A.
,
Jolliffe
I. T.
,
Stephenson
D. B.
&
Trendafilov
N.
2006
In search of simple structures in climate: Simplifying EOFs
.
Int. J. Climatol.
26
,
7
28
.
https://doi.org/10.1002/joc.1243
.
Hannachi
A.
,
Jolliffe
I. T.
&
Stephenson
D. B.
2007
Empirical orthogonal functions and related techniques in atmospheric science: A review
.
Int. J. Climatol.
27
,
1119
1152
.
https://doi.org/10.1002/joc.1499
.
Hannachi
A.
,
Finke
K.
&
Trendafilov
N.
2023
Common EOFs: A tool for multi-model comparison and evaluation
.
Clim. Dyn.
60
,
1689
1703
.
https://doi.org/10.1007/s00382-022-06409-8
.
Helsel
D. R.
,
Hirsch
R. M.
,
Ryberg
K. R.
,
Archfield
S. A.
&
Gilroy
E. J.
2020
Statistical Methods in Water Resources. Supporting Materials: U.S. Geological Survey data release. https://doi.org/10.5066/P9JWL6XR
.
Hurrell
J. W.
1995
Decadal trends in the North Atlantic Oscillation: Regional temperatures and precipitation
.
Science
269
,
676
679
.
https://doi.org/10.1126/science.269.5224.676
.
Kadioğlu
M.
2000
Regional variability of seasonal precipitation over Turkey
.
Int. J. Climatol.
20
,
1743
1760
.
http://doi.org/10.1002/1097-0088
.
Kadıoğlu
M.
,
Tulunay
Y.
&
Borhan
Y.
1999
Variability of Turkish precipitation compared to El Nino events
.
Geophys. Res. Lett.
26
,
1597
1600
.
https://doi.org/10.1029/1999gl900305
.
Kaiser, H. F. 1958 The varimax criterion for analytic rotation in factor analysis. Psychometrika 23, 187–200. https://doi.org/10.1007/BF02289233
Karaca
M.
,
Deniz
A.
&
Tayanç
M.
2000
Cyclone track variability over Turkey in association with regional climate
.
Int. J. Climatol.
20
,
1225
1236
.
https://doi.org/10.1002/1097-0088
.
Kiladis
G. N.
&
Diaz
H. F.
1989
Global climatic anomalies associated with extremes in the Southern Oscillation
.
J Clim.
2
,
1069
1090
.
https://doi.org/10.1175/1520-0442
.
Koçman
A.
1993
Climate of Turkey
.
Aegean University Press
,
İzmir, Turkey
.
(in Turkish)
.
Komuscu
A. U.
2001
An analysis of recent drought conditions in Turkey in relation to circulation patterns
.
Drought Netw. News
22
,
1994
2001
. https://digitalcommons.unl.edu/droughtnetnews/22.
Kömüşcü
A. U.
&
Aksoy
M.
2023
Long-term spatio-temporal trends and periodicities in monthly and seasonal precipitation in Turkey
.
Theor. Appl. Climatol.
51
,
1623
1649
.
https://doi.org/10.1007/s00704-022-04349-1
.
Kömüşcü
A. U.
,
Turgu
E.
&
DeLiberty
T.
2022
Dynamics of precipitation regions of Turkey: A clustering approach by K-means methodology in respect of climate variability
.
J Water Clim. Chang.
13
(
10
),
3578
3606
.
https://doi.org/10.2166/wcc.2022.186
.
Lian
T.
&
Chen
D.
2012
An evaluation of rotated EOF analysis and its application to tropical Pacific SST variability
.
J. Clim.
25
,
5361
5373
.
https://doi.org/10.1175/jcli-d-11-00663.1
.
Lionello
P.
,
Malanotte-Rizzoli
P.
,
Boscolo
R.
,
Alpert
P.
,
Artale
V.
,
Li
L.
,
Luterbacher
J.
,
May
W.
,
Trigo
R.
,
Tsimplis
M.
,
Ulbrich
U.
&
Xoplaki
E.
2006
The Mediterranean climate: An overview of the main characteristics and issues
.
Dev. Earth Environ. Sci.
4
,
1
26
.
https://doi.org/10.1016/S1571-9197(06)80003-0
.
Longobardi
A.
&
Boulariah
O.
2022
Long-term regional changes in inter-annual precipitation variability in the Campania Region, Southern Italy
.
Theor. Appl. Climatol.
148
,
869
879
.
https://doi.org/10.1007/s00704-022-03972-2
.
Makkonen
L.
2006
Plotting positions in extreme value analysis
.
J. Clim. Appl. Meteorol.
45
,
334
340
.
https://doi.org/10.1175/JAM2349.1
.
Mariotti
A.
,
Struglia
M. V.
,
Zeng
N.
&
Lau
K. M.
2002
The hydrological cycle in the Mediterranean region and implications for the water budget of the Mediterranean Sea
.
J. Clim.
15
,
1674
1690
.
https://doi.org/10.1175/1520-0442
.
Marshall
J.
,
Kushnir
Y.
,
Battisti
D.
,
Chang
P.
,
Czaja
A.
,
Dickson
R.
,
Hurrell
J.
,
McCartney
M.
,
Saravanan
R.
&
Visbeck
M.
2001
North Atlantic climate variability: Phenomena, impacts and mechanisms
.
Int. J. Climatol.
21
,
1863
1898
.
https://doi.org/10.1002/joc.693
.
Martinson, D. G. 2018 Empirical Orthogonal Function (EOF) Analysis. In: Quantitative Methods of Data Analysis for the Physical Sciences and Engineering (Martinson, D. G., ed.). Cambridge University Press, Cambridge, pp. 495–534.
Navarra
A.
&
Simoncini
V.
2010
A Guide to Empirical Orthogonal Functions for Climate Data Analysis
.
Springer Science & Business Media
,
Dordrecht.
NCAR
.
2013
NOAA Climate Prediction Center, Northern Hemisphere Teleconnection Patterns. Retrieved August 23, 2022, from www.cpc.ncep.noaa.gov/data/teledoc/telecontents.shtml
.
North
G.
,
Bell
T.
,
Cahalan
R.
&
Moeng
F.
1982
Sampling errors in the estimation of empirical orthogonal functions
.
Mon. Weather Rev.
110
,
699
706
.
O'Reilly
C. H.
,
Woollings
T.
&
Zanna
L.
2017
The dynamical influence of the Atlantic multidecadal oscillation on continental climate
.
J. Clim.
30
,
7213
7230
.
https://doi.org/10.1175/JCLI-D-16-0345.1
.
Öztürk
M. Z.
,
Çetinkaya
G.
&
Aydın
S.
2017
Köppen-Geiger İklim Sınıflandırmasına Göre Türkiye'nin İklim Tipleri. Climate types of Turkey according to Köppen-Geiger climate classification
.
J. Geogr.
35
,
17
27
.
https://doi.org/10.26650/jgeog295515 (in Turkish)
.
Palutikof
J.
,
2003
Analysis of Mediterranean climate data: Measured and modelled
. In:
Mediterranean Climate
(
Bolle
H. J.
, ed.).
Springer
,
Berlin, Heidelberg
, pp.
125
132
.
Partal
T.
2018
Wavelet based periodical analysis of the precipitation data of the Mediterranean region and its relation to atmospheric indices
.
Model. Earth Syst. Environ.
4
,
1309
1318
.
https://doi.org/10.1007/s40808-018-0505-2
.
Partal
T.
&
Kahya
E.
2006
Trend analysis in Turkish precipitation data
.
Hydrol. Processes
20
,
2011
2026
.
https://doi.org/10.1002/hyp.5993
.
Pettitt
A. N.
1979
A non-parametric approach to the change-point problem
.
J. R. Stat. Soc. Ser. C Appl. Stat.
28
,
126
135
.
https://doi.org/10.2307/2346729
.
Raja
N. B.
,
Aydin
O.
,
Türkoğlu
N.
&
Çiçek
İ.
2017
Space-time kriging of precipitation variability in Turkey for the period 1976–2010
.
Theor. Appl. Climatol.
129
,
293
304
.
https://doi.org/10.1007/s00704-016-1788-8
.
R Core Team
.
2019
R: A Language and Environment for Statistical Computing
.
R Foundation for Statistical Computing
,
Vienna, Austria
.
Reale
M.
&
Lionello
P.
2013
Synoptic climatology of winter intense precipitation events along the Mediterranean coasts
.
Nat. Hazards Earth Syst. Sci.
13
,
1707
1722
.
http://doi.org/10.5194/nhess-13-1707-2013
.
Roundy
P. E.
2015
On the interpretation of EOF analysis of ENSO, atmospheric Kelvin waves, and the MJO
.
J. Clim.
28
,
1148
1165
.
https://doi.org/10.1175/JCLI-D-14-00398.1
.
Sang
Y. F.
,
Wang
Z.
&
Liu
C.
2014
Comparison of the MK test and EMD method for trend identification in hydrological time series
.
J. Hydrol.
510
,
293
298
.
https://doi.org/10.1016/j.jhydrol.2013.12.039
.
Sariş
F.
,
Hannah
D. M.
&
Eastwood
W. J.
2010
Spatial variability of precipitation regimes over Turkey
.
Hydrol. Sci. J.
55
,
234
249
.
https://doi.org/10.1080/02626660903546142
.
Şen
Z.
2012
Innovative trend analysis methodology
.
J. Hydrol. Eng.
17
,
1042
1046
.
https://doi.org/10.1061/(asce)he.1943-5584.0000556
.
Şen
Z.
2014
Trend identification simulation and application
.
J. Hydrol. Eng.
19
,
635
642
.
https://doi.org/10.1061/(asce)he.1943-5584.0000811
.
Sezen
C.
&
Partal
T.
2019
The impacts of Arctic oscillation and the North Sea Caspian pattern on the temperature and precipitation regime in Turkey
.
Meteorol. Atmos. Phys.
131
,
1677
1696
.
https://doi.org/10.1007/s00703-019-00665-w
.
Sezen
C.
&
Partal
T.
2020a
Wavelet combined innovative trend analysis for precipitation data in the Euphrates-Tigris basin, Turkey
.
Hydrol. Sci. J.
65
,
1909
1927
.
https://doi.org/10.1080/02626667.2020.1784422
.
Sezen
C.
&
Partal
T.
2020b
The effects of Mediterranean oscillation on temperature and precipitation data in Turkey
.
J. Water Clim. Chang.
11
,
722
743
.
https://doi.org/10.2166/wcc.2019.192
.
Sönmez
İ.
2013
Quality control tests for western Turkey Mesonet
.
Meteorol. Appl.
20
,
330
337
.
https://doi.org/10.1002/met.1286
.
Tatli
H.
,
Dalfes
N. H.
&
Menteş
S. Ş.
2004
A statistical downscaling method for monthly total precipitation over Turkey
.
Int. J. Climatol.
24
,
161
180
.
https://doi.org/10.1002/joc.997
.
Tayanç
M.
,
Karaca
M.
&
Yenigün
O.
1997
Annual and seasonal air temperature trend patterns of climate change and urbanization effects in relation to air pollutants in Turkey
.
J. Geophys. Res. Atmos.
102
,
1909
1919
.
https://doi.org/10.1029/96jd02108
.
Topuz
M.
,
Feidas
H.
&
Karabulut
M.
2021
Trend analysis of precipitation data in Turkey and relations to atmospheric circulation: (1955–2013)
.
Ital. J. Agrometeorol.
2
,
91
107
.
https://doi.org/10.13128/ijam-887
.
Toros
H.
1993
Klimatolojik serilerden Türkiye ikliminde trend analizi. Trend Analysis in Turkish Climate from Climatological Series
.
Master's Thesis
,
Istanbul Technical University
, pp.
30
100
(in Turkish)
.
Trenberth
K. E.
&
Shea
D. J.
2006
Atlantic hurricanes and natural variability in 2005
.
Geophys. Res. Lett.
33
.
https://doi.org/10.1029/2006GL026894
.
Trigo
I. F.
,
Davies
T. D.
&
Bigg
G. R.
1999
Objective climatology of cyclones in the Mediterranean region
.
J. Clim.
12
,
1685
1696
.
https://doi.org/10.1175/1520-0442
.
Türkeş
M.
1996
Spatial and temporal analysis of annual rainfall variations in Turkey
.
Int. J. Climatol.
16
,
1057
1076
.
https://doi.org/10.1002/(sıcı)1097-0088
.
Türkeş
M.
2020
Climate and drought in Turkey
. In:
World Water Resources
,
Vol. 2
(
Harmancioglu
N.
&
Altinbilek
D.
, eds).
Springer
,
Cham
.
https://doi.org/10.1007/978-3-030-11729-0_4
.
Türkeş
M.
&
Erlat
E.
2003
Precipitation changes and variability in Turkey linked to the North Atlantic Oscillation during the period 1930–2000
.
Int. J. Climatol.
23
,
1771
1796
.
https://doi.org/10.1002/joc.962
.
Türkeş
M.
&
Erlat
E.
2006
Influences of the North Atlantic oscillation on precipitation variability and change in Turkey
.
Il Nuovo Cimento
29
,
117
135
.
https://doi.org/10.1393/ncc/i2005-10228-8
.
Türkeş
M.
&
Erlat
E.
2008
Influence of the Arctic Oscillation on the variability of winter mean temperatures in Turkey
.
Theor. Appl. Climatol.
92
,
75
85
.
https://doi.org/10.1007/s00704-007-0310-8
.
Türkeş, M. & Erlat, E. 2009 Winter mean temperature variability in Turkey associated with the North Atlantic Oscillation. Meteorol. Atmos. Phys. 105, 211–225. https://doi.org/10.1007/s00703-009-0046-3
Türkeş
M.
,
Koç
T.
&
Sariş
F.
2009
Spatiotemporal variability of precipitation total series over Turkey
.
Int. J. Climatol.
29
,
1056
1074
.
https://doi.org/10.1002/joc.1768
.
Unal
Y. S.
,
Deniz
A.
,
Toros
H.
&
Incecik
S.
2012
Temporal and spatial patterns of precipitation variability for annual, wet, and dry seasons in Turkey
.
Int. J. Climatol.
32
,
392
405
.
https://doi.org/10.1002/joc.2274
.
Wilks
D. S.
2011
Statistical Methods in Atmospheric Sciences
, 3rd edn.
Academic Press
,
Oxford
, pp.
704
.
ISBN: 9780123850225
.
Yavuz
H.
&
Erdoğan
S.
2012
Spatial analysis of monthly and annual precipitation trends in Turkey
.
Water Resour. Manag.
26
,
609
621
.
https://doi.org/10.1007/s11269-011-9935-6
.
Yılmaz
E.
&
Çiçek
İ.
2016
Thornthwaite climate classification of Turkey
.
J. Hum. Sci.
13
(
3
),
3973
3994
. .
Yılmaz
E.
,
Akdi
Y.
,
Uğurca
E.
,
Çiçek
İ.
&
Atakan
C.
2021
Precipitation cycles in Turkey
.
Theor. Appl. Climatol.
143
,
1299
1314
.
https://doi.org/10.1007/s00704-020-03490-z
.
Web Sources:
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).