Climate

Abstract

Climate plays a crucial role in the development of soils on the surface of the earth. In fact, there is a broad agreement between the distributions of soil and climate types at a global scale.

1 Introduction

Climate plays a crucial role in the development of soils on the surface of the earth. In fact, there is a broad agreement between the distributions of soil and climate types at a global scale. The direct impact of climate on the soil formation is through weathering caused by temperature and precipitation. Both temperature and precipitation affect the rates of the physical, chemical, and biological processes in the soil. Warmer and moister climates favor the weathering, leading to a richer and deeper soil layer. Colder and drier climates slow down the soil formation process, resulting in a relatively poor shallow soil layer. Leaching is enhanced in rainy climates, because of increased percolation rates. Climate also affects the soil formation indirectly via its influence on plant cover. Humid and warmer climates favor the plant growth and microbial activity in the soil, and the balance between them determines the organic matter content of the soil. Climate change is expected to make alterations in temperature and precipitation. Temperatures are projected to increase worldwide, however precipitation projections show increases and decreases for different regions. Moreover, not only its annually averaged values, but also its characteristics such as seasonality, intensity, duration, and frequency are projected to change. Such changes may have profound impacts on soil development processes in some areas, as they are highly effective in weathering. Soil formation rates could also be impacted by the northward and upward (to higher elevation) shifts in the biomes that are expected as a result of temperature rise.

Turkey lies in a transition zone between arid and temperate climates, comprising primarily Mediterranean and continental climate types. It has a mountainous topography, prone to significant erosion rates. The IPCC’s (Intergovernmental Panel on Climate Change) Fifth Assessment Report (IPCC 2013) indicates that Turkey is located in a region that is highly vulnerable to climate change impacts. A northward shift in the storm (of Atlantic origin) tracks is expected to result in less precipitation for the southern parts of Turkey while more precipitation for the northern parts. Such changes, together with temperature increases, will certainly have important implications for the soils of Turkey. It is therefore important to, first, understand the climate of Turkey and the possible changes it could undergo. This chapter, thus, introduces what is known about the present and future climates of Turkey. The rest of this chapter is organized as follows: Next section introduces the present climate of the Turkey and its Köppen–Geiger classification. Then, the future climate of the region is given by the evaluation of CMIP3 and CMIP5 projections for future climate of Turkey. It is followed by a section that introduces CMIP3-based high-resolution projections for Turkey. This section also includes a discussion on the impacts of climate change from soil moisture and drought perspectives.

2 Present Climate of Turkey

The Anatolian peninsula is mainly characterized by subtropical climate, similar to other regions in the Mediterranean, except for minor differences arising from the fact that the region is surrounded by water bodies, the Black Sea in the north, the Aegean Sea in the west and the Mediterranean Sea in the south. Due to its unique location and complex topography, the different subregions or basins have varying climatic characteristics. The southern and western coastal areas have Mediterranean climate , which is mainly associated with the seasonal migrations of the mostly dynamic-originated pressure systems and the alternating tropical and polar air masses between summer and winter, and the dry summer subtropical Mediterranean climates can be found along the west coasts between about 25° and 40° latitudes as a natural part of the mid-latitude mild climates (Türkeş 2010). The region is mostly dominated in summer by dry, stable and subsiding air from the eastern portions of dynamically originated subtropical highs (e.g., Azores high), which are also associated with the sinking branch of the Hadley cell circulation. In winter, the wind and pressure systems shift equator-ward as a result of the equator-ward migration of the Rossby waves and associated polar jet streams and upper air westerlies with the polar front, and the Mediterranean climate regions are influenced by the westerlies with their trailing middle-latitude frontal cyclones (Türkeş 2010).

Besides having different climate types distributed across the country, Turkey is considered as the largest Mediterranean climate region found in the Mediterranean Basin, which can be thus called as the ‘true’ Mediterranean macroclimate. In the Atlantic-ward of the southern Europe and the Mediterranean Sea Basin, there are no north–south-oriented mountain chains and the Mediterranean climates penetrate a great distance from the west basin of the Mediterranean Sea with some parts of the Iberia Peninsula to the East Basin of the Mediterranean Sea and the Mediterranean coastal and inland regions of the Middle East region. Although the term Mediterranean is used to describe the climate synonymous with the region surrounding the Mediterranean Sea, it is not exclusive to this region. The Mediterranean climates are also found, for example, in the northern Iran, California, Chile, Australia, New Zealand and South Africa, in addition to the ‘true’ Mediterranean regions of Portugal, Spain, Coastal North-west Africa (Morocco, Tunisia, Algeria), France, Italy, Greece, Turkey, Lebanon and Israel (Türkeş 2010; Türkeş et al. 2011).

In addition, seasonality is the most dominant and distinctive character and factor of the Mediterranean climates. The Mediterranean climate tends to alternate wet and dry seasons, because it is located in the transitional zone between the dry west coast tropical desert mainly related to the subtropical high pressure systems and the descending segment of the tropical Hadley circulation cell, and the wet west coast climate mainly related with the polar front and associated mid-latitude cyclones.

As we shortly discussed above on the mid-latitude climates, there are many atmospheric features that can be considered as dominating the Mediterranean climates. For instance, the Rossby waves, which are formed by the upper air troughs and lows, and the upper air ridges and highs, controlled the penetration of polar air masses (continental polar—cP, maritime polar—mP, and very rarely continental Arctic—cA) towards equator in certain months of the year, and tropical air masses (continental tropical—cT, and maritime tropical—mT) towards poles in other certain months of the year. The seasonal movements arise from mainly the Sun’s seasonal migration (i.e., Sun’s apparent movement) and thus amount and intensity of the Sun’s radiation result an energy exchange between the poles and the equatorial belt (Türkeş 2010). On the other hand, due to the seasonal movements of the inter-tropical convergence zone (ITCZ) associated closely with the movements of the Sun, the Rossby waves would be closer to the equatorial belt in winter than that in summer. Thus, polar originated weather systems can more strongly fluctuate further equator-ward in winter than in summer. This movement is also responsible for enhancing the seasonal energy contrasts globally particularly over the continents. In the mid-latitudes, while the planetary-scale Rossby waves and the upper air jet streams governing also westerlies and the frontal cyclones are the most powerful features of these climates, these atmospheric controls are also themselves controlled by topography, continentally, land–sea distribution and interactions, air masses and their thermodynamic and mechanical modifications arising from the physical geographical features of the Earth surfaces (Türkeş 2010). Consequently, there is no simple explanation of the mid-latitude climates including the Mediterranean climate as a whole.

An analysis of cyclone paths over the Anatolian Peninsula performed by Karaca et al. (2000) using ECMWF data shows a set of dominant cyclone trajectories: (1) the path extending from northern Turkey to the southwest of Russia (and Balkans), affecting Marmara and the Black Sea region , (2) the path starting in the Genoa Gulf and passing over the Aegean Sea, affecting all of Turkey and (3) the path which originates in western or central Mediterranean (and in some cases north of the Sahara), affecting southern Turkey. Almost all precipitation comes from the frontal cyclones, except for the late spring and early summer convective instability showers and thunderstorms in inland Mediterranean climates of the Anatolia Peninsula and the Middle East region. On the other hand, west-to-east oriented Alp mountains of the South Europe and the North Anatolia and Taurus mountains of Turkey create stronger influence on the westerlies and associated mid-latitude frontal cyclones by means of the orographic lifting of the moist air masses resulting adiabatic cooling and condensation, and occurrence of the precipitation over the mountains, particularly by the time of that faced to southerly and westerly moist air masses (Türkeş 1996, 1998, 2010).

Based on all these explanations above, the climate of the region have following five distinctive characteristics (Türkeş 2010): (1) about half of the modest annual precipitation amount falls in winter, whereas summers are mostly virtually rainless, (2) winter temperatures are unusually mild for the middle-latitudes except some eastern and inland regions, summer air temperatures vary from hot to warm, (3) cloudless skies and intensive sunshine (shortwave solar radiation) are typical particularly in summer months, (4) the seas (Mediterranean and Black Seas) have major influences on the climate of the region and land distribution and the interactions between sea and lands, in addition to the ocean–atmosphere interaction (Bozkurt and Sen 2011, Turunçoğlu 2015), during the year particularly in the ‘true’ or ‘actual’ Mediterranean macroclimate region, and (5) The major characteristic of the Mediterranean climate is of high temporal variability varying from seasonal and inter-annual to centennial scales due to following factors (Türkeş 2010):

• i. It extends in a transition region between temperate and cold mid-latitudes and tropics (i.e., subtropical zone)

• ii. It has been facing significant circulation (associated pressure and wind systems characterizing mid-latitude and tropical/monsoonal weather and climate, respectively) changes between winter and summer

• iii. It is closely associated with several atmospheric oscillation and/or teleconnection patterns such as North Atlantic Oscillation (NAO), Arctic Oscillation (AO), Mediterranean Oscillation (MO), El Niño-Southern Oscillation (ENSO), and North Sea Caspian Pattern (NCP), etc. (Kutiel and Türkeş 2005; Tatlı 2007; Türkeş and Erlat 2003, 2005, 2008, 2009; Erlat and Türkeş 2012).

2.1 The Climate Classification of the Region

The idea of forming climate classes with the combination of long-term temperature and precipitation characteristics, along with setting limits and boundaries fitted to prevailing vegetation and soil distributions, was first suggested in 1918 by Wladimir Köppen. This classification was subsequently revised and improved by Köppen himself and his students, and it becomes the most widely used tool of climatic classifications for the physical geographical and environmental (climatological, biogeographical, ecological, etc.,) purposes. We have accepted here the criteria that follow Köppen’s last publication about his classification system in the Köppen-Geiger Handbook (Köppen 1936), with the exception of the boundary between the temperate (C) and cold (D) climates, because we have used the Köppen–Geiger climate data computed by Peel et al. (2007) who also followed Russell (1931) and used the temperature of the coldest month >0 °C, rather than >−3 °C as used by Köppen in defining the temperate—cold climate boundary (Wilcock 1968; Essenwanger 2001). The Köppen climate system carries out a shorthand code of letters designating major climate groups, subgroups within the major groups, and also subdivisions to distinguish particular seasonal characteristics of temperature and precipitation. For instance, Table 1 indicates two-letter group climates of the Köppen–Geiger classification.

Table 1 Two-letter group climates of the Köppen–Geiger classification. The red solid boxes show climate types that are found in Turkey

To determine the Köppen–Geiger climate classifications for Turkey, the long-term average monthly precipitation total and monthly mean temperature data were calculated by using time series of the station-based climate data of Turkey (Türkeş 1996, 1999; Türkeş et al. 2002b). Detailed information for meta-data and homogeneity analyses applied to long-term precipitation and temperature series of Turkey can be found in Türkeş (1996, 1999) and Türkeş et al. (2009) and Türkeş et al. (2002a), respectively.

The climate of Turkey according to the Köppen–Geiger climate system is very diverse (Fig. 1), as in many other climate classifications. Only the first and second-hand letters in the Köppen–Geiger climate classification system, following the major climate types can be separated (Türkeş 2010):

Fig. 1

Geographical distribution of climate types in Turkey based on the first-, second-, and third-hand letters classification of the Köppen–Geiger climate system (Türkeş 2010)

• (1) Subtropical steppe climate BS (mostly BSk—cold steppe) is found in the mid-part of the continental Central Anatolia region and the Van-Iğdır district over most eastern part of the continental Eastern Anatolia region.

• (2) Temperate rainy or humid temperate west coast climate without dry season Cf (mostly Cfa and Cfb—humid mesothermal) is dominant in the Black Sea coastal region of Turkey with the exception of the western subregion.

• (3) The Marmara, Aegean, Mediterranean and southeastern Anatolia regions, and the western and southern parts of the continental central Anatolia region belong to the dry summer subtropical Mediterranean climate or temperate rainy climate with dry summer (humid mesothermal) Cs (mostly Csa). Csa classification indicates that summers are hot with midsummer monthly averages between 24 and 29 °C and high maximums above 38 °C. Average cold-month temperatures are about 10 °C with occasional minimums below freezing temperatures. Yearly soil moisture deficiency is not characteristic in Group C climates in general, whereas seasonal soil moisture deficiency particularly in summer months is evident in the Mediterranean Csa and Csb climates due to changes in the hemispheric and regional circulation, air mass and pressure systems producing dry conditions in the summer months. In this case, Csb winters are slightly milder than Csa winters, the former having cold-month averages of about 13 °C. Csb regions also have higher humidity, frequent advection and evaporation fogs, and occasional low stratus, altostratus, and nimbostratus overcast.

• (4) A cold snowy forest climate with dry summer (humid microthermal) Ds (mostly Dsa and Dsb) takes place over a relatively large zone lying in the mid-northern parts of the continental central and eastern Anatolia regions of Turkey, whereas a cold snowy forest climate, humid in all seasons (humid microthermal) Df (mostly Dfb) exists over relatively small areas seen in the northern parts of the continental Central Anatolia region and the northeastern Anatolia subregion (mostly Erzurum-Kars subregion) of Turkey.

The maritime influences on the weather and climate conditions of Turkey characterized mainly by the maritime polar (mP) and maritime tropical (mT) air masses, and in winter Mediterranean air masses carried by the westerly air flows (W, SW, and NW) tend to decrease toward the continental inland regions including the central, eastern and southeastern Anatolia regions. Major changes of the inland Mediterranean regions concern decreases of precipitation amounts in winter and number of rainy days, an increase of precipitation amounts in spring months, particularly with the continentally such as in the mid-west and south parts of central Anatolia region. An increase in temperature range is observed as the continental effects are also enhanced. The Northern Anatolia Mountains, on the other hand, separate the Mediterranean, steppe, and cold snowy forest climates from the west coast temperate rainy (Black Sea in Turkey) climate, while the Taurus’ and the Southeastern Taurus Mountains separate the Mediterranean climate (Csa) from the more continental Mediterranean climate (Csb), steppe and cold snowy forest climates to the north and the east of the Anatolian Peninsula (Fig. 1).

As a requirement of the United Nations Convention to Combat Desertification (UNCCD), arid, semiarid and dry subhumid climates were defined as “areas, other than polar and sub-polar regions, in which the ratio of annual precipitation to potential evapotranspiration falls within the range from 0.05 to 0.65” (UNCCD 1995). In the present study, the UNCCD Aridity Index (AI) is used as one of the base methods for determining dry-land types in Turkey and assessing their vulnerability to the desertification processes. Following the UNEP (1993), AI is written as follows:

$$ A = (P/PE) $$

(1)

where P and PE are annual precipitation (mm) and potential evapotranspiration (mm) totals, respectively. The criteria in Table 2 were used to characterize the dry-lands of Turkey (Türkeş 1999). In theory, and according to the UNCCD Aridity Index (AF), AI values below 1.0 generally show an annual moisture (soil water) deficit in average climatic conditions, whereas AI values above 1.0 generally show an annual moisture surplus (Fig. 2). In Turkey, dry subhumid climatic conditions extend throughout most of the continental central Anatolia and southeastern Anatolia regions, some part of the eastern Mediterranean, and eastern and western parts of the continental eastern Anatolia region, while the semiarid climatic conditions are found only in the Konya Plain and the Iğdır district of the Eastern Anatolia region (Fig. 2).

Table 2 Dry-land (arid climate) types in Turkey according to the Aridity Index (AI) and their vulnerability to desertification (Türkeş 1999)

Fig. 2

Geographical distribution of the UNCCD aridity indices over Turkey, in which arid, semi-arid , dry subhumid and subhumid areas of the country are highlighted by hatching with the red color (Türkeş 1999, 2010)

The areas having values 0.65 < AI < 0.80, where an annual soil moisture deficit exists, are also concentrated around the semiarid and dry subhumid areas of Turkey. The arid lands in Turkey having AI values between 0.20 and 0.80 are likely supposed to have been influenced by the desertification processes (Table 2), by considering the existing hydro-climatological conditions, human-induced land degradation , observed and projected climate change and variability, etc. (Öztürk et al. 2012, 2015; Topçu et al. 2010; Şen et al. 2012; Türkeş 1999, 2010; Tatlı and Türkeş 2011, 2014; Türkeş and Akgündüz 2011; Altınsoy et al. 2011; Türkeş et al. 2011).

3 Future Climate of Turkey: Evolution of CMIP3/CMIP5 Models

Human influence on Earth’s climate is now accepted as a well-tested scientific hypothesis, therefore, for a sustainable environment both mitigation and adaptation are imperative at all levels of social organization. Starting point of any mitigation or adaptation effort is the generation of information concerning what the future will bring and how natural and human- built systems will behave under climate change. Such information, in turn, forms the basis for studies of the impacts of current climate variability and future climate change on all aspects of a country’s resources, from water to health, from Agriculture to urban environment. Climate simulations, which are the basis for IPCC’s Fourth and Fifth Assessment Reports (IPCC 2007, 2013), provide such information assuming several greenhouse gas emission scenarios for the future. Before giving details of the analysis of projected climate change over Turkey, it is useful to give the basic definitions behind the emission scenario design and the difference between different emission scenario approaches.

The emission scenarios basically describe the future concentrations of the main anthropogenic emissions of all relevant greenhouse gases (GHGs) and a small set of other gases such as SO₂, CO, and NO _x . From this point of view, emission scenarios can be identified as the main component of any assessment of climate change because they are used to drive global multi-component climate models for future climate change projections, such as those planned under the World Climate Research Programme’s Coupled Model Intercomparison Projects (CMIP3 for AR4, Meehl et al. 2007; CMIP5 for AR5 Taylor et al. 2012). In parallel with the developments in climate science and based on the needs of the climate science community, the designed emission scenarios also evolved from non-mitigation Special Report on Emission Scenarios (SRES) (Nakicenovic and Swart 2000) to the Representative Concentration Pathways (RCPs) (Meinshausen et al. 2011), which are based on scenarios that consider possible approaches to climate change mitigation. In AR4, the 40 different SRES emission scenarios were defined based on the range of global energy related CO₂ emissions from the literature (Nakicenovic and Swart 2000), and they are categorized into four distinct scenario families such as A1, A2, B1 and B2. The scenario family A1 is also subdivided into four scenario groups (i.e., A1C, A1G or A1FI, A1B, and A1T) based on the different future directions of technological change. According to the classification of the SRES scenarios, B1 is defined as the low-forcing scenario (CO₂ concentration of about 550 ppm by 2100), which assumes rapid economic development, low population growth, and introduction of new energy sources and/or technologies. A1B is designed as medium forcing scenario (CO₂ concentration of about 700 ppm by 2100) and A2 is defined as high forcing scenario (CO₂ concentration of about 820 ppm by 2100), which assumes a very heterogonous world with high population growth. The projected changes in the concentrations of the two main GHGs (CO₂ and CH₄) used in SRES scenarios can be seen in Fig. 3a, b, respectively.

Fig. 3

Concentrations of a CO₂ (ppm) and b CH₄ (ppb) under SRES and RCP scenarios

The AR5 differs from the previous report (AR4) in terms of the definition of emission scenarios. The new report builds upon a new concept, called representative concentration pathways (RCPs), which includes approaches to both climate change mitigation and adaptation. Basically, the RCPs assume policy actions that will be taken to achieve certain emission targets (Taylor et al. 2012). For CMIP5 project, the four-emission scenarios or RCPs (RCP2.6, RCP4.5, RCP6.0, and RCP8.5) are defined by considering future population growth, technological development and community responses. According to the definition, the radiative forcing for 2100, which is relative to the preindustrial condition, is estimated for each RCPs, and emission scenarios are created based on these assumptions. For example, radiative forcing reaches about 8.5 Wm⁻² in RCP8.5 high emission scenario with a rising pathway. Unlike RCP8.5, the RCP4.5, and RCP6.0 can be considered as intermediate level emissions scenarios whose radiative forcing’s reach 4.5 and 6.0 Wm⁻², respectively. The low-level scenario, RCP2.6, has its maximum in the middle of the twenty-first century and starts to decrease, thereafter, to reach 2.6 Wm⁻² in 2100. The comparison of CO₂ and CH₄ gas concentrations of the different RCPs and SRES scenarios can be seen in Fig. 3. It is clearly shown that RCP8.5 and SRES A2 scenarios have very similar temporal patterns for CO₂ and CH₄ gas concentrations. Likewise, RCP4.5 and SRES B1 emission scenarios are similar to each other.

The SRES and RCP scenarios are used to force multi-component (i.e., atmosphere, ocean, land, ice) global circulation models (hereafter referred as GCM ) to have an assessment of future climate change and its direct and indirect effects on the water resources, agriculture, soil, human health, etc. This section provides information on the climate change projections for Turkey using available data from the downscaled CMIP3 simulations as well as the original outputs of CMIP3 and CMIP5.

As mentioned previously, GCMs are the primary tools to obtain climate change projections based on emission scenarios. The main shortcoming of the information one can derive from the global circulation modeling studies is related to their spatial resolution, which is usually on the order of a few hundreds of kilometers. However, they can still provide a crude estimation of the regional climate change and its impacts over a studied region. Given the fact that the number of available global simulations is comparatively high, it becomes possible to use an ensemble approach that allows uncertainty analysis. Therefore, the use of the multi-model and/or multi-scenario average approach might give more robust and reliable results. This section basically aims at identifying the future climate change over Turkey by estimating the multi-model ensemble average of surface temperature and precipitation fields from the outputs of the available CMIP3 and CMIP5 simulations.

Figure 4a shows the ensemble average of surface air temperatures over Turkey for three different emission scenarios (B1, A1B, and A2). The ensemble average is calculated from the outputs of the 15 different CMIP3 models after interpolating them into a common 0.5° × 0.5° grid. The reader should also note that the colored regions show the inter-model uncertainty for each emission scenario. As it can be seen from the figure, the surface air temperature has an increasing linear trend for all SRES emission scenarios, and the change in average surface air temperatures by 2100 ranges from 1.94 to 4.31 °C. As expected, the B1 scenario shows the lowest change in surface air temperature, and the maximum change occurs for the A2 scenario. The medium emission scenario (A1B) indicates a 3.33 °C increase by the end of the twenty-first century. Unlike surface air temperatures, associated precipitation projections show decreasing trends (Fig. 4b). The SRES A2 emission scenario shows the highest significant reduction with 0.39 mm/day in 2100, and A1B and B1 follow it with 0.26 and 0.11 mm/day respectively (Fig. 4b). The combination of precipitation loss with increasing surface air temperature might adversely affect the water availability and soil moisture , especially in the regions that are under the risk of aridity (southeast and central Turkey). Kömüşçü et al. (2010) estimated the change in soil moisture as 4–43% loss for 2 °C warming without a change in the amount of precipitation and 8–91% for a 4 °C warming.

Fig. 4

Time series of annual a surface air temperature (°C), and b precipitation (mm/day) of CMIP3 model averages over Turkey for the period 2006–2099. The colors indicate different emission scenario simulations (B1: green, A1B: blue, and A2: red). The colored regions show the inter-model uncertainty for SRES scenario simulations. The solid lines represent the multi-model ensemble average and dashed lines indicate the trend. The numbers give the absolute changes by the end of the twenty-first century

The CMIP5 models can be assumed superior to the CMIP3 models considering their improved model physics (Taylor et al. 2012), finer spatial resolution (1°–2.8°—but still coarse for constructing assessments at a regional scale), integrated dynamic vegetation models and better representation of the components of the Earth system (included interactive atmosphere- chemistry models, representation of glaciers, etc.). The multi-model ensemble of surface air temperature (Fig. 5a) and precipitation (Fig. 5b) for CMIP5 models is calculated using the same methodology that is used in CMIP3 models previously. In this case, 32 models are used to calculate the multi-model ensemble for each RCPs.

Fig. 5

Time series of annual a surface air temperature (°C), and b precipitation (mm/day) of CMIP5 model averages over Turkey for the period 2006–2099. The colors indicate different emission scenario simulations (RCP45: blue, and RCP85: red). The colored regions show the inter-model uncertainty for both RCP scenarios. The solid lines represent the multi-model ensemble average and dashed lines indicate the trend. The numbers give the absolute changes by the end of the twenty-first century

The CMIP5 simulations show similar results to the CMIP3 simulations in terms of the direction of the change (i.e., increase in temperature and decrease in precipitation). The changes in temperature are, however, larger than those in CMIP3. Based on the analysis of the CMIP5 model results, changes in average surface air temperature for Turkey indicates 2.30 and 5.41 °C warming by the end of the twenty-first century for RCP4.5 and RCP8.5 respectively (Fig. 5a). Unlike the surface air temperature, precipitation shows similar signals to CMIP3 models. The maximum end of century change in precipitation is −0.33 mm/day for RCP8.5 scenarios. The RCP4.5 also depicts slight negative slope, and reaches a 0.12 mm/day deficiency by the end of the century (Fig. 5b).

3.1 Dynamical Downscaling Using CMIP3 Models: Future Climate Projections

As mentioned briefly in the previous section, the coarse-resolution GCMs may lead to substantial uncertainties in projected regional climate indicators such as surface air temperature, precipitation and soil moisture . To overcome such problems, regional climate models (RCMs) are used to downscale GCM model outputs to produce higher resolution (both spatial and temporal scale) representations of regional effects over the region including Turkey (Önol and Semazzi 2009; Önol 2012). The first step in a model-based climate change study is the assessment of the model’s performance in reproducing present-day climate conditions (Giorgi and Mearns 1999). The performance analysis of the perfect boundary condition experiment (driving the regional model at the lateral boundaries with fields obtained from reanalysis of observations) and three selected GCMs, ECHAM5, CCSM3, and HadCM3 whose climate change projections used in this study are given in Bozkurt et al. (2012). In their study, RegCM3 (Pal et al. 2007) is used as the regional climate model component. In addition, the projected seasonal and daily changes in these three GCM-driven simulations are also extensively discussed for the wider model domain covering the eastern Mediterranean–Black Sea region by Önol et al. (2014). They reported that winter runoff over Turkey’s mountainous areas increases in the second half of the twenty-first century, because the snowmelt process accelerates where the elevation is higher than 1500 m.

Bozkurt et al. (2012) suggests that the selected two GCMs (ECHAM5 and CCSM3) are highly skilled in simulating the winter precipitations and surface temperatures in Turkey. However, the CCSM3 model produces relatively drier and warmer summer conditions compared to the observations. In general, the two models could be used in the climate change and impact assessment studies as long as their strengths and weaknesses are taken into account.

The rest of this chapter includes mainly the results of the future climate change projections based on the dynamically downscaled ECHAM5 and CCSM3 model results (Önol et al. 2014). The analysis is extended to examine the extreme drought events and their future changes by using the outputs of the downscaled simulations.

3.2 Temperature and Precipitation

Figure 6 shows the annually averaged surface air temperature anomalies with respect to the 1961–1990 reference period for both CCSM3 (A1FI and A2 emission scenario) and ECHAM5 (A2 emission scenario) simulations. As can be seen from the figure, all simulations show similar warming trends in selected future periods (2010–2039, 2040–2069 and 2070–2099). The CCSM3-A2 simulation has, on the average, around 1 °C stronger warming signal than the ECHAM5-A2 simulation. This result is also consistent with the reference simulations of the same models. The CCSM3 model is generally drier and warmer than ECHAM5 (Bozkurt et al. 2012). Substantial increases in surface temperature start to appear in the second selected period (2040–2069) over Turkey and reach a level around 3.5–6.0 °C at the end of the twenty-first century. The model simulations also suggest that the increase in surface temperature over Turkey will not be uniform. The eastern interior, southern, and southeastern parts will experience greater rises in temperatures. The real added value of the dynamical downscaling can be also seen when the RCM results are compared with the GCM ’s raw surface air temperature data. In this case, RCM simulation driven by the same GCM, gives a 1.5–3.0 °C stronger warming signal than the raw data (Fig. 4).

Fig. 6

Annual changes in surface air temperature (°C) relative to climatology of 1961–1990 period

As Table 3 indicates, there are a total of three downscaled simulations (ECHAM5 A2, CCSM3 A2, and A1FI) for Turkey for the period between 2071 and 2099. All three simulations exhibit similar behaviors in the changes of surface temperatures. For instance, the changes are relatively small in winter but they increase in transition seasons and reach a peak in the summer. They mostly indicate larger increases in eastern rather than in western Turkey. The differences between model simulations arise mostly in the magnitudes of their projections. The CCSM3’s estimation of fall temperature increase for Turkey (about 5.4 °C) is larger than ECHAM5 simulation results (about 4.2 °C). A1FI simulation of CCSM3 yields 0.5–1.4 °C larger values than A2 simulation of the same model. It produces an average summer increase of 7.3 °C for the eastern Turkey.

Table 3 Projected seasonal surface temperature changes (°C) in the 2071–2099 period over 1961–1990 period based on different scenario simulations

Figure 7 shows the simulated precipitation anomalies relative to their 30-year climatology (1961–1990) of reference runs. We first note the large differences between ECHAM5 and CCSM3 simulations for the A2 emission scenario. ECHAM5 shows precipitation increase (15–20%) in the first defined future period (2010–2039) while CCSM3 mostly shows decreases. In the mid-century (2040–2069), both models show similar patterns in precipitation change: increase in the Black Sea region (10–20%) and decreases in interior and southern parts of Turkey (15–25%).

Fig. 7

Annual changes in precipitation (%) relative to climatology of 1961–1990 period

Table 4 provides the seasonal changes in precipitation from the three different simulations for the 2071–2099 periods. There are broad agreements between the model estimations of the precipitation changes for the same scenario (i.e., A2). However, the magnitude of the changes may not be fully consistent because areas outside of ‘hot spots’ may show different sensitivity to the increased emissions in different models, and this affects the average values. The projected changes in precipitation are usually stronger in the CCSM3’s A1FI simulation than those in its A2 simulation, especially in fall and spring, which are wet seasons. All models broadly agree that Turkey will have less annual precipitation in the last 30-year period of the twenty-first century compared to the present times.

Table 4 Projected seasonal precipitation changes (%) in 2071–2099 period over the 1961–1990 period based on different scenario simulations

3.3 Soil Moisture

The soil moisture can be defined as the amount of water stored in the soil. The Biosphere-Atmosphere Transfer Scheme (BATS, Dickinson et al. 1993) represents the land-surface processes in RegCM3. In BATS, the soil is characterized by three layers to calculate soil moisture and temperature. In addition, it also includes a one-layer vegetation scheme and simple surface runoff component. The change in soil moisture has an important effect on agriculture, potential evaporation and surface runoff (IPCC 2001). Changes in soil moisture are predominantly related to two major factors: regional climate change and soil characteristics. This section presents the future change in soil moisture under different SRES emission scenarios (A1FI and A2).

The change in simulated soil moisture with respect to the 1961–1990 periods is illustrated in Fig. 8. As can be seen from the figure, the change signal with decreasing trend in soil moisture is much stronger along the Taurus mountain range compared to the surrounding region. The difference reaches 20% in the CCSM3 A1FI and 10% in the ECHAM5 and CCSM3 A2 emissions scenario, but the spatial extent of the change is larger in A1FI simulation. This result is also consistent with the change in surface runoff, which is not shown here (Önol et al. 2014). This is most likely an indication of early snow melting in response to the increased surface temperatures and decreased precipitation in the higher elevation regions. The projected change in soil moisture also reaches higher values (>30%) in the last 30-year period (2070–2099).

Fig. 8

Annual changes in soil moisture (%) relative to the climatology of 1961–1990 period

The soil moisture analysis is extended to analyze the changes in the soil moisture amounts for the 26 major basins in Turkey (Fig. 9). The figure indicates that the CCSM3 and ECHAM5 models (with A2 emission scenario) show different behavior in the first 30-year of the twenty-first century. The decreasing signal in CCSM A2 and A1FI emission scenario (0.5–1.0 mm) is reversed in ECHAM5 A2 scenario, but all the models show similar results after 2040s. The Çoruh and Aras basins also behave differently compared to the other basins. Soil moisture in these basins continues to increase (~0.5 mm) until 2070s before it starts to decrease. As could be seen from the figure, the change in the Seyhan basin is stronger than the other basins. This also suggests that the Seyhan basin is more sensitive (>1.0–1.5 mm) to the change in the regional climate.

Fig. 9

Changes in basin averaged soil moisture (mm) relative to climatology of the 1961–1990 period

3.4 Drought Indices: Palmer Drought Severity Index (PDSI )

Climate indices provide information on the extreme events such as drought and floods that might affect the daily life in a negative way. A drought is an extended period of months or years when a region notes a deficiency in its water supply. Drought, which can be also characterized by a decrease in precipitation over a period of months to years, is one of the most important climate hazards due to its impact on agriculture, water resources and also human health (Dai 2011a). Recent studies investigating drought events in Turkey have mainly focused on their historical trends (Tatlı and Türkeş 2011) and their effects on crop yields. In this respect, studies on corn (Durdu 2012), wheat (Özdoğan 2011), and olives (Tunalıoğlu and Durdu 2012) predict a decrease in production. Thus, this section aims to provide future projections about drought events over Turkey.

The Palmer drought severity index (PDSI , Palmer 1965) is a widely used measure of drought events (Heim 2002). PDSI and its variants have been used to investigate long-term changes of aridity over land in both global (Dai et al. 2004; Dai 2011a, b) and also regional scale, such as the Mediterranean Basin (Sousa et al. 2011). In this study, we used a modified version of the PDSI, i.e., the self-calibrating PDSI (scPDSI). The scPDSI basically replaces the empirically derived climatic characteristic and duration factors with automatically calculated values based upon the historical climatic data of a location (Wells et al. 2004). The scPDSI is selected in this study because it provides more geographically comparable classification of climate divisions than the original PDSI (Wells et al. 2004).

The monthly scPDSI is calculated using both RCM simulated and observed monthly surface air temperature and precipitation. The potential evapotranspiration (PET) is estimated using Thornthwaite’s method (Thornthwaite 1948) rather than Penman-Monteith parameterization, but Van der Schrier et al. (2011) showed that PDSI values based on these two different PET calculations give similar results in terms of regional averages and trends. In addition to the surface air temperature and precipitation, the calculation of scPDSI values also closely depend on the available water-holding capacity of the soil. A soil texture-based water-holding-capacity map from Webb et al. (1993) is used for this purpose. The original dataset of water-holding capacity is in 1° × 1° resolution but it is interpolated into high-resolution RCM grids (27 km) to perform scPDSI calculations. The scPDSI calculations were started 10 years earlier than the beginning of the RCM simulations (1950 for historical and 1990 for future simulations) using climatological monthly mean values for temperature and precipitation at each grid box. This eliminates any spin-up problems related to the scPDSI calculation. The classification used in scPDSI can be seen in Table 5. The values from 0.5 to −0.5 are considered as normal conditions and those smaller than −0.5 are associated with drought events with different severity.

Table 5 Classification of dry and wet conditions as defined by Palmer (1965) for the PDSI

The simulated scPDSI statistics are assessed based on the results of the RCM simulation driven by the NCEP/NCAR Reanalysis dataset (Kalnay et al. 1996) for the 1961–1990 period. The simulated PDSI values are compared to the observation-based scPDSI values, which are calculated using data provided by the Climate Research Unit (CRU) of the University of East Anglia, UK (Mitchell and Jones 2005). The time series of the monthly scPDSI values averaged over Turkey for the model simulation (NCEP.RF) and CRU-based scPDSI values can be seen in Fig. 10.

Fig. 10

The time series of the monthly PDSI estimated by CRU observational data and RCM simulation forced with NCEP/NCAR data during the 1961–1990 period

The correlation between simulated and CRU-based scPDSI values are around 0.6. As can be seen from the figure, the scPDSI values are generally less than −0.5, which indicate that slight dry conditions are generally dominant over Turkey. The reader should also note that the simulated scPDSI values show a stronger drought signal after the first quarter of the 80s, which is not seen in the CRU-based drought index calculations. Despite this, it could be said that RegCM3 is able to satisfactorily reproduce observed drought events and can be used to produce future drought projections for Turkey.

The probability distribution or frequencies of drought events over Turkey in terms of the metric of scPDSI can be seen in Fig. 11. The topmost plot (Fig. 11a) shows the probability of wet and dry years between 1961 and 1990 for model simulations and also observation (CRU)-based scPDSI. It is clearly seen that models are broadly able to reproduce the frequency of the drought events when they are compared with the NCEP/NCAR and CRU-based scPDSI. In general, the frequent occurrence of slight (25%) and moderate (20%) dry events are overriding in Turkey. It should also be noted that the frequency of extreme drought events is around 5% in the reference period.

Fig. 11

Probability distribution of PDSI (in fraction) for a reference period (1961–1990), b 2010–2039, c 2040–2059 and d 2070–2099 for A1FI and A2 future SRES scenarios

Figure 11b–d displays the future projections of drought events for three different time periods (2010–2039, 2040–2069, and 2070–2099). The probability distribution of scPDSI shows a completely different behavior in the first selected period (2010–2039), and the shift to the relatively wet years can be clearly seen in Fig. 11b. The frequency of drought events is around 10–15% for slight and moderate cases and less than 2% for extreme cases in this period. As for the drought events, floods are also very important extreme events. The occurrence of extremely to moderate wet years start to increase and reach 8–15% for the ECHAM5 A2 emission scenario and 2–8% for CCSM3 A2. The A1FI scenario indicates lower probability of wet years than the ECHAM5 A2 scenario, which is also consistent with the analysis of the reference period. In fact, the CCSM3 model is generally drier than the ECHAM5 model (Bozkurt et al. 2012; Önol et al. 2014).

The probability distributions of scPDSI values in the 2040–2069 periods (Fig. 11c) are very similar to the reference period (Fig. 11a). In this case, two different emission scenarios of the CCSM3 model (A2 and A1FI) also behave very similarly, but extreme drought events are doubled in A1FI.

The most significant change in the probability distribution of scPDSI is seen in the last 30 years of the twenty-first century (Fig. 11d). The frequency of extreme drought events reaches up to 40% in ECHAM5 A2 and around 30% in CCSM3 A2 and A1FI simulation in this period. Based on the projections of all the RegCM3 simulations, the occurrence of the strong drought events will be dominant, and occurrence of the dry years will increase in the future.

The spatial extents of the drought events are also important. Figure 12 shows the probability of drought events (scPDSI < −2) for different time slices. Similar to the Fig. 11b–d, the probability of the dry years increases and reaches its maximum value in the 2070–2099 periods. In the same period, the southwestern, southeastern and central parts of Turkey will be affected mostly by extreme drought events. Again, the ECHAM5 A2 simulation shows stronger signal in probability of drought events than the CCSM3 A1FI in the last 30-year.

Fig. 12

The probability of the drought events (PDSI  < −2) estimated by RCM simulations forced by CCSM3 A2, A1FI and ECHAM5 A2 data for 2010–2039, 2040–2069, and 2070–2099 periods

4 Conclusions

This chapter aims to provide detailed information concerning the main features of the climate of Turkey, which is mainly effected by the surrounding large water masses (i.e., Mediterranean and Black Sea) and Atlantic and Mediterranean (mainly the Genoa Bay and Adriatic Sea) originated cyclonic systems. In addition to the detailed information about the regional climate system and its main drivers, the Köppen-Geiger climate classification system and an Aridity Index analysis are also presented to extend the given analysis from the perspective of land surface and soil. The results indicated that the climate of Turkey according to the Köppen- Geiger climate system is very diverse. The sub-tropical steppe climate (BS) is found in the central Anatolia region as well as the farthest eastern part of eastern Anatolia. The coastal Black Sea region is mainly dominated by a temperate rainy or humid temperate west coast climate without a dry season (mostly Cfa and Cfb). The regions around the coast of the Mediterranean Sea (Marmara, Aegean, and southwestern Anatolia regions) belong to the dry summer subtropical Mediterranean climate or temperate rainy climate with dry summer (mostly Csa). Lastly, a cold snowy forest climate with dry summer (mostly Dsa and Dsb) takes place over a relatively large zone lying along the mid-northern portions of the continental regions of Turkey (central and eastern Anatolia), whereas a cold snowy forest climate humid in all seasons (humid microthermal, mostly Dfb) exists over relatively small areas seen in the northern portions of the continental central Anatolia region and the northeastern Anatolia subregion of Turkey.

As for future climate of Turkey, an analysis involving the century-long changes obtained directly from the CMIP3 and CMIP5 GCM simulations and a more detailed analysis based on the dynamically downscaled outputs of some of the CMIP3 simulations are included. The number of the scenario/model simulations in the latter is limited, which hinders a thorough analysis including the estimations of the uncertainty metrics. Nevertheless, given the fact that these simulations provide much higher resolution climate parameters, they are extremely beneficial for the illustration of the spatial distributions for Turkey that has a very heterogeneous topography and landscape, which cannot be resolved by the coarse-resolution General Circulation Models. It is also important to mention here the changes occurring in projections for the early periods of the 21st century. Because the projected emissions in these periods are still close to the present-day emissions, the changes (i.e., the signals) should be considered with caution. The signal-to-noise ratio improves toward the end of the century, and the climate change signal becomes stronger and spatially more distinctive. The agreement between different model simulations of the same emission scenario also increases.

Having said these, it is possible to draw the following relatively robust conclusions from the climate change projections that were obtained from different model simulations:

• 1. All simulations agree on a temperature increase in Turkey in the twenty-first century. The different scenario-based CMIP3 simulations indicate increases between 1.94 and 4.31 °C by the end of the century, while the CMIP5 simulations yield increases between 2.50 and 5.41 °C. The downscaled simulations of CMIP3 give information about the spatial distribution of the temperature changes. They indicate larger increases towards the central and eastern parts compared to the coastal parts of Turkey.

• 2. All simulations agree on a precipitation reduction in Turkey in the twenty-first century. The reductions estimated by both CMIP3 and CMIP5 simulations for the end of the century broadly agree in magnitude (between 0.11 and 0.39 mm/day), as well. For regional changes, the high-resolution simulations indicate reductions in the Mediterranean region of Turkey while they consistently exhibit increases for the eastern Black Sea region .

• 3. Although not shown here, all simulations agree on a reduction in the total runoff of Turkey by the end of the century. Moreover, the simulations indicate a reduction in the spring runoff while an increase in winter runoff in eastern Anatolia in response primarily to increased temperatures is foreseen. These results may have important implications for the irrigation of the agricultural fields in southeastern Anatolia as well as the energy production that depend on the snow-fed rivers of the region.

• 4. In parallel to the increased surface air temperature and decreased amount of precipitation, the frequency of the extreme drought events is projected to increase by the end of the twenty-first century. Consequently, the assessment of the impacts of climate change on water resources and agriculture will become crucial for a sustainable future.