It has been argued that dam construction and irrigation projects in Turkey and Syria would diminish the flow of the Euphrates River and increase salinity levels. However, previous studies have not systematically defined the baseline changes in the quantity and quality of the river water. In this study, we associate variations in flow quantity and quality of water with natural aspects. We analysed flow data from 1970 to 2010 for the river at the Turkish-Syrian border to assess the effect of upstream dam construction, correlating these data with precipitation variation in Syria in the same period. In addition, we compared flow data between Syria and Iraq in a period with no dam construction in those countries (2000 to 2010). Regarding water quality, we collected samples from the Turkish-Syrian and Syrian-Iraqi borders to identify levels of salinity through laboratorial analysis. We complemented the data analysis with fieldwork along the river basin, and with evidence from the literature. We conclude that the slight diminishing of flow as well as the increasing salinity towards downstream are both associated with natural aspects of the landscape, more than human actions.

Introduction

References to the link between decreasing water flow and dam construction in the Euphrates River are commonplace, as is the attribution of increasing salinity to irrigation projects and reservoirs formed by dams. Conclusions have invariably led to predictions of Iraq being subjected to drought and salty water caused by Turkish and Syrian dam and irrigation projects. Although the Inventory of Shared Water Resources in Western Asia (United Nations Economic and Social Commission for Western Asia (UN-ESCWA), 2013) considers climate and soil information, it tends to emphasise the human impacts and only regards natural conditions as vague influences on lowering flow and raising salinity. Moreover, water pollution is often related to irrigation because of the use of fertilisers that infiltrate the river plain. However, these assertions neglect the effect of natural causes, and are based on imprecise data on the quantity and intensity of pollutants.

In this study, we address this disagreement by demonstrating that natural aspects of the geographical landscape are more strongly associated with the variation of flow (quantity) and salinity (quality) of the Euphrates River than human interventions.

We first examined 41 years of river flow and precipitation data. These two variables showed a stronger correlation with each other than the one between flow and upstream dam constructions. Then we compared these data with other sources to reinforce our conclusions. Next, we analysed the water quality using samples collected from the Turkish-Syrian and Syrian-Iraqi borders to identify levels of salinity, and we compare these findings with the results of other studies. Then, we examined a set of climate features and geomorphological and hydrological influences that may explain the gradual increase of salinity and a slight diminishing of flow from upstream to downstream. We complement the analysis with data from the literature, statistics and findings from images and fieldworks to add observational evidences.

This blend of methodologies was necessary to compensate for the lack of substantial data. In other words, it would be useful, for instance, to carry out a water balance, as a recognisably recommended methodology for river basin studies. To do so, however, it would be necessary to add specific data such as evapotranspiration and infiltration, which is not available for the entire basin. Likewise, modelling would be another suitable methodology choice, which could not only provide a rigorous explanation about the basin but could also identify trends in variables. Again, we had to renounce this for the same reasons as previously mentioned. As a result, the methodology was adapted to data availability, resulting in a broader approach regarding the landscape features, rather than specific data.

Even so, we could construct a logical argument that allowed us to conclude that the variation of the Euphrates flow is a seasonal phenomenon, more associated with natural factors than with human-made factors. Additionally, we concluded that the gradual increase of salinity towards downstream is also a natural tendency rather than the effect of dams or irrigation projects, although human actions can exacerbate this process.

Once this article has addressed the natural aspects of the basin and recognised the natural fragilities and tendencies of the landscape, the conclusions drawn here are likely to be relevant to regional water management, which currently appears to focus on engineering and political matters.

The article is organised as follows. The next section describes the geographical aspects of the Euphrates River basin. The basin is divided into three sections with distinctive features: the upstream (Turkey), middle stream (Syria) and downstream (Iraq). A quantitative analysis is then presented based on literature information, statistical data and observational evidence. The subsequent section explains the qualitative analysis based on empirical observation and laboratorial results. Then, conclusions are drawn from the presented evidence.

Background information of the geographical landscape

The Euphrates River basin lies between latitudes 30° N and 40° N and longitudes 37° E and 49° E (Figure 1). Its area totals 440,000 km2, of which 28% is in Turkey (123,200 km2), 22% in Syria (96,800 km2) and 47% in Iraq (206,800 km2).
Fig. 1.

Map of the Euphrates basin. Source:UN-ESCWA (2013, pp. 50–51).

Fig. 1.

Map of the Euphrates basin. Source:UN-ESCWA (2013, pp. 50–51).

The total length of the Euphrates River from its source to the confluence with the Tigris is 2,780 km. The confluence of the Euphrates and Tigris forms the river Shatt Al-Arab, which flows approximately 192 km before its outflow into the Persian Gulf.

Upstream – Turkey

The convergence of Karasu and Murat rivers forms the Euphrates River in the Anatolian uplands, at more than 3,000 m above sea level. In this Turkish stretch, the river crosses a mountainous relief and receives several perennial tributaries until it reaches Syria, 434 km later.

The climate in this upstream region presents low average temperatures and an annual rainfall average of 570 mm/year. The river also receives an important contribution from the Anatolian snow that melts from April to June. All these natural features explain the hydroelectric potential of the river in this stretch.

Middle stream – Syria

The Euphrates River enters Syria through the north at 500 m of altitude, flowing towards the south-east and crossing a sedimentary low plateau covered with steppes (Figure 2). This is a transitional landscape from the colder and moister Turkey to the hotter and more desert Iraq.
Fig. 2.

The Syrian sedimentary plateau. The central figure source: http://www.freeworldmaps.net/asia/syria/map.html (accessed 23 August 2011).

Fig. 2.

The Syrian sedimentary plateau. The central figure source: http://www.freeworldmaps.net/asia/syria/map.html (accessed 23 August 2011).

In Syria, the river receives just three perennial tributaries: the rivers Balikh and Khabur on its left bank, with average annual flows of 6.7 m3/s and 9.3 m3/s respectively, and the river Sajur (average flow of 4.1 m3/s) on its right bank in northern Syria.

After approximately 610 km, it reaches the Iraqi border just below the city of Abukamal. Although this descent is much smoother and more gradual than the previous stretch in Turkey, the river has excavated valleys in the sedimentary structure, occasionally favourable for dam construction, such as the Al-Assad reservoir (Figure 3), the largest in Syria, with 63,000 ha of water surface.
Fig. 3.

Al-Assad Dam (Google Earth (a) and local photograph (b), February 2011).

Fig. 3.

Al-Assad Dam (Google Earth (a) and local photograph (b), February 2011).

Along its crossing through the Syrian plateau, the Euphrates River forms a valley of 2 to 12 km in width, with fluvial plains 250 m wide (on average) on both banks. The climatic characteristics of this stretch reflect the same regional tendencies of the whole basin: increasing of aridity towards downstream, with gradual reduction in precipitation and air humidity due to a higher evaporation tax. According to the Climatic Atlas of Syria (1977), the annual average rainfall where the Euphrates River enters Syria at the Turkish border (250 mm) decreases downstream reaching only 50 mm/year at the Syria-Iraq border. Similarly, the relative humidity (annual average) decreases along the Syrian stretch from 70% to 40%. However, the potential evapotranspiration increases from 330 mm at the Turkish border to 3,000 mm at the Iraqi border.

Soils in the river basin show gypsum and layers of impermeable clay. By bringing up salts through capillarity, irrigation accentuates the salinity of soils and water. Gypsiferous soils are strongly related to salinity and constitute one of the main groups of salted soils. Shainberg and Levy reaffirm that ‘under arid and semiarid conditions, and in regions of poor natural drainage, there is a real hazard of salt accumulation in soils’ (Shainberg & Levy, 2005, p. 434).

Integrating the pedologic, climatic and hydrographic features (including the lack of perennial tributaries), we can observe an increasing trend towards aridness, which must be considered in the explanation of any decreasing of river flow and an increasing salinity towards downstream.

Downstream Euphrates – Iraq

After the river enters Iraq it no longer receives contributions from any tributaries until its convergence with the Tigris River in southern Iraq. Here, the river flow depends exclusively on the water coming from upstream regions and on the low annual precipitation. In the Iraqi stretch, the landscape is similar to the Syrian sedimentary plateau but gradually more arid. The topography forms a barren plain devoid of rivers except for the Euphrates itself and certain natural and man-made canals. The altitude decreases from 165 m at the Syrian border to sea level approximately 1,000 km later, producing a low fluvial gradient of only 18 cm/km. Continuous fluvial and lacustrine plains are characteristic of the region as well as canals built for irrigation and water diversion purposes.

The region where the Euphrates and Tigris converge constitutes a large alluvial delta common to both rivers. Immediately before the confluence of the rivers, there is a wide swampy area with approximately 15,000 m2 of marshes known as Hawr al Hammar. In this region, the waters of the two rivers have a high load of silt that forms a depositional delta just before their confluence. It is estimated that the surface of this inland delta has risen by approximately 20 cm in the last century due to sedimentation. Because of this depositional process, this region is subject to regular and intense flooding, which also increases the evaporation surface.

The poor drainage, flooding and irrigation in this region have resulted in salt concentrations in the upper soil horizons, which is accentuated by higher rates of evaporation and low levels of rainfall.

The flood season ends in June, followed by the onset of a hot and dry summer. As a result, the flow is lowest when the water demand is highest, and pressure from downstream resources increases. These patterns produce, in both middle stream and downstream, a high discharge amplitude, in which there are periods of high flow that are many times higher than the periods of low flow in the dry season. There is also extreme variability from year to year, as in 1954 when the flow was ten times higher than it was in 1953. And also in the same year flow can vary severely, as in 2006, when the lowest flow was 118 m3/s and the highest was 1,762 m3/s (15 times higher), in the Syrian stretch. In this sense, the dams provide control over the flow by moderating the extremes of high and low flow. Thus, Iraq is affected by both floods and droughts and was the first country to begin engineering work along the Euphrates River. In 1914, the Hindiyah Dam was built to deviate floods to a lower land area, forming Lake Razzaza. In 1948, the Ramadi Dam was built to divert floods to Lake Habbaniyah (Habbaniyah Escape). The water from Habbaniyah was then returned to the Euphrates River or diverted to Lake Razzaza by the Mujarra Canal, which was built in 1957. The last and greatest work in the Iraqi stretch of the Euphrates River was the Haditha Dam, which formed Lake Al Qadisiyah, with 500 km2 of water mirror and a storage capacity of 8.2 billion cubic metres (BCM, 10¹² m³) (UN-ESCWA, 2013, p. 63).

Landscape conclusions

The climatic characteristics of the basin clearly indicate an increasingly arid tendency from upstream to downstream. Upstream (in Turkey), the average annual precipitation can reach 1,000 mm, whereas in the middle stream (Syria), precipitation is approximately 250 mm, and it decreases to less than 100 mm downstream (Iraq), as shown in Figure 4.
Fig. 4.

Average annual precipitation in the Euphrates River basin.

Fig. 4.

Average annual precipitation in the Euphrates River basin.

Graphs of the annual variation of air temperature (°C) and precipitation (mm) of Turkey, Syria and Iraq confirm these climatic characteristics of the Euphrates River basin (Figure 5). At the weather station of Erzinkan, temperature and precipitation measurements indicate a mountainous Mediterranean climate, with hot and dry summers (between June and September) and cold and wet winters. In these uplands, precipitation extends from autumn until spring, passing through the winter (December to March) with rain and blizzards.
Fig. 5.

Graphs of annual variation of temperature and precipitation of Erzinkan (Turkey), Deir Ez-Zor (Syria) and Basrah (Iraq) illustrating the gradual increase in aridity and rainfall. Source:UN-ESCWA (2013, p. 56).

Fig. 5.

Graphs of annual variation of temperature and precipitation of Erzinkan (Turkey), Deir Ez-Zor (Syria) and Basrah (Iraq) illustrating the gradual increase in aridity and rainfall. Source:UN-ESCWA (2013, p. 56).

The influence of the Mediterranean climate decreases downstream and Syria features a transition from Mediterranean to desert climate, which is predominant downstream in Iraq when rainfall becomes much rarer and temperatures can reach 50 °C.

On the one hand, all these natural features, such as: gradual increase in temperature and evaporation rate; the gradual reduction of air humidity and rainfall; decrease to total absence of tributaries, must be considered to explain occasional decreases in flow and increases in salinity as a natural tendency. On the other hand, the increasing requirements for irrigation water (because of a soaring aridity) towards downstream can also intensify these natural processes. All these arguments will be reinforced by data analysis in the subsequent sections.

Water quantity and quality in the Euphrates river

Quantity of water

Here we analyse the annual discharge data of the Euphrates River, based on three types of information: literature, statistical data and empirical observations.

Literature

Several studies of the regional geography prior to dam construction were undertaken and published. Chaieb (1955) recorded the annual average flow of the Euphrates River of 525 m3/s in 1955 in the Syria stretch. In the same decade, Gourou (1953, p. 483) reported an average flow of 710 m3/s. Much earlier, however, Blanchard (1929, p. 223) described slightly different flow data, referring to the lower reaches in Iraq: 400 m3/s in October, reaching 2.750 m3/s at the end of April, when the water level can be 3.5 m higher than October. These data confirm the high seasonal amplitude of discharge variation shown in the above section. However, the historic literature does not show evidence of either increasing or decreasing flow. To identify such tendencies, we have to compare data prior to the construction of the dams along the Euphrates to post-dam construction data, which will be discussed in the next section in an attempt to prove the hypothesis that changes in quantity and quality of water are due to a natural tendency.

Data and statistical analysis

Initially, we consider data from the Inventory of Shared Water Resources in Western Asia (UN-ESCWA, 2013) as the most recent and pertinent document on this matter (Table 1).

Table 1.

Annual average flow of the Euphrates River between 1930 and 2011 measured at different points.

Station Period Mean (BCM) Minimum (BCM) Maximum (BCM) CVa [-] 
Jarablus (Syria) 1938–2010 26.6 12.7 56.8 0.33 
1938–1973 30.0 15.0 56.8 0.29 
1974–1987 24.9 12.7 34.1 0.27 
1988–1998 25.5 14.4 50.1 0.42 
1974–1998 25.1 12.7 50.1 0.34 
1990–2010 22.8 14.4 32.6 0.34 
Hussaybah (Iraq) 1981–2011 20.0 8.9 47.6 0.44 
1988–1998 22.8 8.9 47.6 0.54 
1999–2010 15.5 9.3 20.7 0.27 
1990–2010 16.8 8.9 30.7 0.39 
Hit (Iraq) 1932–1998 27.1 9.0 63.0 0.36 
1938–1973 30.6 15.1 63.0 0.30 
1974–1987 23.1 9.3 31.2 0.32 
1988–1998 22.4 9.0 46.6 0.51 
1974–1998 22.8 9.0 46.6 0.40 
Hindiyah (Iraq) 1930–1999 17.6 3.1 40.0 0.4 
1938–1973 19.8 6.6 40.0 0.35 
1974–1987 15.3 3.1 24.1 0.45 
1988–1998 13.8 7.7 27.9 0.48 
1974–1998 14.7 3.1 27.9 0.46 
Station Period Mean (BCM) Minimum (BCM) Maximum (BCM) CVa [-] 
Jarablus (Syria) 1938–2010 26.6 12.7 56.8 0.33 
1938–1973 30.0 15.0 56.8 0.29 
1974–1987 24.9 12.7 34.1 0.27 
1988–1998 25.5 14.4 50.1 0.42 
1974–1998 25.1 12.7 50.1 0.34 
1990–2010 22.8 14.4 32.6 0.34 
Hussaybah (Iraq) 1981–2011 20.0 8.9 47.6 0.44 
1988–1998 22.8 8.9 47.6 0.54 
1999–2010 15.5 9.3 20.7 0.27 
1990–2010 16.8 8.9 30.7 0.39 
Hit (Iraq) 1932–1998 27.1 9.0 63.0 0.36 
1938–1973 30.6 15.1 63.0 0.30 
1974–1987 23.1 9.3 31.2 0.32 
1988–1998 22.4 9.0 46.6 0.51 
1974–1998 22.8 9.0 46.6 0.40 
Hindiyah (Iraq) 1930–1999 17.6 3.1 40.0 0.4 
1938–1973 19.8 6.6 40.0 0.35 
1974–1987 15.3 3.1 24.1 0.45 
1988–1998 13.8 7.7 27.9 0.48 
1974–1998 14.7 3.1 27.9 0.46 

BCM is billion cubic metres (10¹² m³) and CV is the coefficient of variation that is defined as the ratio of the standard deviation to the mean for each period. Source:UN-ESCWA (2013).

The data show the average flow of the Euphrates River in distinct periods and measurements at different points along the river: one in Syria (at the border with Turkey) and three in Iraq. Although the periods are not the same for each measuring station because of discrepancies in the available data, we can identify a main division of time: the so-called ‘natural period’ that refers to the time prior to dam construction, which is before 1973, and the subsequent period that refers to the time after the main dam construction, which means, post 1973. The explanatory text in the Inventory states that, ‘Measured flow characteristics changed with the filling of the Keban Dam Reservoir in Turkey and Lake Assad in Syria in the winter of 1973–74. This is reflected in the discharge downstream’ (UN-ESCWA, 2013, p. 58). Further, the text states that ‘Before 1973, the mean annual flow of the Euphrates at the Syrian-Turkish border (Jarablus) was approximately 30 BCM, but this figure dropped to 25.1 BCM after 1974 and fell to 22.8 BCM after 1990’ (p. 59).

Although the Inventory also points to a possible climatic contribution to the decreasing flow along the dams, it does not distinguish the contributions of these two variables to this process. In saying that the decrease of the flow may be ‘possibly reflecting a combination of drier weather conditions and the effects of extensive dam building’ (p. 59), or that ‘droughts and the construction of dams accounts for the diminishing flow’ (p. 59), the Inventory leaves this question as an open hypothesis to be verified. Even so, the way the data are presented (‘this figure dropped to 25.1 after 1974’) leads us to believe that the reduced flow is conclusively a consequence of the dams. However, we cannot accept that because of three reasons.

Firstly, we can observe a decreasing flow in the ‘natural period’ before the construction of dams. By comparing the 1938–1973 average flow measured in Jarablus (Syria, of 30 BCM) to the same period of flow in Hindiyah (Iraq, of 19.8 BCM on average) we have strong evidence that this reduction is due to natural factors, since there were no significant human interventions in the river during this period. Secondly, focusing only on the 1974–1987 period flow of Jarablus (24.9 BCM), Hit (23.1 BCM) and Hindiyah (15.3 BCM), we can build further evidence of this natural tendency. Within this period of 13 years we had construction only of the Keban dam, in Turkey, 1975. The next dam (Karakaya, Turkey) would not be constructed until 1987 and with only a small storage capacity (9.5 km3). Although the Keban dam has a greater storage capacity (31 km3) the subsequent decreasing of flow from Jarablus (Syria) to Hit (Iraq) was of only 1.8 BCM during the period, while the decrease in Iraq (much further downstream from the Keban dam than Jarablus) was from 23.1 BCM (Hit) to 15.3 BCM (Hindiyah). This drop of 8.2 BCM in the flow between the two Iraqi points cannot be more attributed to the Keban construction than to climate factors, remembering that the Iraqui stretch of the Euphrates flows in desert conditions. Secondly, by presenting average flow within periods, the Inventory may camouflage this diminishing tendency shown above. Hence, what is identified as a drop from one period to another could be a gradual and continuous decrease due to natural factors and, consequently, might not be inextricably linked to the dam construction.

The third logical reason by which we can question the Inventory conclusions is that the statistical analysis of 1970–2010-flow data in Jarablus, downstream from all Turkish dams, shows a stationary temporal series, i.e., the river flow maintains an approximately constant average over time, reflecting a stable equilibrium, even with the construction of dams.

We employed three non-parametric tests widely used in hydrological studies (Lettenmaier et al., 1994; Moraes et al., 1998; Yue et al., 2002; Kahya & Kalaycu, 2004) to detect trends and changes in annual streamflow for the period of 1970 to 2010 in Jarablus station.

The Mann-Kendall and Spearman tests detected trends in time series data and the Pettitt test identified the existence of change points in streamflow.

A time trend, possibly present in a hydrological series, can be detected by correlation between the number and the time index. This is the essential idea of the non-parametric Spearman test described by Nerc (1975). The statistical result of the Spearman test was −1.52. This value indicates that the hypothesis of stationarity cannot be rejected (for a 0.05 significance level).

The sequential form of the Mann-Kendall test, consisting of the application of the test to all the series starting with the first term and ending with the ith and to those starting with the ith one and ending with the first, was also used for a progressive analysis of the series. In the absence of any trend, the graphical representation of the direct (U(t)) and the backward (U*(t)) series obtained with this method gives curves that overlap several times. A detailed description of the method is found in Moraes et al. (1998).

Figure 6 shows graphic U(t) and U*(t) curves of the Mann-Kendall test. The horizontal lines represent significance levels of 10% and −10% (continuous lines) and 5% and −5% (dashed lines). For example, when statistic U(t) (bold line) intersects the 10% horizontal axis we have a positive trend with 90% significance.
Fig. 6.

Mann-Kendall test for annual stream flow in Jarablus station.

Fig. 6.

Mann-Kendall test for annual stream flow in Jarablus station.

Figure 6 shows a positive trend statistically significant trend only in 1982–1983, which does not persist over time. Although time series presents a decline after 1983, the variability is wholly in the non-significant interval. This result shows that mean annual stream flow in Jarablus station does not present statistically significant increased or decreased trends in the observed period.

The Mann-Kendall test can also identify abrupt changes in time series. This can be identified by crossing U(t) and U*(t) lines. However, many intersections in time series can complicate interpretation of the results. Therefore, we performed a Pettitt test to analyse change-points in streamflow, which is more sensitive than the Mann-Kendall test.

The central idea of the test is to compare the distribution of features from successive sequences of data analysis. Considering a sequence of random variables X1, X2,…, XT, then the sequence is said to have a change-point at τ by Pettitt test if Xt for t = 1,…,τ has a distribution function F1(x) and Xt for t = τ + 1,τ + 2,…,T has a distribution function F2(x) and F1(x)F2(x) (Pettitt, 1979).

The Pettitt test statistic can be represented graphically and the change-points occur when the statistic crosses the line of a given significance level. This test did not identify statistically significant change-points for the time series (Figure 7). Therefore, we cannot assert that the streamflow time series of Jarablus station has a ‘natural period’ (before dam constructions) or that dam operation has a strong influence over river flow. Figure 7 shows a time series of Pettitt statistics (u(t,n)) for Jarablus station.
Fig. 7.

Pettitt test for annual stream flow in Jarablus station.

Fig. 7.

Pettitt test for annual stream flow in Jarablus station.

In the 41-year series of data (Figure 6), we observe some remarkable variations of flow from one year to another, such as from 1987 to 1988. We also notice that certain periods of decline are followed by recovery of the flow, such as from 1973 to 1976, and from 1990 to 1994. First, we considered whether these decreases followed by recovery might reflect periods of dam filling. Although the completion of a dam is relatively fast (depending on the storage potential), it can affect the average annual flow of the downstream stretches. However, after reaching its storage limit, dams must resume releasing water and the river gradually regains its previous flow. To illustrate this, we have plotted, in Figure 8, the dates of the main Turkish dams built upstream of the Syrian city of Jarablus, where the flow measurements were made.
Fig. 8.

Flow of the Euphrates River between 1970 and 2010 and Turkish dams upstream of the point of measurement (Jarablus, Syria, bordering Turkey). (*) Dams built on tributaries of the Euphrates River. Sources:Statistical Abstracts (1970 to 2011) and UN-ESCWA (2013).

Fig. 8.

Flow of the Euphrates River between 1970 and 2010 and Turkish dams upstream of the point of measurement (Jarablus, Syria, bordering Turkey). (*) Dams built on tributaries of the Euphrates River. Sources:Statistical Abstracts (1970 to 2011) and UN-ESCWA (2013).

Relying on these 41 years of flow data, noting the main dams, we have produced four arguments to challenge the automatic relationship commonly assumed between dams and low flow.

First, considering the years 1989 (1,590 m3/s) and 1990 (482 m3/s), the flow fell to less than a third from one year to the next, recovering in 1994. This drop in flow may initially be related to the construction of Ataturk Dam, the largest in Turkey in water storage volume. However, the Ataturk Dam was completed in 1992 and it is decisively unlikely that the filling of the reservoir began 2 years before completion of the dam. The same observation was made for the Keban and Karakaya dams.

The second argument is that, if we accept the causal relationship between the construction of three major dams and the decrease in flow, we should also accept that the subsequent recovery is a result of the completion of the filling of the reservoirs. However, sometimes the river flow recovers to higher levels than previously, as occurred after the completion of Keban and Karakaya reservoirs.

The third argument refers to the fact that the average flow of the 5 years after the construction of the last dam (2006 to 2010) is equivalent to the average flow of the 5 years prior to the construction of the first dam (1970–1974). The 5-year period before the construction of the first big dam, in the city of Keban, in 1975, shows an average flow of 677 m3/s, whereas in the 5-year period after the construction of the last dam, in the city of Kayacik, in 2005, the average flow rate remained at 615 m3/s, which illustrates stability.

The fourth argument against the Inventory conclusions is quite simple: there are periods of low flow followed by recovery when no dams were constructed. This means that there must be other strong factors influencing the river flow, such as precipitation (Figure 9). In fact, the significant positive correlation between streamflow and rainfall in Jarablus station for Spearman's correlation test (for a 0.05 significance level) suggests that the average volume of water is predominantly ruled by natural dynamics, despite human interventions.
Fig. 9.

Average annual precipitation in the region of the middle course of the Euphrates River (Syria) between 1970 and 2007. Source:Statistical Abstracts (1970 to 2007).

Fig. 9.

Average annual precipitation in the region of the middle course of the Euphrates River (Syria) between 1970 and 2007. Source:Statistical Abstracts (1970 to 2007).

Comparing the river flow between Syria (Jarablus) and Iraq (Husaybah) from 2000 to 2010 (Figure 10), which is a period with no dam construction in this stretch, we verify a flow drop of 22.9% on average for the period.
Fig. 10.

Euphrates river flow between Syria (Jarablus) and Iraq (Husaybah) from 2000 to 2010. Sources:Statistical Abstracts (1970 to 2011) and Ministry of Water Resources (Iraq).

Fig. 10.

Euphrates river flow between Syria (Jarablus) and Iraq (Husaybah) from 2000 to 2010. Sources:Statistical Abstracts (1970 to 2011) and Ministry of Water Resources (Iraq).

The Spearman coefficient for correlation between the Jarablus and Husaybah flow test was 0.92. This result shows that, in the annual cycle, the dams along the river do not significantly alter the dynamics of the river downstream.

The fact that there is no dam construction or lake filling during this period is irrefutable evidence that the flow diminishing is due to natural dynamics.

We must also challenge the way the impact of dams is assessed when considering the relationship between the total storage capacity of reservoirs and the total water flow in the period. The Inventory of UN-ESCWA presented several accounts, concluding that, ‘The maximum storage capacity of major dam and reservoirs exceeds the natural annual flow volume of the river (30 BCM) by four to five times’ (UN-ESCWA, 2013, p. 62). The alarmist way that this information is presented leads to the conclusion that Turkey and Syria could even block the Euphrates flow for at least four or five years. Yet, according to the Inventory, only the Ataturk Dam would be able to store the volume of a year of discharge of the Euphrates River: ‘Ataturk is large enough to store the entire annual discharge of the Euphrates’ (p. 63). However, this conclusion is not sustainable because the dams were built over four decades and are all already filled.

Therefore, the Inventory accounts should consider four decades of flow, not just a single year. Considering an average annual flow of 778.5 m3/s this is equivalent to an annual flow of 24.5 km3, or 24.5 BCM. Over 41 years the total water flow is of 1,004.5 km3 or 1,004.5 BCM. The total storage capacity of the top five Turkish dams on the Euphrates River is 90.5 km3, or 90.5 BCM, which represents 9% of the entire flow of the period. This number may seem significant, but it is worth remembering that the dams were not built simultaneously and each one filled once, whereas the river flows continuously (except for short periods when filling large dams). Only large scale water consumption, such as irrigation and water diversions, could significantly diminish the flow of a river such as the Euphrates. Even so, some water used in irrigation infiltrates and returns to the basin. Diversion of water could also be substantial, but also, if the water is used within the basin limits, it would be returned to it through infiltration, although it may be reduced by losses in evaporation. However, such huge diversions would be unlikely because of the political, economic and environmental consequences. The 90 tons of water transported from the Ataturk Dam to the dry southern plains (through the Suruc Tunnel, opened in 2013) will largely infiltrate again into the basin. Thus, the dams do not represent consumption use except by increasing the evaporation surface.

Empirical information

We undertook direct observation of the Euphrates landscape in Syria during three different fieldworks. The first was in January 2010 (Figure 11) in Deir Ez-Zor, Syria, approximately 130 km from the border with Iraq.
Fig. 11.

Photograph of the Euphrates River in the city of Deir Ez-Zor (January 2010).

Fig. 11.

Photograph of the Euphrates River in the city of Deir Ez-Zor (January 2010).

Although this observation occurred prior to the melting snow of Anatolia, when the volume of water reaches its maximum levels, the river level was already quite high. If the level rose to 4 metres, as the literature proved was likely, several riverine occupations would be flooded. Between February and April 2011, two additional fieldworks confirmed the suspicion raised by the first observation: there was no sign of drought, even in the period before the high flow, which starts in April. Comparing the pictures at the same point in 2010 and 2011, we observe exactly the same characteristics of flow and the apparent quality of the water (Figure 12).
Fig. 12.

Pictures of the same stretch of the Euphrates River in Deir Ez-Zor in (a) January 2010 and (b) February 2011. Arrows indicate the fluvial island sighted from the bridge.

Fig. 12.

Pictures of the same stretch of the Euphrates River in Deir Ez-Zor in (a) January 2010 and (b) February 2011. Arrows indicate the fluvial island sighted from the bridge.

Comparing these previous photos with an aerial image dated to 1920 (Figure 13), we can identify a clear similarity despite the different angle and scale of the images.
Fig. 13.

Panoramic photograph of the Euphrates River and the city of Deir Ez-Zor in 1920. The horizontal arrow points to the river island that is observed at the centre of previous photographs. Source:Blanchard (1929, p. 214).

Fig. 13.

Panoramic photograph of the Euphrates River and the city of Deir Ez-Zor in 1920. The horizontal arrow points to the river island that is observed at the centre of previous photographs. Source:Blanchard (1929, p. 214).

Approximately 120 km downstream near the border with Iraq, the landscape did not change and water was still abundant (Figure 14).
Fig. 14.

Euphrates in the town of Abukamal, 10 km from the border with Iraq (February 2011).

Fig. 14.

Euphrates in the town of Abukamal, 10 km from the border with Iraq (February 2011).

At this point, statements by residents indicated that the river remains the same through the time. A man stated to us: ‘I was born and have always lived here. I never noticed anything different on the river. It always rises at the same time … and decreases at the same time.’ When this stretch of the Euphrates River is compared to another more than 500 kilometres upstream at the Turkish border, the landscape is quite similar, despite having three main Syrian dams and three tributaries between these two points (Figure 15).
Fig. 15.

Pictures of the Euphrates River at Jarablus, Syria (border with Turkey), obtained from the same point on the shore. The first (a) is towards upstream and the second (b) is towards downstream (March 2011).

Fig. 15.

Pictures of the Euphrates River at Jarablus, Syria (border with Turkey), obtained from the same point on the shore. The first (a) is towards upstream and the second (b) is towards downstream (March 2011).

Apart from the relief that shows sloped banks and a narrow or absent fluvial plain, all other aspects of the landscape related to vegetation and water remain the same between the Turkish border and the Iraqi border. In the middle way of the borders, the transition from the plain to the sedimentary plateau is bounded by cliffs that are 50 to 100 m high (Figure 16). This landscape was described by Dudley Stamp as follows: ‘The borders are hard, gravelly plains with patches of sand and there is frequently a scarp of 50 to 100 feet in height which marks clearly the commencement of the Euphrates-Tigris plains’ (Stamp, 1959, p. 142).
Fig. 16.

Relief of the basin of the Euphrates River in the middle course (February 2011).

Fig. 16.

Relief of the basin of the Euphrates River in the middle course (February 2011).

From one border to another, the plains alongside the Euphrates are largely cultivated with crops under traditional irrigation systems, whereas the plateau shows arid features that strengthen the contrasts within the landscape. These features confirm that most water used in irrigation remains in the river plain.

Water quality

The analysis of water quality from the Euphrates River relied on field observation and laboratorial data, as detailed below. Considering that there are human interventions along the entire river, from its formation until the confluence with the Tigris River, we could not find an immaculate stretch to be compared with a contaminated one. For this reason, we chose the Syrian stretch through which we could reveal the quality of water coming from Turkey and flowing towards Iraq.

The first field observation of the river was at the city of Deir Ez-Zor in January 2010. On that occasion, in full sunlight, the colour of the water appeared to have emerald-green tones (Figure 17), which could be the reason that this city is known as the Emerald of the Euphrates.
Fig. 17.

Emerald-green colouration of the waters of the Euphrates (January 2010).

Fig. 17.

Emerald-green colouration of the waters of the Euphrates (January 2010).

During the second and third fieldworks, we sought out natural and man-made aspects of the landscape from Jarablus to Abukamal that could indicate, preliminarily, good quality of the water along the basin. First (February 2011), we travelled from Deir Ez-Zor to Abukamal stopping at different points along the river to observe the landscape. Between these cities, the primary land use is agriculture undertaken by traditional systems of plantation and irrigation, which is not mechanised to a significant degree. Small towns and villages fringe the river bank. We expected to find a polluted river in Abukamal since it was the last Syrian point downstream. In addition, we were in the low flow season meaning the pollutants could be concentrated by the lower quantity of water. However, we found clear water, similar to that we would find hundreds of kilometres upstream at Jarablus. The samples we collected were limpid, clear and odourless. We also observed pumps withdrawing water for household purposes without any treatment. The absence of industrial activities may explain the apparently good quality of water, which was confirmed later by analysis of the samples. Even in larger cities such as Deir Ez-Zor (212,000 inhabitants, according to the 2004 census), we did not find any evidence of pollution. Conversely, people used the water for different purposes, such as fishing and swimming, which indicates that any water pollution remains at a tolerable level. Other features indicate good quality of water, such as the abundant presence of aquatic animals, including fish and ducks (Figure 18).
Fig. 18.

Many ducks inhabit the Euphrates River in Deir Ez-Zor (February 2011).

Fig. 18.

Many ducks inhabit the Euphrates River in Deir Ez-Zor (February 2011).

In certain backwaters, however, we observed an accumulation of urban waste, as shown in Figure 19.
Fig. 19.

Plastic trash accumulated in small backwaters in the city of Deir Ez-Zor. The same point viewed from different angles and dates. (a) January 2010. There is a nora, or hydraulic apparatus, used to withdraw water from the river. (b) February 2011.

Fig. 19.

Plastic trash accumulated in small backwaters in the city of Deir Ez-Zor. The same point viewed from different angles and dates. (a) January 2010. There is a nora, or hydraulic apparatus, used to withdraw water from the river. (b) February 2011.

Then, we travelled towards upstream to the city of Ar-Raqah, reaching the Al Baath dam and, further, the big Al-Assad dam. Apart from the dam, the landscape along the river did not change significantly.

During the third field trip to Jarablus in March 2011, the water of the Euphrates was very clear, odourless and apparently acceptable for drinking. This impression was reinforced by observing families having supper along the river and swimming. We were also encouraged by locals to drink the water at that point, and fortunately, the quality was later confirmed by the laboratorial analysis.

It is probable that there are discharges of sewage from occupation along the river, but we do not know the amount of pollutants that these discharges contribute. In the entire Syrian stretch of the Euphrates River, the water was apparently of good quality. Irrigation channels are also common throughout the Syrian stretch, as shown in Figure 20.
Fig. 20.

Water pumping and redistribution by channels for traditional agriculture activities (March 2011).

Fig. 20.

Water pumping and redistribution by channels for traditional agriculture activities (March 2011).

Laboratorial data

We collected samples in Jarablus and Abukamal, which were analysed by a Brazilian laboratory, according to Ordinance n. 518 of the Brazilian Ministry of Health. The appraisals of both samples indicated that the water was within parameters established by the ordinance, based on international standards.

Among the 24 parameters analysed in the original report, 12 showed the same level for both samples, four were minimally more favourable for the downstream Abukamal sample and eight were minimally more favourable for the sample of Jarablus. None of the parameters received a negative mark according to the ordinance, which indicates that the water in natura from the Euphrates River at that time and in both locations was of suitable quality for drinking. In broader terms, we can affirm that the quality of water from the Euphrates River in Turkey entering Syria at Jarablus is the same as that leaving Syria for Iraq at Abukamal.

Some of the climatic features previously mentioned, such as the gradual decrease in rainfall and increased evaporation rate, have been confirmed by the laboratory analysis, specifically by the conductivity parameter (Table 2).

Table 2.

Analysis of conductivity and TDS of Jarablus and Abukamal samples.

Analysis results Jarablus sample Abukamal sample 
Conductivity 302.0 651.0 
TDS (total dissolved solids) 146.0 329.0 
Analysis results Jarablus sample Abukamal sample 
Conductivity 302.0 651.0 
TDS (total dissolved solids) 146.0 329.0 

Source:RR Acqua Service (Analysis n. 132-6/2011 and n.131-6/2011, respectively).

From Jarablus to Abukamal, conductivity and total hardness parameters increased 349 and 183 points, respectively. These parameters are related to the increasing concentration of salts towards downstream. This occurs due to a gradual higher evaporation and lower precipitation, which corroborates the increasing aridity of the landscape towards downstream. The following graph (Figure 21) also confirms this natural trend in salinity.
Fig. 21.

Gradation of TDS (total dissolved solids) between the Ataturk Dam (Turkey) and the city of Nasiryah (Iraq). Source:UN-ESCWA (2013).

Fig. 21.

Gradation of TDS (total dissolved solids) between the Ataturk Dam (Turkey) and the city of Nasiryah (Iraq). Source:UN-ESCWA (2013).

The Inventory of UN-ESCWA (2013) states that the increasing salinity is caused by human activities such as fertilised irrigation and sewage disposal along the river (without specifying quantities). It attests that salinity ‘is probably the result of upstream pollution from Turkish irrigation projects, and Syrian agricultural activities in the flood plains; of the Euphrates River’ (p. 67, emphasis added).

However, the most used fertilisers in the basin are predominantly based on nitrogen, which has a short life span and loses its chemical characteristics after a few metres and a few days of infiltration. Sewage that may reach the river is obviously a source of pollution, although its intensity and pollution potentiality remain unknown. However, there are no major cities or industrial activities along the Euphrates River. The largest cities are Ar-Raqah (approximately 220,000 inhabitants), Deir Ez-Zor (approximately 212,000 inhabitants) and Abukamal (approximately 43,000 inhabitants), which are distributed along more than 300 km.

The Inventory also mentions the gypsum concentration of Syrian soil helping in the mobilisation of salts: ‘In Syria, the Euphrates flows through areas with gypsiferous soils, which have a high potential for salt mobilisation and thus contribute to further salinisation’ (p. 67). Once again, the causal relationship between the variables is unconfirmed. Even so, the Inventory uses conclusive assertions, such as in the statement ‘salinity often results from directly agricultural activities […]’ (p. 68). It also associates salinity to dams as follows:

The large peak in TDS values observed at Hussaybah in the period 1989–1993 coincides with reduced flow of the Euphrates entering Iraq, possibly due to the filling of upstream reservoirs, such as the Baath Dam (1987) and the much larger Ataturk Dam (1990)’ (pp. 68–69, emphasis added).

In the years 1988 and 1989, which are subsequent to the construction of the mentioned Baath dam, the Euphrates River presented the greatest flow of the 41-year period in Syria, reaching 1,590 m3/s. Therefore, the assertion that relates this dam with low flow and high salinity is not true.

The increasing salinity results from the increasing temperature and evaporation, associated with the decreased rainfall and absence of tributaries, resulting in a higher water conductivity. Some authors have been reinforcing the natural tendency to salinity in this region. Abdelfattah (2013) studying soils in the Middle East attests that ‘the slow accumulation of atmospheric inputs in arid areas may lead to net mass gains of salts’ (p. 280); in arid regions, the lack of water results in weak dissolution and leaching of the materials in the soil. As a consequence, ‘climate characteristics influence pedogenesis through accumulation of materials such as calcium carbonate, gypsum and salts’ (Abdelfattah, 2013, p. 286).

Decisive evidence in favour of this natural tendency is provided by the salinity rates prior to the construction of any dams or large irrigation projects. In 1971, the index of salinity at Tabqa (near Ar-Raqah city) was 333 mg/L, whereas at Deir Ez-Zor (approximately 180 kilometres downstream), it was 413 mg/L; in 2010, these figures were 277 and 441 mg/L, respectively. Because the levels of salinity were quite similar before and after dam construction, the argument that salinity results from human activities is unlikely to hold true. Moreover, the trend line of Figure 21 reflects a continuous phenomenon with a gradual increase of salinity. If the salinity were directly related to dams, then the trend line should show some irregularity. After all human interventions, samples of Jarablus and Abukamal TDS showed rates of 146 and 329 mg/L respectively, noting that these two cities are more than 400 kilometres apart.

Conclusions

The analysis of water quantity of the Euphrates River over 41 years shows that the flow is a stationary variable that remains at an approximate average of 778.5 m3/s despite all dams constructed within this period. Within the 41-year period, data series show peaks and valleys not always coinciding with dam constructions (low correlation) but highly coincident to precipitation rates (high correlation).

Even so, in general terms we could observe a gradual and regular decreasing of flow towards downstream. Again, this is not due to the dams, since, despite their dimensions, they all have a limit of storage and do not constitute a consumption use, whereas the seasonal feeding of the river particularly by the melting snow, is constant. Therefore, we related this slight diminishing flow to climatic factors and the gradual decrease and ultimately the complete absence of tributaries, apart from the scattering of waters through various downstream channels. These natural features show a tendency to aridness, which is accentuated by some human factors. As the end of the flood season is followed by a dry summer, the demand for irrigation increases towards downstream in Iraq, when the river has less water.

All these factors led us to conclude that changes in flow are more closely related to natural features than to human interventions.

Concerning the water quality, the laboratory data showed that the quality of the water that enters Syria from Turkey is similar to the water that flows from Syria to Iraq, always with a good level of drinkability apart from an increasing salinity. The salinity, in turn, increases due to a higher evaporation tax associated with a decreasing precipitation and perhaps, due to the occurrence of gypsum associated with clay. The natural salinity is confirmed by data prior to dam construction and irrigation projects. The absence of industrial cities along the Euphrates River and the predominance of traditional agricultural activities contribute to the maintenance of the water quality and negate the relation between salinity and human activities. As a consequence, riparian communities make direct use of the water, as shown by the fieldworks.

Recognising the natural dynamics and trends associated with landscape-scale processes is vital to reduce occasional tensions between riparian countries. Recognising that Turkey assures an accorded flow to Syria, and this country assures the flow to Iraq, inclusively maintaining the quality of water is of paramount importance for the region. This context opens a path contrary to the idea of a water war and can create opportunities for common and efficient management.

References

References
Blanchard
R.
(
1929
).
Asie Occidentale
. In:
Géographie Universelle
.
La Blache
P. V.
(ed.).
Librairie Armand Colin
,
Paris
.
Chaieb
F.
(ed.) (
1995
).
Syria
.
Direction Générale de ĺInformation
,
Damascus
.
Climatic Atlas of Syria
(
1997
).
Ministry of Defence, Damascus
.
Gourou
P.
(
1953
).
L'Asie
.
Hachette
,
Paris
.
Kahya
E.
Kalaycu
S.
(
2004
).
Trend analysis of streamflow in Turkey
.
Journal of Hydrology
289
,
128
144
.
Lettenmaier
D. P.
Wood
E. F.
Wallis
J. R.
(
1994
).
Hydro-climatological trends in the continental United States, 1948–88
.
Journal of Climate
4
,
586
607
.
Moraes
J. M.
Pellegrino
G. Q.
Ballester
M. V.
Martinelli
L. A.
Victoria
R. L.
Krusche
A. V.
(
1998
).
Trends in hydrological parameters of a southern Brazilian watershed and its relation to human induced changes
.
Water Resources Management
4
,
295
311
.
Nerc (Natural Environment Research Council)
(
1975
).
Flood Studies Report 1
,
London
.
Pettitt
A. N.
(
1979
).
A non-parametric approach to the change-point problem
.
Applied Statistics
28
,
126
135
.
Shainberg
I.
Levy
G. J.
(
2005
).
Salination Processes
.
The Volcani Center
,
Bet Daganm
.
Stamp
S. L. D.
(
1959
).
Asia: A Regional and Economic Geography
.
Methuen & Co
,
London
.
Statistical Abstracts
(
1970-2011
).
Central Bureau of Statistics, Damascus
)
.
UN-ESCWA & BGR (United Nations Economic and Social Commission for Western Asia; BundesanstaltfürGeowissenschaften und Rohstoffe)
(
2013
).
Inventory of Shared Water Resources in Western Asia, Beirut
.