Water quality and discharge of the Lower Jordan River

Journal of Hydrology 527 (2015) 1096–1105

Contents lists available at ScienceDirect

Journal of Hydrology journal homepage: www.elsevier.com/locate/jhydrol

Water quality and discharge of the Lower Jordan River Noa Hillel a, Stefan Geyer b, Tobias Licha c, Saed Khayat b, Jonathan B. Laronne a,⇑, Christian Siebert b a

Ben-Gurion Univ. of the Negev, Israel Helmholtz Center for Environmental Research – UFZ, Germany c Georg-August-University Göttingen, Germany b

a r t i c l e

i n f o

Article history: Received 3 November 2014 Received in revised form 24 May 2015 Accepted 1 June 2015 Available online 6 June 2015 This manuscript was handled by Laurent Charlet, Editor-in-Chief, with the assistance of Pedro J. Depetris, Associate Editor Keywords: Lower Jordan River Water quality Salinity Isotopes Water discharge

s u m m a r y The fresh surface water of the Lower Jordan River (LJR) has been limited in the past several decades due to damming of its main tributaries, which reduced the annual flow by 90%, leaving a mixed flow of polluted and saline sources. A monitoring and sampling hydrometric station was installed on the southern LJR to track the temporal variations of its discharge (Q) and hydrochemistry. In addition to manual water sampling, the station includes an automatic water sampler and cellular transmitting pressure and EC sensors, allowing real time observation. All samples were analyzed for major ions (Na+, Ca2+, Mg2+, K+, Cl, SO2 4 ,  34 18 Osulfate, 15Nnitrate, 18Onitrate, NO 3 , Br ) and several samples were analyzed for selected isotopes ( Ssulfate, 2 Hwater, 18Owater) as tracers. A general inverse seasonal trend was found between EC and water level although extreme values relate to flood events during the wet period. High values of EC (up to 40.3 mS/cm), high concentration of major ions, and flood events characterized by clockwise EC–Q hysteretic relations likely relate to the dissolution of precipitated salts in the basin. Isotope analyses reveal lithology and sewage as the respective major contributors of salinity; they were used to identify events unrelated to runoff (i.e., to precipitation in the area). The continuous monitoring is an essential tool for understanding long term changes of such a dynamic system but is critical for identifying extreme events occurring rarely and rapidly, possibly having a drastic effect on fauna and flora. Ó 2015 Elsevier B.V. All rights reserved.

1. Introduction Water quality in many rivers worldwide has degraded due to anthropogenic activities. Municipal and industrial polluted wastewater, salinization and the control of water regime demonstrate some of the aspects damaging water quality. For example, saline agricultural return flows caused salinization in the Amu-Daria River in central Asia (Crosa et al., 2006), geomorphic changes caused by damming of the McKenzie river resulted in the reduction of the salmon population (Ligon et al., 1995), and the Rhine River became known as ‘‘the sewer of Europe’’ by the end of 1960s due to former severe industrial and municipal pollution, causing most flora and fauna to disappear (Wieriks and Schulte-Wulwer-Leidig, 1997). The Lower Jordan River (LJR, Fig. 1a) runs through the Jordan Dead Sea Rift from the Sea of Galilee to the Dead Sea along an aerial distance of 105 km (220 km along its meanders; Nir, 1989) and is incised in two Pleistocene floodplain units. The current day floodplain of the former river Zor incised into the Samra formation (marl, sand, conglomerate, limestone and chalk), while the higher, ⇑ Corresponding author. E-mail address: [email protected] (J.B. Laronne). http://dx.doi.org/10.1016/j.jhydrol.2015.06.002 0022-1694/Ó 2015 Elsevier B.V. All rights reserved.

ancient floodplain termed Ghor consists of the Lake Lisan formation (aragonite, marl, detrital sediments and gypsum; Nir, 1989; Farber et al., 2004). The LJR is located in a (semi-) arid region with high evapotranspiration (Farber, 2005). The two major sources of the LJR, the Sea of Galilee and the Yarmouk River, were dammed during the 1960s to meet the water needs of the region’s growing population. Accordingly, flow volumes decreased dramatically, causing changes in the river’s physical and ecological characteristics, as well as in its water quality (Calvo and Ben-Zvi, 2005). Current water sources of the LJR – mainly diverted saline springs, agricultural return flows, partially treated waste water, discharge from fish ponds, ground water and storm water – contribute to varying salinity and pollution. Nevertheless, only few publications have addressed this issue, as most of the LJR is a trans-boundary river, sensitive politically and difficult to approach physically due to remaining land-mines along parts of its course (Knesset site, 2013). Farber et al. (2004) defined three distinct segments (northern, middle and southern) based on salinity and sources. Salinization trends could not be explained by a sole source; the groundwater component in the northern section contributes the largest volume of water, while in the southern section groundwater contributes a higher solute concentration (Farber et al., 2004). A drastic decrease in water discharge increased both

N. Hillel et al. / Journal of Hydrology 527 (2015) 1096–1105

1097

Fig. 1. (a) Map of the Lower Jordan River (LJR). (b) Qasser Al-Yahud.

the salinity and the organic pollution and the loss of natural habitats together with a 50% decrease in biodiversity (Gafny et al., 2010). The water appendix in the Jordan–Israel peace agreement (26.10.1994) states the commitment of the parties to monitor water quality along the border within 3 years from the agreement. In addition, wastewater and saline water are to be treated before entering the LJR within 4 years. An initial monitoring scheme operated by state authorities began in July 2012 with two new hydrometric stations. A treatment plant aiming to treat waste and saline water has been constructed and is expected to start operating during 2015. Rehabilitation efforts were initiated in May 2013 with a daily allotment of 24,000 m3 of treated water entering the LJR from the Sea of Galilee (Israel Water Authority, 2013). The Jordan River is the largest continuous flowing source of water to the Dead Sea and its discharge is vital for the water balance of the dwindling Dead Sea (Calvo and Ben-Zvi, 2005). In addition, the LJR is intensely used for agriculture (Segal-Rozenhaimer et al., 2004). Apart from hydrologic and economic aspects, the Jordan River is of high importance to religious groups baptizing in its waters. The objectives of this research are (1) to address the causes of annual water quality changes by understanding the chemical composition of the water, using continuous water monitoring and high-frequency sampling combined with geochemical tools and (2) to quantify the relevant and temporally varying water discharge. Relations between water quality parameters should enable to further understand the characteristic behavior of the

LJR at base flow and also in rain-fed and anthropogenically induced runoff events. 2. Methods A hydrometric station was installed in February 2010 at the ‘‘Qasser al-Yahud’’ baptism site, 8 km north of the Dead Sea (Fig. 1a and b), including an automatic water sampler (Andress & Hauser Liquiport 2000), independent pressure sensors (Waterpilot FMX21, Schlumberger), EC-sensor (Condumax CLS12) and a turbidity sensor (TurbiMax WCUS41). High correlation (r2 = 0.93) between data of the atmospherically corrected MicroDivers and the Waterpilot enabled interpolation of water level data. Information on precipitation events were obtained from the database of the Israel Water Authority internet site. 2.1. Water discharge Water discharge (Q) was calculated using the continuity equation Q ¼ UA, where A is the cross sectional area (m2) and U is the mean water velocity (m/s). Velocity was measured by two non-contact methods using a Decatur Electronics portable surface velocity radar (SVR) with a 0.3–9.1 m/s range and 5% accuracy, and separately also by floats. As both surface velocity methods are not equivalent to average water velocity, results were multiplied by a coefficient (0.85) for SVR measurements (Yorozuya et al., 2010) and a lower coefficient (0.64) for floats (Marjang and Merkley,

1098

N. Hillel et al. / Journal of Hydrology 527 (2015) 1096–1105

2009). The cross section was measured by a Sokkia SET5X total station and the cross sectional area was calculated by Eq. (1):

A¼

n X

ai

ð1Þ

i¼1

ai representing one trapezoidal area in the cross section (Eq. 2):

ai ¼

  h1 þ h2  dh1 ;h2 2

ð2Þ

where h1 and h2 represent the length of each of the trapeze perpendiculars and d is the distance between them. 2.2. Sampling Altogether 428 water samples representing the time period 28/8/2010–20/8/2012 were collected manually in 500 ml PE double capped bottles. For each sample, 5 ml of water were filtered through a 0.22 lm filter (PVDF) into a PP vial for major ion analysis. Onsite measurements of EC and temperature were done using a Lutron EC meter. 2.3. Chemical analyses 2 Anions (Cl, Br, NO 3 , SO4 ) were analyzed by IC Dionex + + DX-320 and cations (Na , K , Mg2+, Ca2+) were analyzed applying IC Dionex DX-500 both with conductometric detection. Prior to IC, samples were diluted according to the linear working range. The analytical error was estimated to be 8%. Isotope analyses were performed on 40 samples: 10 random summer samples, 10 random winter samples and five samples from four flood events. For analyses of 34S and 18O on sulfate, 30 ml of sample were filtered using 0.45 lm CA filter, acidified with 6 N HCl, heated to 100 °C and 5 ml BaCl2 were added. The precipitated BaSO4 was removed by wet filtering (0.45 lm CA membrane), dried at 95 °C, burned at 600 °C, homogenized and weighed. For 34S analysis, 350–370 lg were placed in a zinc cup with V2O5; for 18O, 400–450 lg were placed in a silver cup with NiCO3 and analyzed using a Euro-EA elemental analyser. 34S and 18 O values have a respective analytical precision of 0.4‰ and 0.6‰, and are reported relative to V-CDT and V-SMOW, respectively. Nitrate isotope (15N, 18O) analyses were realized using the denitrifier method (Casciotti et al., 2002) with an analytical precision of 0.4‰ and 1.6‰ for 15N and 18O, respectively. 15N values are reported relative to air. Water isotope analyses were performed using a Piccaro cavity ringdown spectrometer with a respective analytical precision of 1.5‰ and 0.4‰ for 2H and 18O. Stable isotopes of water are reported relative to V-SMOW.

3. Results and discussion 3.1. EC and discharge relations During the 2-year observation period the LJR water level fluctuated with distinct events, which led to abrupt changes in stage (Fig. 2 and Table 1a). Water level rise due to precipitation is clearly recognized. Nevertheless, the reasons for observable level rises unrelated to precipitation are difficult to identify, and are at times unknown (Table 1b). The average electrical conductivity of the LJR is 12 mS/cm, at least one order of magnitude higher than values typical for freshwater rivers (10–1000 lS/cm; Chapman, 1996). Dry periods are characterized by high solute concentrations, which commonly decline during wet periods as a result of dilution (Knighton, 1984; Glover and Johnson, 1974). Although this description is generally valid, the largest EC values occurred during the wet season (Table 2), with

maximum concentrations likely due to solute accumulation on the ground surface caused by weathering, percolation and evaporation during the dry season and between events, thereafter transported to the stream by runoff events (Choudhari and Sharma, 1984). Generally, the 2010–2011 hydrologic year was drier and with only few and small water level rises compared to the following year. Seasonality in EC (Fig. 2) is recognizable by higher average values during the dry period (16.0 and 14.4 mS/cm for 2011 and 2012, respectively) than during the wet period (9.7 and 10.3 mS/cm for 2010/2011 and 2011/2012, respectively). In most cases, the temporal variation of EC was related to variations in water level: they either varied sympathetically (correlate positively) or opposite. Rises in water level preceding the December-14 2010 event were not caused by flow events. In the first major event (1) of the wet period water level rose by 48 cm and dropping EC mirrored the water level trend. Since that rain event was the first considerable event of the year, higher EC values were expected as the first runoff event of the season tends to have higher TDS due to flushing of salts accumulated on the surface by capillary rise and evaporation. Events 2 and 3 showed increased water levels by 55 and 28 cm, respectively. In both cases, EC maxima were recorded a day before the discharge peaks, possibly representing initial solute flushing. Event 4 (May 13–14 2011; 11.7 cm rise) occurred after a long dry period and characterizes the transition season between March/April and the end of June when rain occurs only sporadically. An extreme case of flushing is demonstrated; this precipitation event generated an increase in EC with a relatively low increase in water level. The high increase in EC implies considerable flushing of a very dry drainage basin with surface runoff (Alexandrov et al., 2003). Event 5 represents several EC peaks that occur during the dry period and cannot be explained by rain events. Remaining sources include wastewater, groundwater, saline springs and agricultural return flows. Since correlative water table data are missing for the period of event 5, it is difficult to evaluate which of the mentioned sources of salinty could have caused the observed EC-fluctuations. A larger number of significant precipitation events occurred in the 2011–2012 hydrologic year. Water level rises unrelated to precipitation were also recorded. The first event of the wet period (6; 106 cm water rise) after a long dry period presents a strong flushing effect (24.6 mS/cm). A 64 cm gradual increase in water level (7) was generated by three small rainfall events and by fishpond releases (pers. comm. Amitai Geva, Ministry of Agriculture, 2012). Two consecutive discharge peaks are characterized by EC increase during the first peak and an EC decrease for the second peak as a result of former dilution. The following event (8) was unrelated to precipitation; indeed, December 2011 was recalled to be the driest since 1993 (Israel Water Authority, 2013). This event began on 30/12/2011 and lasted 21 days while maintaining a water level approximately one m higher than common for this period. Displaying an initial very high electrical conductivity (40.3 mS/cm), this event had overall values >21 mS/cm, whereas in the rest of the wet period, average EC was 10 mS/cm. A release from the Karama Dam is thought to be the cause of this event. The Karama reservoir was constructed in Wadi Mallaha in 1995 for agricultural uses. The reservoir, with a 55  106 m3 capacity, lies on the Lisan formation having a 6% halite content. Additional solutes contributing to the high salinity reach the reservoir include water from the King Abdulla Canal, agricultural return flow, saline springs, and from leaching of thick salt crusts covering the bed and banks of Wadi Mallaha. In the Karama Dam, EC ranges from 6 to 35.5 mS/cm with an average of 20 mS/cm. Due to its high salinity, the water cannot be used for the intended purpose of irrigation and is occasionally discharged to the Jordan River (Salameh, 2004).

1099

N. Hillel et al. / Journal of Hydrology 527 (2015) 1096–1105

Jordan River stage at bapsm site 26.8.10-20.8.12 8/10/2010 500

11/18/2010

2/26/2011

6/6/2011

9/14/2011

12/23/2011

4/1/2012

7/10/2012 45

stage gaps EC precipitaon event

450

40 35

8

400

4

350

6

5

25

7

1 300

20

2

11 10

3 250

EC (mS/cm)

stage (cm)

30

15

9 10

200

5

150 8/10/2010

11/18/2010

2/26/2011

6/6/2011

9/14/2011

12/23/2011

4/1/2012

7/10/2012

0

date Fig. 2. Water level, conductivity and dates of precipitation for the period 28/8/10–20/8/2012. Thick black lines represent gaps filled by ‘‘Diver’’ data. The width of precipitation relates to the length of the event (days); only precipitation events mentioned in this paper are shown. Numbers refer to peaks in water level discussed in the text.

Table 1a Major events of water level rise during the period of study (26/8/2010–20/8/2012). No.

1 2 3 4 5 6 7 8 9 10 11

Initiation of rise

14/12/2010 12:16 29/01/2011 17:31 07/02/2011 17:18 14/05/2011 14:00 11/06/2011 08:00 25/09/2011 18:00 02/11/2011 22:00 30/12/2011 12:00 01/03/2012 03:08 06/03/2012 11:11 22/03/2012 18:00

Level (cm)

Peak

265

19/12/2010 12:49 01/02/2011 21:31 11/02/2011 02:33 15/05/2011 04:00 21/08/2011 06:00 26/09/2011 04:00 28/11/2011 03:04 04/01/2012 19:20 02/03/2012 05:09 11/03/2012 06:00 25/03/2012 14:00

234 240 227 n.a. 231 266 294 321 360 344

Total rise (cm)

Level (cm)

EC at peak (mS/cm)

48.2

314

6.13

55.2

289

9.58

28.1

268

8.26

11.7

239

7.88

n.a.

n.a.

19.6

105

336

24.6

63.6

329

99.8

393

157

479

2.25

85.0

445

3.24

22.6

367

4.34

EC values of events 9, 10, 11 mirror the trend in the very high water level demonstrating dilution (Alexandrov et al., 2003). Event 9 (158 cm in 26 h) represents the largest precipitation event of the season with the largest water level rise during the documented 2-year period. The following water level rises (10, 11) were unrelated to precipitation. The event represented by (10) started five days after the last rain event and ended two days before the next rain event, while event (11) commenced six days after the last rain event. Having very low EC values, these two events suggest the release of fresh water. Rapid changes in values of stream water EC occur during flow events and can be described by EC–Q relations (Fig. 3). Nine of the 15 presented relations are clockwise or semi-clockwise

6.65 25.4

End of recession

23/12/2010 08:28 07/02/2011 11:33 12/02/2011 16:25 19/05/2011 08:00 n.a. 27/09/2011 20:00 28/12/2011 18:00 23/01/2012 04:00 06/03/2012 07:11 15/03/2012 22:00 27/03/2012 18:00

Duration

Cause/comment

Rise h

Fall

120

91.6

Runoff event

76.0

134

Runoff event

81.2

37.9

Runoff event

14.0

100

Runoff event

94.0

n.a.

Dry period

10.0

40.0

Runoff event

605

734

127

440

Runoff event, fishponds? Karama dam?

26.0

98.0

Runoff event

114

112

dam?

68.0

52.0

Unknown

hystereses, each event showing a slightly different behavior. This relation occurs when an EC peak precedes the peak discharge, EC declines with an increase in discharge, and thereafter increases. This response results from the dissolution of soluble deposits on the surface when high solute concentrations occur during the first flush (Choudhari and Sharma, 1984). After the first flush, EC values decrease as water discharge rises. Groundwater reaching the surface as springs and enriched with solutes causes an EC increase at the end of a flow event with decreasing water discharge. Additionally, an EC increase can also result from the dissolution of transported suspended sediment, mostly during the beginning of an event when suspended sediment concentrations are high (Laronne and Shen, 1982). Although the 13–23/12/2010 event is

1100

N. Hillel et al. / Journal of Hydrology 527 (2015) 1096–1105

Table 1b Potential sources and their effect on level and EC. Potential source

Expected level change

Expected EC change

SDC (constant inflow) Runoff after dry period Consecutive runoff event Sewage (constant inflow) Sewage (pulses) Saline springs

No change Increase Increase No change Slight increase Very slight increase

Fresh water reservoir discharge Saline water reservoir discharge Fishpond release Dry season, unknown sources

Increase

No change Increase Decrease No change Slight increase Potentially high increase Decrease

Increase Slight increase Decrease

Potentially high increase Slight increase Increase

the first major event of the wet period and displays a clockwise hysteresis, it is quite unusual due to the initially observed decrease in EC with Q, suggesting dilution. Three among the four anti-clockwise EC–Q hysteresis events occurred in proximity after a previous event and started with a decrease in EC as Q increased. Consecutive events have lower EC values and the relations are anti-clockwise. These relations occur when a minimum in EC precedes peak discharge, thereafter increasing during a recession (Alexandrov et al., 2003). This response can be ascribed to the removal of solutes by the initial flushing of the previous event (Choudhari and Sharma, 1984). Nevertheless, an exceptional anti-clockwise hysteresis was documented on September 25–28 2011. Unlike the other instances, this first event of the season demonstrates an increase in both EC and Q followed by a slower decrease in EC relative to that in Q, shaping the anti-clockwise pattern. The reason for this unique relation is the high solute concentration after a long dry period. A linear EC–Q relation was observable between May 11 and May 19 2011, when a small change in Q caused a sharp EC increase reflecting salt accumulation during the long dry period prior to the event. An ‘‘8’’ hysteresis in the EC–Q relationship occurred in the February 01–16 2012 event. The initial increase in Q is accommodated by a slight EC increase, followed by EC increasing as Q decreases. A slight second discharge increase included a sharp drop in EC, which rises again with decreasing discharge, closing the ‘‘8’’.

3.2. Major ions Major ion concentrations in the LJR (Table 3) are very elevated in comparison with world averages for rivers and the range in their magnitudes is particularly wide. Maximum concentrations are often 20–40 times higher than the minima of long term monitoring (Hem, 1985). Changes in ion concentration occurred between seasons and during changing discharge. Particularly in arid areas, evaporation and evapotranspiration are very high, causing increased concentration during summer, but cannot explain the total salinity increase (Chapman, 1996; Farber, 2005). In fact, maximum concentrations occurred in the wet season. The process of salt accumulation and its flushing by surface runoff is typical of

Table 2 Seasonal statistics of electrical conductivity of the LJR water. Season

Dry Wet

EC, mS/cm min

max

average

stdv

8.1 2.2

25.0 40.3

15.5 10.0

3.0 4.7

arid lands, accordingly affecting water salinity (Choudhari and Sharma, 1984). Chloride concentrations are the highest: the range is very wide (451–13,408 mg/l), with 80% between 2172 and 6093 mg/l, on average 3658 mg/l. Minimum and maximum values are related to unusual events. Although median values are higher in the dry season for all ions, maximum ionic concentrations are attained in the wet season. The proximity between average and median values of ionic ratios implies a normal distribution, whereas the distance between the percentiles to the minimum and maximum implies a narrow range of values. Hence, extreme events are uncommon, as is expected for a 2-year record. Most ion concentrations follow a common trend, suggesting a common source. High correlations occur between all ions and chloride excluding nitrate (Figs. 4 and 5a), implying an independent source of the latter. Data for major ions are compared (Table 4) with similar data collected during 1999–2005 (Farber, 2005). The presented database is by far the largest for the LJR. Comparison of the two databases displays considerable differences between average values (particularly Na, Cl and d18O and maximum values). While Farber (2005) documented a maximum Cl concentration of 4672 mg/l, this study revealed peaks up to 13,408 mg/l Cl. Excluding potassium, the maximum values of our data are 2–3 times higher than previously observed. From the above, it is concluded that continuous monitoring is crucial to understand the long-term quality status of a river, particularly during extreme events, which occur infrequently but may have a drastic effect on fauna and flora. 3.3. Ion ratios Ionic ratios reveal information about processes in water bodies more visibly than concentrations (Siebert et al., 2014) and are given in the following as molar ratios. Rain in Israel has a Na/Cl molar ratio of 0.86–1.0 (Vengosh and Rosenthal, 1993) whereas the average LJR ratio is 0.791 (the range is 0.674–0.911), with 80% of all results in the 0.752–0.826 range (no relation between Na/Cl and Q). This means that the LJR has sources with a lower Na/Cl ratio than that of rain. The ratio in major saline spring water diverted to the LJR is 0.580–0.727 (Bergelson et al., 1999). The Na/Cl-ratio of the most upstream input to the LJR, consisting of water from the saline drainage carrier (SDC, diverting saline springs from the Sea of Galilee) and sewage range from 0.655 to 0.724 and its influence may be represented mainly in dry periods when sources are fewer (Fig. 5a). In the section of the Yarmouk below the diversion of most of its water to the King Abdulla Canal the Na/Cl ration is 0.78–0.96. For fishponds the ratio is 0.68–0.77. Stream water inflowing into the LJR from the West and from the East have respective ratios of 0.68–0.75 and 1.0 (Farber, 2005). Although the values of the different sources are close to the average value found in this study, as far as 5 km upstream of the Baptism Site a 30 km reach has a ratio < 0.75, and only in the lowest 10 km of the river the ratio is Na/Cl > 0.75 (Farber, 2005). This implies a highly saline source in the lower section of the river, most likely groundwater with a high chloride concentration (Farber, 2005). Very high values of Na/Cl result from precipitation events. The highest value, 0.911, is in the typical range of rainwater and relates to a dam opening due to heavy rainfall. The lowest value, 0.674, is the outcome of a ‘‘first rain of the year’’ event. Although both values are the result of rainfall, the first represents the fingerprint of the rain whereas the second is affected by other causes: solutes flushed after a long dry period. The K/Cl ratio of fresh ground water is similar to the value in sea water (0.018; Vengosh and Rosenthal, 1993), whereas the average ratio in this study is considerably higher – 0.028. Discharge and chloride concentrations do not explain changes in the K/Cl ratio (r2 = 0.15, p < 0.01; r2 = 0.12, p < 0.01 respectively). Optional

1101

N. Hillel et al. / Journal of Hydrology 527 (2015) 1096–1105

9

13-23.12.10

9.2 9.0 8.8 8.6 8.4 8.2 8.0 2.0

(c)

EC (mS/cm)

8 7 6 5

3

4

5

discharge

7

3.5

4.0

0

1

2

3

EC (mS/cm)

4

5

5 1.0

6

(m3/sec)

29.12.11-24.1.12

(e)

8

10

6 7

8

9

10 11 12 13

4

24-28.2.12

5

9

5

8

4

7 5.0

7.5

8.0

8.5

9.0

(o)

22-28.3.12

7

1

8

9

8.8 2.4

2.6

3.0

3.2

1-16.2.12

(j)

6 5.5

6.0

6.5

7.0

4

5

6

7-14.3.12

(n)

8

2

0 10

12

discharge

14

(m3/sec)

7

8

9

10

11

12

discharge (m3/sec)

2

discharge

(g)

8

0

(m3/sec)

6

10

2 11

5

12

4

10

2.8

14

(k)

4

9

4

discharge (m3/sec)

6

4

3

10-15.10.11

9.6

6

6

2

discharge (m3/sec)

(h)

16-21.2.12

8

8

8

0

discharge (m3/sec)

discharge (m3/sec) 10

6

11

6

7.0

2.0

(d)

9.0

10

6.5

4.0

9.2

7

6.0

3.5

30.1-6.2.11

discharge (m3/sec)

(l)

3.0

9.4

discharge (m3/sec) 8

1.8

12 10

6

1.6

13.11-2.12.11

14

(i)

20

5

1.4

18 16 14 12 10 8 6

discharge (m3/sec)

30

0

2.5

discharge (m3/sec)

11-19.5.11

1.2

8.2 2.0

(m3/sec)

10 5

40

EC (mS/cm)

3.0

15

discharge

EC (mS/cm)

2.5

20

15

(a)

8.4

25

25

23.11-4.12.10

9.0

8.6

30

(f)

(b)

8.8

discharge

25-28.9.11

35

EC (mS/cm)

6

(m3/sec)

7-11.12.10

16

0

(m)

28.2-7.3.12

6

8

10

12

discharge

14

16

18

20

(m3/sec)

Fig. 3. Relations between EC and water discharge for events of increase in water level.

potassium sources in the drainage basin are sewage (Nödler et al., 2011) and fertilizers such as KNO3 and KCl (Oren et al., 2003). The NO3/Cl ratios range between 6.7  104 and 2.6  102, with an average at 3.7  103, displaying the largest range among all ion ratios; the lowest value relates to the Karama dam event. Water released from this reservoir has extremely high chloride concentrations compared to nitrate (Salameh, 2004). The average ratio for this event was 1.14  103, whereas the average for the entire research period was 7.31  103. Events unrelated to rainfall had the highest values of the ratio, displaying very low chloride concentrations with a slight decrease in nitrate. A considerable decrease in concentration of all other ions implies a source with a stable concentration relative to chloride. Previous data (Farber, 2005) have a higher average NO3/Cl ratio (9  103) than the current larger data set, although nitrate concentrations are higher in the more recent samples, which may have been related to lower average chloride concentrations in the historic data.

The Br/Cl ratio in fresh water (1.56  103) is much lower than in brines. In salinization processes the ratio reflects the chemical composition of the saline end member. Bromide is conservative; hence the ratio is a reliable parameter for identifying salinization sources (Vengosh and Rosenthal, 1994). The average Br/Cl ratio in the LJR is 4  103, while a previously documented average Br/Cl ratio is slightly higher (4.48  103) than in our study due to the previous lower chloride concentrations (Farber, 2005). In evaporated water the ratio increases with evaporation rate and is not influenced by diagenetic processes. The Br/Cl ratio of brines derived from these processes is relatively high (2  103  1.2  102; Vengosh & Rosenthal, 1994). Nevertheless, previous data show a higher average Br/Cl ratio in the northern river reach (5.13  103) than at the Baptism Site (4  103; Farber, 2005), explainable by the general increase in chloride concentration in the 10 km river reach immediately upstream of the Baptism Site. Both NO3/Cl and Br/Cl do not correlate with Q. However, both

1102

N. Hillel et al. / Journal of Hydrology 527 (2015) 1096–1105

Table 3 Descriptive statistics of major ion concentrations and ion ratios (in mM).

All data Average Median Min Max 10% 90%

Na/Cl (molar ratio)

K/Cl (molar ratio)

NO3/Cl (molar ratio)

Br/Cl (molar ratio)

Ca/SO4 (molar ratio)

Mg/Ca (molar ratio)

33.4 25.5 5.04 159 16.7 61.3

0.791 0.674 0.911 0.752 0.826

0.0283 0.0182 0.0461 0.0236 0.0329

0.00725 0.00067 0.0258 0.00446 0.0102

0.00394 0.00280 0.00593 0.00351 0.00472

0.658 0.481 1.60 0.580 0.882

1.89 0.59 2.51 1.52 2.14

40.4 38.7 16.6 92.6 20.1 63.4

0.785 0.724 0.847 0.745 0.817

0.0280 0.0216 0.0355 0.0251 0.0332

0.00633 0.00079 0.0104 0.00462 0.00883

0.00402 0.00330 0.00515 0.00362 0.00473

0.629 0.481 0.808 0.569 0.748

2.01 1.73 2.51 1.82 2.20

26.5 21.5 5.04 159 15.8 38.8

0.793 0.674 0.911 0.768 0.834

0.0279 0.0182 0.0461 0.0224 0.0322

0.00848 0.00067 0.0258 0.00507 0.0115

0.00387 0.00280 0.00593 0.00349 0.00459

0.695 0.488 1.59 0.606 0.962

1.78 0.59 2.50 1.47 2.01

Na (mg/l)

Mg (mg/l)

Ca (mg/l)

K (mg/l)

Cl (mg/l)

SO4 (mg/l)

NO3 (mg/l)

Br (mg/l)

1818 1497 235 6452 1070 3005

426 359 49.6 1438 245 735

379 313 65.1 1475 227 604

113 93.7 9.53 412 53.2 193

3658 3014 451 13408 2172 6093

1357 1147 239 4133 659 2281

39.6 40.2 5.83 160 24.6 52.4

429 424 208 926 250 604

137 140 46.7 235 71.3 196

4374 4288 2017 8194 2537 6208

1636 1642 648 2969 841 2322

43.0 42.3 5.83 65.8 32.0 55.0

328 273 65.1 1475 216 488

90.6 79.4 9.53 412 48.5 125

2978 2569 451 13408 2006 4230

1099 953 239 4133 602 1567

36.5 38.0 7.00 160 19.9 48.2

Dry seasons Average 2156 Median 2158 Min 1050 Max 3817 10% 1271 90% 3056

521 503 253 1013 290 749

Wet seasons Average 1494 Median 1295 Min 235 Max 6452 10% 1001 90% 1965

339 298 49.6 1438 223 459

Fig. 4. (a) Relation between potassium and sodium to chloride. (b) Relation between nitrate and chloride.

weakly correlate with the Cl-content. While in a logarithmic regression NO3/Cl correlates negatively with Cl (r2 = 0.59, p < 0.01,), Br/Cl shows a linear positive relation (r2 = 0.43, P < 0.01) with Cl. Changes in the Br/Cl ratio with level are seen in Fig. 5b. Although dry season values are mostly in the SDC range (4  103  7.1  103) stressing its waters as a main origin, similar values characterize runoff events, likely related to dilution of Cl from a different source. The Mg/Ca ratio for water flowing in a limestone or chalk aquifer is 0.5–0.7. Dissolution of limestone or gypsum results in values less than 0.5 (Vengosh & Rosenthal, 1994). The average value found in our 2-year database is 1.86; in fact, it is higher than 1, except for very few cases related to flow events, with minimum and maximum values occurring in the respective dry and wet seasons. An initial increase in the ratio occurs with increasing chloride concentrations up to 4000 mg/l, followed by a slight decrease. The previous database (Farber, 2005) has an average value of 1.49, but a much lower (0.58) value at the upstream end of the LJR (at Old Gesher). In salinization processes the Mg/Ca ratio reflects the chemical composition of the saline end member. For instance, evaporated seawater is characterized by a high Mg/Ca ratio due to gypsum and aragonite precipitation (Vengosh & Rosenthal, 1994). An average value of 1.03 was found in ground water in the Negev (south of Israel), whereas the ratios in rain water, in floods and in the unsaturated zone are lower than 0.25. Based on

the stratigraphic column of this region, it can be concluded that Mg-rich saline groundwaters within the northern Lower Jordan Valley (Möller et al., 2011) cause the observed high ratios in the LJR. Pure gypsum dissolution is indicated by Ca/SO4 = 1 (Vengosh & Rosenthal, 1994). The average ratio in our samples is 0.7 (with min and max values in the dry and wet seasons, respectively). An initial decrease occurs in the ratio with an increase in chloride of up to 4000 mg/l, followed by a slight increase in the ratio. Previous data include a 3.4 ratio at the Old Gesher site and 0.9 at the Baptism Site (Farber, 2005), implying an influence of additional factors on sulfate and calcium concentrations along the river. Beside fine-grained clastic minerals, the geological formations in the central and southern Jordan Valley host abundant evaporites (aragonite, gypsum and halite). Water flowing through these layers dissolves the gypsum. However, in the river water most of the additionally inserted dissolved calcium may precipitate as CaCO3, reducing the residual calcium concentrations (Farber, 2005). 3.4. Stable isotopes in LJR water Sulfate isotope signatures range between 6.80‰ and 11.66‰ for d34S and between 10.50‰ and 15.40‰ for d18O, typically of marine sulfates. The majority of enriched d34S values were found in

1103

N. Hillel et al. / Journal of Hydrology 527 (2015) 1096–1105

(a)

0.95

500

Na/Cl NO3 stage

9 10

450

0.90

0.85

400

11

350

7 1

5

2

Na/Cl

0.80

6

300 250

3

0.75

4

200

NO3 (mg/L), stage (cm)

8

150

0.70

100 0.65 50 0.60 3/23/2010

7/1/2010

10/9/2010

1/17/2011

4/27/2011

8/5/2011

11/13/2011

2/21/2012

5/31/2012

0

date

(b) 1.00 0.90

500

Na/Cl Br/Cl stage

9 10

450

8

0.80

400

350

7 1

0.60

6

2 3 4

0.50

300

5

250

0.40

200

0.30

150

0.20 3/23/2010

7/1/2010

10/9/2010

1/17/2011

4/27/2011

8/5/2011

stage (cm)

Na/Cl, Br/Cl*100

11 0.70

11/13/2011

2/21/2012

5/31/2012

100

date Fig. 5. (a) Changes of Na/Cl and NO3 concentration with level. Grey area defines the range of Na/Cl recorded previously in the SDC. (b) Changes of Na/Cl and Br/Cl ratios with level. Grey area defines the range of Br/Cl recorded previously in the SDC.

summer samples, implying sources of depleted isotopic values during the wet period. This is not always the case for d18Osulfate, the values of which were highest during the Karama Dam event. The isotopic composition of sulfate is neither affected by water level nor by sulfate concentration. Plotting d18Osulfate versus d34S results in the range of seawater evaporites, tangent to atmospheric deposition (Cook and Herczeg, 2000). Intense remobilization of Lisan sediments within the rift valley leads to atmospheric deposition (Herut et al., 1995) that may either enter the aquifer due to groundwater recharge or become flushed to rivers by overland flow. Additionally, sulfate in groundwater originates from dissolution (Cook and Herczeg, 2000). Thus, the main source of sulfate to the southern LJR is geogenic during both baseflow and flood events. This conclusion is supported by the local stratigraphic column,

which is composed of evaporites within the Samra and Lisan formations (Nir, 1989). The d15N values range between 9.9‰ and 21.3‰ with two extreme values of 26.2‰ and 29.8‰ (Fig. 6), which refer to flood events and compatible with peak discharge. The d18Onitrate ranges between 7.7‰ and 28.1‰. On a plot of d18Onitrate versus d15N (Michener and Lajtha, 2007) results fall in the range of manure and septic wastes. Manure undergoes processes of volatilization of the lighter isotope of nitrogen and d15N values increase to 10‰–20‰. These values are similar to those for human feces; hence, they cannot be separately identified. During denitrification, a dissimilation of the light nitrate isotope results in the increase of d15N and a decrease in nitrate concentration (Cook and Herczeg, 2000). The case of the one extreme value fits the description above:

1104

N. Hillel et al. / Journal of Hydrology 527 (2015) 1096–1105

Table 4 Major ion statistics: comparison with previous results.

The highlighted values represent the importance of continuous monitoring.

δ18O (NO3) ‰ (V-SMOW)

30 26 22 18 14 10 6

8

12

16

20

24

28

32

δ15N (NO3) ‰ (air) Fig. 6. Nitrate isotope results. A general clustering and two extreme cases related to flood events and compatible to peak discharge are shown.

100

δ2H ‰ (V-SMOW)

results GMWL

80

LMWL

60 y = 4.41x - 3.59 r² = 0.93

40 20

-6

-5

-4

-3

-2

0 -1 0 -20

1

2

3

4

5

6

7

8

-40

δ18O ‰ (V-SMOW) Fig. 7. Results of water isotope signatures of the LJR compared to the global meteoric water line (GMWL) and the Levantine meteoric water line (LMWL). The trend line of our samples is shown in black.

in the January 28–February 01 2011 event there is a maximum value of 29.8‰ with a concomitant minimum nitrate concentration, implying an influence of microbial processes affecting nitrate concentrations and d15N values. The second extreme value, 26.2‰, relates to the first rain of the season. In this case nitrate concentration increased as did the concentration of the rest of the ions and the d15N value was unaffected by nitrification. Previous d15N data also fit the category of manure and septic waste inputs (Farber, 2005).

The water of the Lower Jordan River shows isotopic signatures, which always fall below both meteoric water lines, the global (GMWL) and the Levantine (LMWL; Fig. 7). Their trend line is well represented by d2H = 4.41  d18O  3.59, showing a much flatter slope compared to the GMWL d2H = 8  d18O + 10 (Craig, 1961) and LMWL d2H = 8  d18O + 22 (Dansgaard, 1964). The d18O signatures range between 4.5‰ and +7‰. The lowest signatures are connected to the assumed release of floodwaters from dams during and after heavy rainfalls in Feb-Mar 2012. This conclusion is supported by the fact that (i) the range of d18O signatures during that period (2.3‰ to 4.5‰) fits the isotopic range of d18O signatures in rain over Israel (2‰ to 9‰; Gat and Dansgaard, 1972) and (ii) the results are unique to this event. The highest isotopic signatures are related for the most to the Karama dam event. d18O signatures from that event range from +3.8‰ to 5.2‰, suggesting an input of highly evaporated water. A random winter sample from December-19 2011 shows d18O = 7‰, the heaviest recorded d18O signature; interestingly, it occurred during a recession. Since it is a random sample and there are no other data from this date, the cause for this extreme value cannot be ascertained: possible sources include reservoir or fishpond releases, both subjected to high evaporation rates. Previous studies (Farber, 2005) revealed for the same sampling location d18O signatures ranging from 4.70‰ to 1.40‰, being on average (3.04‰), isotopically more depleted than our samples (0.5‰). This difference may be due to distinct differences in the number of samples (Farber: n = 8; present study n = 428), randomly taken over the seasons, compared to daily samples of that study. d2H values range between 23.5‰ and 23.5‰. Deuterium values in precipitation are mostly negative in Israel, though positive values (up to 46.3‰) were found in the Negev and the Arava valley (Gat and Dansgaard, 1972). The deuterium excess (‘‘d’’) ranges between 38 and 12.6 with 60% of the samples lower than 0, although the characteristic values for precipitation in Israel are higher than 15 (Gat and Dansgaard, 1972). It is known that negative d values are correlated with aridity (Kendall and Coplen, 2001). The low d-values of the LJR reflect the aridity of the Jordan valley, which can be seen in the isotopic composition of the water.

4. Conclusions The LJR has been anthropogenically affected, altering water quantity and quality. Continuous monitoring and manual sampling

N. Hillel et al. / Journal of Hydrology 527 (2015) 1096–1105

enabled identification of annual trends and inter-seasonal trends in water quality. While water level and EC display a general inverse seasonal trend, extreme values are related to flood events during the wet period. EC–Q relations are mainly characterized by clockwise hystereses, implying that most of the events are affected by a washing effect. Major ion and isotopic analysis reveal non-precipitation events and anthropogenic influences such as the flushing from the Karama dam. High nitrate concentrations, supported by nitrate isotopic values, are indicative of a sewage source. It is concluded that continuous monitoring is required for understanding the long-term hydrochemical situation, especially during extreme events which occur infrequently, but may have a drastic effect on fauna and flora. Although results are a step forward in terms of our knowledge of the LJR hydrochemistry, such a dynamic system and its rehabilitation require several locations of continuous monitoring along its course in order to better understand the sources, their quality and the contribution of each source. Indeed, further investigations are required to better characterize the manner in which different sources influence the LJR waters and how these vary spatiotemporally. Acknowledgements This study was supported by the BMBF (German Federal Ministry of Education and Research) as part of the German SUMAR and DESERVE projects and by the Israel Water Authority. Thanks to Yehoshua Ratzon, Danny Girkevich, Daniel Zamler, Yaniv Munwes, Yuval Lorig, Dror Paz, Ariel Chen, Erez Shmerler, Inbal Zamir, Adi Shatkai, Ron Nativ, Roy Naor, Gal Litman, Eitan Hillel and Roni Livnon. The manuscript has benefitted from the comments of two anonymous reviewers. References Alexandrov, Y., Laronne, J.B., Reid, I., 2003. Suspended sediment concentration and its variation with water discharge in a dryland ephemeral channel, northern Negev, Israel. J. Arid Environ. 53 (1), 73–84. Bergelson, G., Nativ, R., Bein, A., 1999. Salinization and dilution history of ground water discharge into the Sea of Galilee, the Dead Sea Transform, Israel. Appl. Geochem. 14, 91–118. Calvo, R., Ben-Zvi, A., 2005. Spatial analysis of the Lower Jordan River Drainage Area and an assessment of the volume of its surface flow into the Dead Sea. Geological Survey of Israel, Ministry of Infrastructures. Jerusalem. pp. 24–32. Casciotti, K.L., Sigman, D.M., Galanter Hastings, M., Bohlke, J.K., Hilkert, A., 2002. Measurement of the oxygen isotopic composition of nitrate in seawater and freshwater using the denitrifier method. Anal. Chem. 74, 4905–4912. Chapman, D. (Ed.), 1996. Water Quality Assessments – A Guide to Use of Biota, Sediments and Water in Environmental Monitoring, second ed. E&FN Spon, Cambridge, Great Britain, pp. 19–39. Choudhari, J.S., Sharma, K.D., 1984. Stream salinity in the Indian arid zone. J. Hydrol. 71, 149–163. Cook, P.G., Herczeg, A.L. (Eds.), 2000. Environmental Tracers in Subsurface Hydrology. Kluwer Academic Publishers, Boston. pp. 195–232, 261–298. Craig, H., 1961. Isotope variations in meteoric waters. Science 133 (3465), 1702– 1703. Crosa, G., Froebrich, J., Nikolayenko, V., Stefani, F., Galli, P., Calamari, D., 2006. Spatial and seasonal variations in the water quality of the Amu Darya River (Central Asia). Water Res. 40, 2237–2245.

1105

Dansgaard, W., 1964. Stable isotopes in precipitation. Tellus 16, 436–468. Farber, E., 2005. The Lower Jordan River: River Stalinization, Relationship with Adjacent Groundwater and Future Management. Unpublished PhD Dissertation: Dept. of Geological & Environ. Sci., Ben-Gurion University of the Negev. pp. 1–110. Farber, E., Vengosh, A., Gavrieli, I., Marie, A., Bullen, T.D., Mayer, B., Holtzman, R., Segal, M., Shavit, U., 2004. The origin and mechanisms of salinization of the Lower Jordan River. Geochimica et Cosmochimia Acta 68 (9), 1989–2006. Gafny, S., Talozi, S., Al Sheikh, B., Ya’ari, E., 2010. Towards a Living Jordan River: An Environmental Flows Report on the Rehabilitation of the Lower Jordan River. EcoPeace/Friends of the Earth Middle East, Amman, Bethlehem, Tel-Aviv, pp. 13–69. Gat, J.R., Dansgaard, W., 1972. Stable isotope survey of the fresh water occurrences in Israel and the Northern Jordan Rift Valley. J. Hydrol. 16, 177–212. Glover, B.J., Johnson, P., 1974. Variations in the natural chemical concentration of river water during flood flows, and the lag effect. J. Hydrol. 22, 303–313. Hem, J.D., 1985. Study and Interpretation of the Chemical Characteristics of Natural Water. U.S. Geological Survey Water Supply Paper 2254, Washington DC, USA. pp. 42–57. Herut, B., Spiro, B., Starinsky, A., Katz, A., 1995. Sources of sulfur in rainwater as indicated by isotopic d34S data and chemical composition, Israel. Atmosph. Environ. 29 (7), 851–857. Israel Water Authority: public announcement: (retrieved 4/2014). Kendall, C., Coplen, T.B., 2001. Distribution of oxygen-18 and deuterium in river waters across the United States. Hydrol. Process. 15, 1363–1393. Knesset site: (retrieved 4/2014). Knighton, D., 1984. Fluvial Forms and Processes. Edward Arnold, London, UK, pp. 44–81. Laronne, J.B., Shen, H.W., 1982. The effect of erosion on solute pickup from Mancos Shale hillslopes. J. Hydrol. 59, 189–207. Ligon, K.L., Dietrich, W.E., Trush, W.J., 1995. Downstream ecological effects of dams. Bioscience 45 (3), 183. Marjang, N., Merkley, G.P., 2009. Surface velocity coefficients for application of the float method in rectangular and compound open channels. Irrig. Sci. 27, 457– 470. Michener, R.H., Lajtha, K. (Eds.), 2007. Stable Isotopes in Ecology and Environmental Science, second ed., Blackwell Scientific Publications, USA. pp. 1–50, 334–435. Möller, P., Siebert, C., Geyer, S., Inbar, N., Rosenthal, E., Flexer, A., Zilberbrand, M., 2011. Relationship of brines in the Kinnarot Basin, Jordan-Dead Sea Rift Valley. Geofluids 12, 166–181. Nir, D., 1989. Geomorphology of Israel. Publication of the Hebrew University Student Union, Jerusalem. pp. 51–53, 305–321 (in Hebrew). Nödler, K., Licha, T., Fischer, S., Wagner, B., Sauter, M., 2011. A case study on the correlation of micro-contaminants and potassium in the Leine River (Germany). Appl. Geochem. 26, 2172–2180. Oren, O., Yechieli, Y., Bohlke, J.K., Avraham, D., 2003. Contamination of groundwater under cultivated fields in an arid environment, central Arava Valley, Israel. J. Hydrol. 290, 312–328. Salameh, E., 2004. The tragedy of the Karama Dam project, Jordan. Acta Hydrochimica et Hydrobiologica 32 (3), 249–258. Segal-Rozenhaimer, M., Shavit, U., Vengosh, A., Gavrieli, I., Farber, E., Holtzman, R., Mayer, B., Shaviv, A., 2004. Source and transformation of nitrogen compounds along the Lower Jordan River. J. Environ. Qual. 33, 1440–1451. Siebert, C., Möller, P., Geyer, S., Kraushaar, S., Dulski, P., Guttman, J., Subah, A., Rödiger, T., 2014. Thermal waters in the Lower Yarmouk Gorge and their relation to surrounding aquifers. Chemie Erde – Geochem. http://dx.doi.org/ 10.1016/j.chemer.2014.04.002. Vengosh, A., Rosenthal, E., 1993. Saline groundwater in Israel: Its bearing on the water crisis in the country. J. Hydrol. 156, 389–430. Vengosh, A., Rosenthal, E., 1994. Saline groundwater in Israel: Its bearing on the water crisis in the country. J. Hydrol. 156, 389–430. Wieriks, K., Schulte-Wulwer-Leidig, A., 1997. Integrated water management for the Rhine river basin, from pollution prevention to ecosystem improvement. Nat. Resources Forum 21 (2), 147–156. Yorozuya, A., Kanno, Y., Fukami, K. & Oodaira, K., 2010. Development of automatic water discharge measurement system in Christodoulou, G.C. & Stamou, A.I. (Eds.). Environmental Hydraulics. Taylor & Francis Group, London. pp. 839–844.