HESSHydrology and Earth System SciencesHESSHydrol. Earth Syst. Sci.1607-7938Copernicus PublicationsGöttingen, Germany10.5194/hess-20-4031-2016Predicting the salt water intrusion in the Shatt al-Arab estuary using an
analytical approachAbdullahAli D.a.abdullah@unesco-ihe.orgalidinar77@gmail.comGisenJacqueline I. A.van der ZaagPieterSavenijeHubert H. G.https://orcid.org/0000-0002-2234-7203KarimUsama F. A.MasihIlyasPopescuIoanaDepartment of Integrated Water System and Governance, UNESCO-IHE Institute
for Water Education, Westvest 7, 2611 AX, Delft, the NetherlandsWater Resources Section, Delft University of Technology, Delft, the
NetherlandsDepartment of Civil Engineering, Basra University, Basra, IraqCentre for Earth Resources Research and Management, University Malaysia
Pahang, Lebuhraya Tun Razak, Gambang, 26300 Kuantan, Pahang, MalaysiaFaculty of Civil Engineering and Earth Resources, University Malaysia
Pahang, Lebuhraya Tun Razak, Gambang, 26300 Kuantan, Pahang, MalaysiaDepartment of Civil Engineering, University of Twente, P.O. Box 217, 7500
AE, Enschede, the NetherlandsFaculty of Civil Engineering, Politehnica University of Timisoara, RomaniaAli D. Abdullah (a.abdullah@unesco-ihe.org, alidinar77@gmail.com)6October201620104031404222March201628April201621September201622September2016This work is licensed under a Creative Commons Attribution 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by/3.0/This article is available from https://hess.copernicus.org/articles/20/4031/2016/hess-20-4031-2016.htmlThe full text article is available as a PDF file from https://hess.copernicus.org/articles/20/4031/2016/hess-20-4031-2016.pdf
Longitudinal and vertical salinity measurements are used in this study to
predict the extent of inland seawater intrusion in a deltaic river estuary. A
predictive model is constructed to apply to the specific tidal, seasonal, and
discharge variability and geometric characteristics of the Shatt al-Arab
River (SAR) situated along the border of Iraq and Iran. Reliable hydrologic
simulation of salinity dynamics and seawater intrusion was lacking prior to
this study. Tidal excursion is simulated analytically using a 1-D analytical
salt intrusion model with recently updated equations for tidal mixing. The
model was applied under different river conditions to analyse the seasonal
variability of salinity distribution during wet and dry periods near spring
and neap tides between March 2014 and January 2015. A good fit is possible
with this model between computed and observed salinity distribution.
Estimating water abstractions along the estuary improves the performance of
the equations, especially at low flows and with a well-calibrated
dispersion–excursion relationship of the updated equations. Salt intrusion
lengths given the current data varied from 38 to 65 km during the year of
observation. With extremely low river discharge, which is highly likely
there, we predict a much further distance of 92 km. These new predictions
demonstrate that the SAR, already plagued with extreme salinity, may face
deteriorating water quality levels in the near future, requiring prompt
interventions.
Introduction
Discharge of fresh river water into the ocean is closely related to vertical
and longitudinal salinity variations along an estuary (e.g. Savenije et al.,
2013; Whitney, 2010; Becker et al., 2010; Wong, 1995; MacKay and Schumann,
1990). River discharge also has a noticeable effect on the tidal range,
primarily through the friction term (the amount of energy per unit width lost
by friction) (Savenije, 2005). A decrease in river discharge into an estuary
could increase the tidal range and the wave celerity, and consequent increase
in salinity levels (Cai et al., 2012). Upstream developments of large dams
and water storage facilities change the nature of river flow and subsequently
alter river hydrology and quality (Vörösmarty and Sahagian, 2000;
Helland-Hansen et al., 1995). The Shatt al-Arab River (SAR) which discharges
through its estuary at the border between Iran and Iraq into the gulf is
facing serious reductions in freshwater inflows upstream and from its
tributaries, as well as significant salt intrusion downstream (Abdullah et
al., 2015). The alteration of river discharge also affects the estuarine
ecosystem in terms of sediments, nutrients, dissolved oxygen, and bottom
topography (Sklar and Browder, 1998). All these problems are strongly
featured in the SAR.
The increases in salinity along the SAR, particularly caused by salt
intrusion, have become a threat to the people and environment alike.
Generally, salt water intrusion makes the river water unfit for human
consumption and unacceptable for irrigation practices (Abdullah et al., 2016;
Al-Tawash et al., 2013). Saline water in the SAR estuary comes from both
natural (seawater intrusion) and anthropogenic sources. Thus, the pattern of
the salinity variation is complex because of the dynamic spatial and temporal
interaction between salinity sources. Available studies on the SAR identify
the escalating pressure of salinity increment and its consequences for water
users as well as the ecosystem (e.g. Abdullah et al., 2016; Al-Tawash et al.,
2013; Fawzi and Mahdi, 2014), but detailed information on the extent of salt
water intrusion under different conditions is lacking. Hence, there is a need
to investigate the impact of seawater intrusion among other sources on the
river salinity, and to analyse the dynamics of the saline
water–freshwater interface for effective water management.
Different approaches have been used to study the relationship between saline
water and freshwater in estuaries. Alber (2002) proposed a conceptual model
for managing freshwater discharge into estuaries. Wang et al. (2011) used an
empirical approach, conducting three hydrological surveys along six locations
around the Yellow River mouth to investigate the effect of abrupt changes in
the river discharge on the salinity variations. Using a numerical model,
Bobba (2002) analysed the mechanism of salt water and freshwater flow in the
Godavari Delta and found that freshwater withdrawals contribute to the
advance in seawater intrusion. Liu et al. (2004) applied a
2-D model to estimate the salinity changes in the Tanshui River, showing
that the significant salinity increase is a result of reservoir construction
and bathymetric changes. A 3-D model was used by Vaz et al. (2009) to study
the patterns of saline water in the Espinheiro tidal channel. The result
indicates that the model underestimated the salinity distributions for high
river inflow. Das et al. (2012) used a hydrology–hydrodynamics model to
examine salinity variations under different water diversion scenarios in the
Barataria estuary, and discovered that the diversions have a strong impact on
salinity in the middle section of the estuary and a minor impact in the upper
section.
Analytical approaches describing salinity distribution in estuaries have been
used by Ippen and Harlemen (1961), Prandle (1985), and Savenije (1986). An
analytical solution is able to provide important knowledge about the
relationship between the tide, river flow, and geometry of the tidal channel.
The 1-D modelling is usually based on a number of assumptions to simplify the
set of equations. Several available models generally assumed a constant tidal
channel cross section to linearize and simplify the calculation processes. In
this study the 1-D analytical salt intrusion model proposed by
Savenije (1986, 1989, 1993) is considered, which uses the more natural
exponential geometry and requires a minimal amount of data. The model has
been successfully applied to several single-channel estuaries worldwide (e.g.
Risley et al., 1993; Horrevoets et al., 2004; Gisen et al., 2015a). Moreover,
it can also describe the tidal propagation in multi-channel estuaries (Zhang
et al., 2012) as well as estuaries with a slightly sloping bottom (Nguyen and
Savenije, 2006; Cai et al., 2015).
The aim of this study is to determine the real extent of salt intrusion into
the SAR estuary. This is done by applying the 1-D analytical salt intrusion
model combined with the revised predictive equations for tidal mixing of
Gisen et al. (2015b). Then the predictive model was used to examine the
consequences of changes in river flow for the salinity distribution.
Research area
The SAR is located in southern Iraq and its estuary is connected to the gulf
(Fig. 1). The total length of the river is 195 km, of which the last 95 km
serve as a boundary between Iraq and Iran. The estuary receives freshwater
from four main tributaries. The Tigris and Euphrates rivers originate in
Turkey and form the SAR at their confluence near the city of Qurna, Iraq. The
other two tributaries, Karkheh and Karun, originate in Iran. The Karkheh is
connected with the SAR through a system of marshes, while the Karun
discharges into the SAR at approximately 87 km from the mouth.
The salient features of the Shatt al-Arab region (left) and the
aerial view of the estuary from Google Earth with the measurement locations
(not to scale) (right).
The estuary experiences a tidal cycle of approximately 12 h 25 min with
notable flood and ebb tides (Fig. 2). The estuary has a mixed-diurnal and
semi-diurnal tide with successive spring and neap tide. The tidal range (the
difference between the water levels at high water (HW) and low water (LW))
varies from 1 m (neap) to 3 m (spring). Salinity levels fluctuate at an
hourly scale depending on the tide cycles and freshwater discharge. Salinity
increases during flood tides and decreases during ebb. The impact of
freshwater inflows can be clearly recognized during neap tide and ebb
periods. The salinity level also varies along the year; for example, the
highest value measured in the year 2014 was 40 kg m-3 during summer
and the lowest value was 0.7 kg m-3.
Tidal elevation at Faw station in June 2014.
The SAR is the main surface water source for daily consumption and
agricultural uses in the region and serves around 3 million people, the
majority living in Basra. Rural communities live along the river and around
the marshes and derive their livelihoods mainly from agriculture and
livestock. The main agricultural lands extend along the river banks with
large date palm plantations. A variety of human activities along the SAR and
its tributaries deteriorates the water quality and has significantly
increased the salinity concentration over time. In addition, the decreases in
freshwater inflows into the estuary due to upstream water withdrawals have
allowed the seawater to intrude further upstream. Currently the Tigris is the
main source of freshwater for the SAR; its discharge ranges between 30 and
100 m3 s-1. The total discharge from other tributaries, except
the Karun, ranges between 0 and 10 m3 s-1. The available
information on discharge of the Karun is limited and inconclusive. Most
relevant is Ahvaz station in Iran (UN-ESCWA and BGR, 2013; Salarijazi et al.,
2012; Afkhami et al., 2007), the most downstream gauging station but still
located approximately 200 km upstream of the confluence with the SAR. Due to
large-scale water developments, the mean annual discharge of the Karun has
experienced a consistent negative trend from 818 to 615 m3 s-1
before and after 1963 respectively (UN-ESCWA and BGR, 2013).
Whereas Salarijazi et al. (2012) reported a mean annual river discharge at
Ahvaz of 1442 m3 s-1 for the period 1954–2005, the mean monthly
river discharge for the period between 1978 and 2009 was only
667 m3 s-1 (Meysam Salarijazi, personal
communication, 2015).
However, the Karun River discharge into the SAR is believed to have decreased
even more in recent years due to continued increases in water abstractions
upstream. The combination of tide and fluctuating river discharge makes it
difficult to recognize the real extent of salt intrusion and its impact on
the horizontal salinity pattern along the river under different conditions.
Adding to the complexity of studying salt intrusion in the Shatt al-Arab is
that it is the border river between Iraq and Iran, with strict security
conditions. This does not make it easy to organize hydrometric surveys by
speedboat and carry out salinity observations during an entire tidal cycle.
As a result, the field data collected during this study and the results
obtained by the analytical model form a unique data set for the region.
Theory of the analytical model
During a tidal cycle, the tidal velocity is near zero just before the tidal
current changes direction. This situation is known as high water slack (HWS)
just before the direction changes seaward, and low water slack (LWS) just
before the direction changes landward. The model originally proposed by
Savenije (1989), calibrated with measurements made at HWS, describes the
salinity distribution in convergent estuaries as a function of the tide,
river flow, and geometry, using Van der Burgh's coefficient (K) and the dispersion
coefficient (D0) at the mouth. A conceptual sketch of the 1-D model of
salt intrusion is shown in Fig. 3.
Sketch of the estuary, the longitudinal profile, and the top view.
The geometry of an estuary can be presented by exponential functions
describing the convergence of the cross-sectional area and width along the
estuary as
A=Aoexp-xa1for0<x≤x1,A=A1exp-(x-x1)a2forx>x1,B=Boexp-xb1for0<x≤x1,B=B1exp-(x-x1)b2forx>x1,
where Ao and Bo are the cross-sectional area
(L2) and width (L) at the estuary mouth (x=0), A1 and
B1 are the cross-sectional area and width at the inflection point
(x=x1), and a1,2 and b1,2 are the cross-sectional and
width convergence lengths (L) at x≤x1 and x>x1
respectively.
Combining Eq. (1) with Eq. (2), and Eq. (3) with Eq. (4), describes the
longitudinal variation of the depth:
h=hoexp-x(a1-b1)a1b1for0<x≤x1,h=h1exp-(x-x1)(a2-b2)a2b2forx>x1,
where h, ho, and h1 are the cross-sectional average
water depths (L) at distance x from the mouth, at the estuary mouth, and
at the inflection point respectively.
Integrating the geometry equations into the salt balance equation of Van der
Burgh (1972) yields a steady-state longitudinal salinity distribution along
the estuary (see Savenije, 2005) under HWS condition:
S-Sf=So-SfDDo1Kfor0<x≤x1,S-Sf=S1-SfDD11Kforx>x1,
where Do, D, and D1 (L2 T-1) are the dispersion coefficient at the
estuary mouth, at any distance x, and at the inflection point, S0,
S1 and S (M L-3) are the salinity at the estuary mouth,
inflection point, and distance x respectively, Sf is the
freshwater salinity, and K is the Van der Burgh coefficient which according
to Savenije (2005) has a value between 0 and 1, where
DDo=1-βoexpxa1-1for0<x≤x1,andDD1=1-β1expx-x1a2-1forx>x1,withβo=Ka1QfDoAofor0<x≤x1,β1=Ka2QfD1A1forx>x1.βo and β1 are the dispersion reduction rate (–) at
the estuary mouth and at the inflection point respectively, and Qf
is the freshwater discharge.
The salt intrusion model is used to estimate the salt intrusion length, which
can be determined using low water slack (LWS, the lower extreme salt
intrusion), high water slack (HWS, the upper salt intrusion), or tidal
average (TA, the average of the full tidal cycle). Savenije (2012) proposed
to calibrate the model on measurements carried out at HWS. This is to obtain
the maximum salt intrusion over the tidal cycle. The salinity distribution
can be computed at LWS and TA based on the relation between salinity
distributions during the three conditions. The salt distribution curve at HWS
could be shifted downstream over a horizontal distance equal to the tidal
excursion length (E) and half of the tidal excursion length (E/2) to
obtain the salt distribution curve at LWS and TA conditions respectively. The
model variables can be determined from field observations and shape analysis,
while the two parameters K and D0 remain unknown, in addition to
Qf, which is difficult to determine in the tidal region. To
facilitate the calibration process, D0 and Qf are combined in
one variable, the mixing coefficient α0 (L-1):
α0=D0Qf
After model calibration, the values for K and α0 are known and
the salinity at any point along the estuary can be calculated. Finally the
salt intrusion length (L) during HWS is obtained by
LHWS=x1+a2ln1β1+1.
The calibration parameters can be obtained based on field measurements, but
to turn the model into a predictive model, a separate equation for D0 is
required. A predictive equation for D0 was presented by Savenije (1993)
and then improved by Gisen et al. (2015b), who moved the boundary condition
to a more identifiable inflection point x1, based on observations made
for a large number of estuaries worldwide as
D1=0.1167E1υ1NR0.57
with
NR=ΔρghQfTρAEυ2
and
E=υTπ.NR is the estuarine Richardson number (–), the ratio of
potential energy of the buoyant freshwater to the kinetic energy of the tide,
ρ and Δρ (M L-3) are the water density and the density
difference over the intrusion length, g is the gravitational acceleration
(L T-2), T is the tidal period (T), υ is the velocity
amplitude (L T-1), and E is the tidal excursion (L).
This study tests the predictive performance of the 1-D analytical salt
intrusion model, combined with new revised predictive equations to analyse
the real extent of seawater intrusion in the SAR estuary under different
river discharge conditions.
Data collection
The 1-D analytical salt intrusion model is based on a number of parameters
that can be obtained through field surveys. Variables such as K and
D0 are not directly measurable, and therefore they are obtained by
calibrating the simulated salinity curve to the data sets from the salt
intrusion measurements. For this study four measurement campaigns were
conducted, mainly measuring salt concentrations and water levels. The
measurements took place during the wet and dry periods at spring and neap
tides. These were on 26 March (neap-wet), 16 May (spring-dry), 24 September
2014 (spring-dry), and 5 January 2015 (spring-wet).
Salinity measurements were conducted at the moment just before the flow
changes direction (HWS and LWS). The HWS and LWS represent the envelope of
the vertical salinity variation during tidal cycles, and are also used to
determine the longitudinal tidal excursion. A moving boat technique was used
in the field survey in which the boat moved with the speed of the tidal wave
to capture the slack moment. Starting from the mouth of the estuary and in
the middle of the course, the salinity variations during the tidal cycle were
observed. A conductivity meter, YSI EC300A (https://www.ysi.com) with a
cable length of 10 m, was used to measure the vertical salinity profile for
each meter depth from the bottom to the surface, and it was done repetitively
at an interval of 3–4 km (longitudinally) until the river salinity was
reached (in this case 1.5 kg m-3).
The required information on river discharge and cross-sectional profiles was
provided by the local water authority. It is difficult to measure the
discharge accurately in an estuary considering the tidal fluctuation. Hence,
the discharge data from the nearest (most downstream) station
were used in the analysis. The daily
streamflow data of all the tributaries within the country were obtained from
the Department of Water Resources in Iraq. However, there were no data on the
discharge of one tributary, the Karun River, located in
neighbouring Iran. Experts of the
water resources authority in Basra indicated that the average discharge of
the Karun River was estimated at 40 m3 s-1. River cross-sectional
data were collected based on the last survey carried out in 2012 by the GDSD
(General Directorate of Study and Design).
A0 and A1 are cross-sectional areas at the mouth
and inflection point respectively. B0 and B1 are channel widths
at the mouth and inflection point respectively, and a1, a2, and
b1, b2 are locations of the convergence length of the
cross-sectional area and width respectively. h‾ is the average
depth over the estuary length (of 60 km).
Vertical salinity distribution of the estuary measured between 0 and
58 km at HWS.
Salinity modellingGeometric characteristics
Results of the cross-sectional area, width, and depth are presented in a
semi-logarithmic scale plot in Fig. 4. This figure shows a good agreement
between the computed cross-sectional areas A, width B, and depth
h based on Eqs. (1)–(6) and the observed data, except for the part
between 40 and 50 km, which is shallower in comparison to the rest of the
estuary. The cross-sectional area A and width B are divided into two
reaches with the convergence lengths a1 and a2 of 22 and 26 km
respectively (see Table 1). The geometry changes in a decreasing pattern
landwards following an exponential function. In an alluvial estuary, the wide
mouth and shorter convergence length in the seaward part is generally
wave-dominated, while the landward part with longer convergence length is
tide-dominated. The average depth h is almost constant, with a very
slight decrease along the estuary axis (a depth convergence length of
525 km).
Vertical salinity profile
In Fig. 5 the results of the observed vertical
salinity profile at HWS are presented. It can be seen that the salt
intrusion mechanism is well mixed for the entire observation period. During
the wet period when river discharge is relatively high, a partially mixed
condition can be observed particularly in the downstream area (Fig. 5a
and d). In the neap-wet condition as shown in Fig. 5a, there is more
stratification and the partially mixed pattern occurs in almost the entire
stretch of the estuary. This is because at neap tide, the tidal flows are
small compared to the high freshwater discharge during the wet season.
Conversely, during the spring-dry period when the river discharge is
significantly low and the tidal range is large (Fig. 5b and c), the vertical
salinity distribution along the estuary is well mixed.
Predicted and measured salinity distribution during HWS, TA, and
LWS.
Characteristic values of the estuary, including the maximum salinity
at the mouth So, the river discharge Qf, tidal excursion
E, Van der Burgh coefficient K, the dispersion coefficient Do,
mixing number αo, and salt intrusion length L.
PeriodSoQfEKDoαoL(kg m-3)(m3 s-1)(km)(m2 s-1)(m-1)(km)26 March 201424109100.654033.73216 May 20142891100.654735.24224 September 201434.64815.50.654429.2655 January 20152853100.652815.342Longitudinal salinity profile
The measurements of salinity during HWS and LWS are presented in Fig. 6.
Calculations of the longitudinal salinity profiles are based on
Eqs. (7)–(14), where the dispersion D decreases over x until it
reaches zero at the end of the salt intrusion length. Coefficients K,
D0, and E were calibrated to obtain the best fit between measured
salinity data and simulated salinity variations. The longitudinal salinity
distributions during a tidal cycle are demonstrated by three curves: (1) the
maximum salinity curve at HWS; (2) the minimum salinity curve at LWS; and
(3) the average of HWS and LWS representing the average salinity curve at TA.
Tidal excursion (E) is determined from the horizontal distance between the
salinity curves of HWS and LWS. This distance is considered constant along
the estuary axis during the tidal cycle. In this study, the tidal excursion
is found to be 14 km on 24 September and 10 km for the other observations
(Table 2).
The results show good agreement between measured and simulated salinity
profiles, with few deviations between the observed and modelled salinities.
The small deviations may be due to the timing errors in which the boat
movement speed did not coincide exactly with the tidal wave. In Fig. 6 (a and
d), it can be seen that the measured salinity at distances 20 and 24 km
during HWS are higher than the simulated values. There is a sub-district
(with considerable agricultural communities) and a commercial
harbour, and it is believed that all
of their effluents and drainage water are discharged into the river. This
could be the reason for the salinity being a little higher than expected. In
Fig. 6c, the last measurement point is lower than the simulated one. This may
be due to the relatively shallow stretch between 40 and 50 km, which can
substantially reduce the salt intrusion. Also, a timing error may be an
explanation for this deviation: the boat did not move fast enough as it was
delayed for short stops at police checkpoints.
Comparison between the predicted and calibrated values of D0
and L.
All the field surveys indicate that the maximum salinity at the mouth ranged
from 24 to 35 kg m-3 (Table 2). The lowest maximum salinity is during
the neap-wet period and the highest is during the spring-dry period. It can
be seen that the seawater intrudes furthest in September (spring-driest
period) and shortest in March (neap-wet). These findings are logical because
during the wet season, the estuary is in a discharge-dominated condition and
the lower tide (neap) can be easily pushed back by the river discharge. On
the other hand, during the dry season the estuary is tide-dominated and the
higher tide (spring) managed to travel further inland without much
obstruction (low freshwater discharge). The tidal ranges recorded during
field surveys are 1.7, 3.2, 2.1, and 2.6 respectively as the same date shown
in Fig. 6a–d.
Besides seawater intrusion, human activities in the upstream part of the
estuary also contribute to the salinity levels along the river. From
observations, the river salinity in the inland part varies in space and time
between 1 and 2 kg m-3. Thus, the salt concentrations are the result
of a combination of anthropogenic and marine sources (Abdullah et al., 2016).
The findings from the longitudinal salinity distribution indicate that there
is a need to analyse and classify the effects of natural and anthropogenic
factors on estuary salinity.
The predictive model
The dispersion coefficient D is not a physical parameter that can be
measured directly. It represents the mixing of saline water and freshwater,
and can be defined as the spreading of a solute along an estuary induced by
density gradient and tidal movement. Knowing the river discharge is crucial
for determining a dispersion coefficient D from Eq. (9). However, it is
difficult to measure the river discharge accurately in the tidal region due
to the tidal fluctuation. In this study, the river discharge data on the days
of the measurements were used from the gauging station located at the most
downstream part of the river network.
For the situation where measured salinity is known, the dispersion
coefficient D0 and the salt intrusion length L at HWS were calibrated
by fitting the simulated salinity curve (Eqs. 7–14) against the field data.
In case no field data are available, the dispersion coefficient D1 was
estimated using Eq. (15). The predicted D1 then was used to determine the
predicted D0 (using Eq. 9) and L (using Eq. 14). Comparisons between
the calibrated and predicted values were done to evaluate the performance of
the model. The prediction performance was evaluated with two model accuracy
statistics: the root mean squared error (ERMS) and the
Nash–Sutcliffe efficiency (ENS) (Eqs. 18 and 19 respectively).
ERMS=1n∑i=1n(Pi-Oi)2,ENS=1-∑i=1N(Oi-Pi)2∑i=1N(Oi-O‾)2,
where P and O are the predictive and observed variables respectively,
and O‾ is the observed mean. The index (ENS) ranges
from -∞ to 1. It describes the degree of accurate prediction. An
efficiency of one indicates complete agreement between predicted and observed
variables, whereas an efficiency of less than zero indicates that the
prediction variance is larger than the data variance.
Measured and adjusted river discharge considering water consumptions
on the days of measurements.
DateMeasured riverAdjusted riverdischarge1discharge2(m3 s-1)(m3 s-1)26 March 201410911416 May 2014919624 September 201448585 January 20155363
1 not counting water abstractions and excluding the Karun
inflows; 2 deducting water abstractions and including the Karun inflows.
The predicted and calibrated values of Do and L
considering measured and adjusted river discharge.
Measured river discharge Adjusted river discharge Calibrated values Predicted values Calibrated values Predicted values Do (m2 s-1)L (km)Do (m2 s-1)L (km)Do (m2 s-1)L (km)Do (m2 s-1)L (km)40332495384223250937473425074449942528434416541670533655296528142395513354240048
Comparison between the predicted and calibrated values of D0
and L using the improved discharge data.
Figure 7 presents poor correlations between the calibrated and predicted
values of D; the situation is better in the case of L values. Table 4
displays the correlation between predicted and measured values. The
ENS obtained for D is -0.09 and reflects weak predictive
performance. Generally the model appears to overestimate the values of the
dispersion coefficient compared to the calibrated ones during the wet period
and to underestimate the value during the drought period in September. This
could be due to the use of the measured discharge at the end of the tidal
domain, which gives higher or lower values than the exact freshwater
discharging into the estuary, as it does not account for the discharge of the
Karun River at the downstream end and the water consumption and water losses
within the system (see Table 3). The SAR is the main freshwater source for
irrigation, domestic, and industrial activities in the region. Hence, water
consumption could highly affect the performance of a predictive model,
especially in the region where water withdrawals can considerably reduce
river discharge into the estuary.
In order to reduce the uncertainty in the discharge data, some alternative
approach has to be adopted. Gisen et al. (2015a) estimated the discharges for
the downstream areas by extrapolating the correlation of the gauged area with
the ungauged areas. Cai et al. (2014) developed an analytical approach to
predict the river discharge into an estuary based on tidal water level
observations. This method is only applicable in estuaries with a considerable
river discharge compared to the tidal flows. In this study, a simple approach
has been used to assess the discharge in the SAR estuary by deducting the
water withdrawals in the downstream region from the discharge data collected
at the lowest gauging points. In a similar way the average discharge of the
Karun was also estimated (Table 3).
Data on water withdrawals were collected from the water resources authority
and water distribution departments. Besides irrigation and domestic supply,
the industrial sector, including the oil industry, is also a significant
water user. Unfortunately this study could not obtain information on water
usage and disposal by the oil industry.
Results of the model performance in terms of root mean squared error
(ERMS) and Nash–Sutcliffe efficiency (ENS).
NSERMSEMeasured river discharge D0-0.0976 m2 s-1L0.756 kmAdjusted river discharge D00.4660 m2 s-1L0.94 km
The adjusted river discharge data are then applied in the predictive model to
evaluate the improvement of these changes in predicting values of D and
L (Table 4). The results obtained after the adjustment are shown in Fig. 8.
The figures demonstrate the improvements in predicting the dispersion and
maximum salt intrusion length and show the importance of computing the
freshwater discharge accurately. Furthermore the correlations between
predictive and observed values are improved for both D0 and L, 0.46
and 0.9 respectively, and the ERMS also reduced to
60 m2 s-1 and 4 km for D0 and L respectively (Table 5).
The prediction performance of the model is demonstrated in Fig. 9, where the
salinity curves were computed from the predictive equation of D1 and the
adjusted river discharges. Figure 9 shows that the prediction salinity curves
perform very well compared to the calibrated one during all periods, except
January 2015. This could be because the average discharge used for the Karun
River, in which the value is lower than the actual discharge, being the month
of January, is in the middle of the wet season. At such a time the SAR is
expected to receive high flow rates from the Karun River. On the other hand,
during this season more return flows are drained into the SAR from the large
irrigation scheme serviced by the Karun water system, increasing
anthropogenic salinity levels. Accurate estimation of river discharge into
the estuary is important in improving the predictive skill of the model.
Comparing the salinity curves of the calibrated results (dashed
lines) and the predicted results (solid lines) to the observed salinity
during the four periods of 2014.
The ultimate objective of the modelling is to assess the influence of upstream development on the estuarine
environment, and also to find the real extent of salt intrusion. The salinity
distribution along the estuary is highly linked to upstream conditions, such
as flow regulation and water withdrawals. For the purpose of improving the
SAR estuary management, the model can lead to estimation of a salt intrusion
length for a given freshwater discharge. This is useful for water supply
managers to determine the appropriate location (salinity-free region) for
water intake stations. Figure 10 demonstrates the salt intrusion length (L)
associated with different river discharges corresponding to water released
from the Tigris River into the SAR. The salt intrusion lengths are plotted
against a range of freshwater discharges from 5 to 120 m3 s-1.
The main finding is that the length of salt intrusion increases in a
non-linear way with decreasing river discharge. The salt intrusion length is
very sensitive to river discharge when the flow is low. From the plot it can
also be seen that the maximum salt intrusion could reach 92 km from the SAR
estuary at 5 m3 s-1 river discharge. This outcome exceeds a
preliminary estimate by Abdullah et al. (2016) based on a 1-year data series,
who found the salinity to reach up to 80 km considering the annual salinity
peaks along the river. An 80 km intrusion length corresponds to a measured
river discharge of 58 m3 s-1, whereas for the predictive model
this distance corresponds to a much lower discharge (7 m3 s-1).
It is, however likely that the true river discharge was lower, since during
the lowest discharge the irrigation demand is relatively high. It should also
be realized that in the region of 40–50 km the depth and cross-sectional
areas are substantially less. Such a shallow reach can reduce the salt
intrusion length substantially, as can be seen from Eq. (14), where
β1 is inversely proportional to A.
Relationship between river discharge and predicted salt intrusion
length.
Conclusions
A 1-D analytical salt intrusion model was applied to the SAR estuary based on
four survey campaigns in 2014 and 2015. This model is used to determine
longitudinal salinity distribution and the length of salt intrusion. The
analytical model is shown to describe well the exponential shape of the
estuary in the upstream direction. Moreover, the results show good agreement
between computed and observed salinity under different river conditions. This
indicates that the analytical model is capable of describing the extent of
seawater intrusion along the SAR estuary.
Results for the dispersion coefficient Do indicate that the
measured river discharge out of the tidal range is higher than the real
discharge into the estuary. This can be attributed to water withdrawals along
the tidal domain. In the case of low river discharge, water withdrawals have
a considerable effect on the predicted salt intrusion length. The river
discharge into the estuary was revised considering water withdrawals of the
irrigation and domestic sectors. Using adjusted river discharge improved the
performance of the predictive equations. For further improvement, it is
recommended to obtain more accurate estimation of the river discharge into
the estuary.
Seawater intrusion is driven by the discharge kinetics from tidal seawater
and the hydrostatic potential energy from freshwater fluctuations. Intrusion
lengths of 38, 40, 65, and 43 km correspond to tidal ranges of 1.7, 3.2,
2.1, and 2.6 m during March, May, and September 2014, and January 2015
respectively. The longer salt intrusion distance is caused by low river
discharge, as evident for September (dry period).
The SAR is the main source of freshwater for daily consumption and
irrigation. Decreased freshwater discharge and increased seawater intrusion
will exacerbate an already critical situation in that important agricultural
and ecological region. The model shows a scenario in which decreasing river
discharge, considered a likely event, can result in an increase in seawater
intrusion further upstream to a distance of 92 km. Additional salinity
sources from anthropogenic activities will diminish the volume of freshwater,
leading to very serious health problems and water and food insecurity.
Calibration of the model can be enhanced with further monitoring of discharge
and salinity from all the tributaries and used to make new estimates of
longitudinal salinity distribution under extreme conditions. Preventing salt
intrusion of these magnitudes can only be achieved if the water quantity and
quality of the upstream sources as well as along the SAR are promptly and
strictly regulated.
Data availability
Data used in this paper is available upon request to the correspondence
author.
Acknowledgements
The authors highly appreciate the support of Meysam Salarijazi
providing required information. We acknowledge the editor M. Vanclooster
for the valuable comments. We thank the two anonymous reviewers for their
interesting comments and suggestions.
Edited by: M. Vanclooster Reviewed by: two anonymous
referees
References
Abdullah, D. A., Masih, I., Van der Zaag, P., Karim, U. F. A., Popescu, I.,
and Al Suhail, Q.: The Shatt al-Arab System under Escalating Pressure: a
preliminary exploration of the issues and options for mitigation, Int. J.
River Basin Manage., 13, 215–227, 2015.
Abdullah, A. D., Karim, U. F. A., Masih, I., Popescu, I., and van der Zaag,
P.: Anthropogenic and tidal influences on salinity levels and variability of
the Shatt al-Arab River, Int. J. River Basin Manage., submitted, 2016.
Afkhami, M., Shariat, M., Jaafarzadeh, N., Ghadiri, H., and Nabizadeh, R.:
Development a Water Quality Management
Model for Karun and Dez Rivers, Iran. J. Environ. Health. Sci. Eng., 4,
99–106, 2007.
Alber, M.: A Conceptual Model of Estuarine Freshwater Inflow Management,
Estuaries, 25, 1246–1261, 2002.
Al-Tawash, B., Al-Lafta, S. H., and Merkel, B.: Preliminary Assessment of the
Shatt al Arab Riverine Environment, Basra Government, Southern Iraq, J. Nat.
Sci. Res., 3, 120–136, 2013.
Becker, M. L., Luettich, R. A., and Mallin, M. A.: Hydrodynamic Behaviour of
the Cape Fear River and Estuarine: A Synthesis and Observational
Investigation of Discharge–Salinity Intrusion Relationships, Estuar. Coast
Shelf S., 88, 407–418, 2010.
Bobba, A. G.: Numerical Modelling of Salt-water Intrusion due to Human
Activities and Sea-level Change in the Godavari Delta, India, Special Issue:
Towards integrated Water Resources Management for sustainable Development,
Hydrol. Sci., 47, S67–S80, 2002.
Cai, H., Savenije, H. H. G., Yang, Q., Ou, S., and Lei, Y.: Influence of
River Discharge and Dredging on Tidal Wave Propagation: Modaomen Estuary
Case, J. Hydraul. Eng., 138, 885–896, 2012.Cai, H., Savenije, H. H. G., and Jiang, C.: Analytical approach for
predicting fresh water discharge in an estuary based on tidal water level
observations, Hydrol. Earth Syst. Sci., 18, 4153–4168,
10.5194/hess-18-4153-2014, 2014.Cai, H., Savenije, H. H. G., and Gisen, J. I. A.: A coupled analytical model
for salt intrusion and tides in convergent estuaries, Hydrol. Sci. J., 61,
402–419, 10.1080/02626667.2015.1027206, 2015.
Das, A., Justic, D., Inoue, M., Hoda, A., Huang, H., and Park, D.: Impacts of
Mississippi River Diversions on Salinity Gradients in a Deltaic Louisiana
Estuary: Ecological and Management Implications, Estuar. Coast Shelf S., 111,
17–26, 2012.
Fawzi, M. N. and Mahdi, A. B.: Iraq's Inland Water Quality and Their Impact
on the North-Western Arabian Gulf, Marsh Bulletin, 9, 1–22, 2014.
Gisen, J. I. A., Savenije, H. H. G., Nijzink, R. C., and Wahab, A. K.:
Testing a 1-D analytical salt intrusion model and its predictive equations in
Malaysian estuaries, Hydrol. Sci. J., 60, 156–172, 2015a.Gisen, J. I. A., Savenije, H. H. G., and Nijzink, R. C.: Revised predictive
equations for salt intrusion modelling in estuaries, Hydrol. Earth Syst.
Sci., 19, 2791–2803, 10.5194/hess-19-2791-2015, 2015b.
Helland-Hansen, E., Holtedahl, T. and Liye, K. A.: Enviromental Effects. Vol.
3. Hydropower Development, Norwegian Institute of Technology, Trondheim,
1995.
Horrevoets, A. C., Savenije, H. H. G., Schuurman, J. N., and Graas, S.: The
Influence of River Discharge on Tidal Damping in Alluvial Estuaries, J.
Hydrol., 294, 213–228, 2004.
Ippen, A. T. and Harlemen, D. R. F.: One-Dimensional Analysis of Salinity
Intrusion in Estuaries, Technical Bulletin No. 5, Committee on Tidal
Hydraulics, Corps of Engineers, US Army, 1961.
Liu, W., Hsu, M., Wu, C., Wang, C., and Kuo, A. Y.: Modeling Salt Water
Intrusion in Tanshui River Estuarine System-Case-Study Contrasting Now and
then, J. Hydraul. Eng., 130, 849–859, 2004.
MacKay, H. M. and Schumann, E. H.: Mixing and circulation in the sundays
river estuary, South Africa, Estuar. Coast Shelf S., 31, 203–216, 1990.Nguyen, A. D. and Savenije, H. H.: Salt intrusion in multi-channel estuaries:
a case study in the Mekong Delta, Vietnam, Hydrol. Earth Syst. Sci., 10,
743–754, 10.5194/hess-10-743-2006, 2006.
Prandle, D.: On Salinity Regimes and the Vertical Structure of Residual Flows
in Narrow Estuaries, Estuar. Coast Shelf S., 20, 615–635, 1985.
Risley, J. C., Guertin, D. P., and Fogel, M. M.: Salinity Intrusion
Forecasting System for Gambia River Estuary, J. Water Res. Pl.-ASCE, 119,
339–352, 1993.
Salarijazi, M., Akhond-Ali, A.-M., Adib, A., and Daneshkhah, A.: Trend and
change-point detection for the annual stream-flow series of the Karun River
at the Ahvaz hydrometric station, Afr. J. Agric. Res., 7, 4540–4552, 2012.
Savenije, H. H. G.: One-Dimensional Model for Salinity Intrusion in Alluvial
Estuaries, J. Hydrol., 85, 87–109, 1986.
Savenije, H. H. G.: Salinity Intrusion Model for High-Water Slack, Low
Water-Slack, and Mean Tide on Spread Sheet, J. Hydrol., 107, 9–18, 1989.
Savenije, H. H. G.: Predictive Model for Salt Intrusion in Estuaries, J.
Hydrol., 148, 203–218, 1993.
Savenije, H. H. G.: Salinity and tides in alluvial estuaries, Amsterdam:
Elsevier, 1–150, 2005.Savenije, H. H. G.: Salinity and tides in alluvial estuaries, completely
revised 2nd Edn., Delft University of Technology, available at:
http://salinityandtides.com (last access: 20 January 2015), 2012.Savenije, H., Cai, H., and Gisen, J.: Developing a coupled analytical model
for analyzing salt intrusion in alluvial estuaries, AGU Fall Meeting
Abstracts 1, 1664, 2013.
Sklar, F. H. and Browder, J. A.: Coastal Environment Impacts Brought about by
Alteration to Freshwater Flow in the Gulf of Mexico, Environ. Manage., 22,
547–562, 1998.
UN-ESCWA and BGR (United nations Economic and Social Commission for Western
Asia; Bundesanstalt für Geowissenschaften und Rohstoffe): Inventory of
Shared Water Resources in Western Asia, Beirut, 2013.
Van der Burgh, P.: Ontwikkeling van een methode voor voorspellen van
zoutverdelingen in estuaria, kanalen en zeeen, Rijkswaterstaat Rapp, 10–72,
1972.
Vaz, N., Dias, J. M., and Leitao, P. C.: Three-Dimensional Modelling of a
Tidal Channel: The Espinheiro Channel (Portugal), Cont. Shelf Res., 29,
29–41, 2009.
Vörösmarty, C. J. and Sahagian, D.: Anthropogenic Disturbance of the
Terrestrial Water Cycle, BioScience, 50, 753–765, 2000.
Wang, Y., Liu, Z., Gao, H., Ju, L., and Guo, X.: Response of Salinity
Distribution around the Yellow River mouth to Abrupt Changes in River
Discharge, Cont. Shelf Res., 31, 685–694, 2011.
Whitney, M. M.: A Study on River Discharge and Salinity Variability in the
Middle Atlantic Bight and Long Island Sound, Cont. Shelf Res., 30, 305–318,
2010.Wong, K. C.: On the relationship between long-term salinity variations and
river discharge in the middle reach of the Delaware estuary, J. Geophys.
Res., 100, 20705–20713, 10.1029/95JC01406, 1995.Zhang, E. F., Savenije, H. H. G., Chen, S. L., and Mao, X. H.: An analytical
solution for tidal propagation in the Yangtze Estuary, China, Hydrol. Earth
Syst. Sci., 16, 3327–3339, 10.5194/hess-16-3327-2012, 2012.