Arab J Geosci (2016) 9:472 DOI 10.1007/s12517-016-2442-6

ORIGINAL PAPER

Intervention scenarios to manage seawater intrusion in a coastal agricultural area in Oman Edda Kalbus 1,2 & Slim Zekri 3 & Akbar Karimi 4

Received: 24 January 2016 / Accepted: 22 March 2016 # Saudi Society for Geosciences 2016

Abstract The Batinah coastal plain in northern Oman has experienced a severe deterioration of groundwater quality due to seawater intrusion as a result of excessive groundwater abstraction for agricultural irrigation. Upgrading all farms to fully automated irrigation technology based on soil moisture sensors may significantly reduce the water demand and lead to recovering groundwater levels. This study compares the effects of smart irrigation technology, recharge dams, and a combination of both on seawater intrusion in the coastal aquifer of the Batinah. A groundwater flow and transport model is used to simulate the effect of reduced pumping rates on seawater intrusion for various intervention scenarios over a simulation period of 30 years, and an economic analysis based on cost-benefit analysis is conducted to estimate the potential benefits. Results indicate that a combination of smart irrigation and recharge dams may prevent further deterioration of This article is part of the Topical Collection on Water Resources in Arid Areas * Edda Kalbus [email protected] Slim Zekri [email protected] Akbar Karimi [email protected] 1

Department of Applied Geosciences, German University of Technology in Oman, Muscat, Sultanate of Oman

2

Present address: Research and Evaluation Unit, Auckland Council, Auckland, New Zealand

3

Department of Natural Resource Economics, Sultan Qaboos University, Muscat, Sultanate of Oman

4

Department of Built and Natural Environment, Caledonian College of Engineering, Muscat, Sultanate of Oman

groundwater quality over the next 30 years. In conjunction with increased efficiency, this combination also generates the highest gross profit. This outcome shows that the problem of seawater intrusion needs to be tackled by a comprehensive, integrated intervention strategy. Keywords Seawater intrusion . Smart irrigation . Simulation . Cost-benefit analysis

Introduction Excessive pumping of groundwater and the resulting depletion or deterioration of groundwater resources are a major concern in arid areas. The Batinah coastal plain in Northern Oman is particularly affected because extensive agricultural areas in this region require year-round irrigation due to low precipitation rates. The water demand is primarily met by groundwater which is pumped from shallow alluvial aquifers. Extensive mechanical pumping of groundwater for irrigation started in the 1970s, and since then, agriculture has expanded continuously in the Batinah. Nowadays, more than 90 % of Oman’s renewable water resources are used for agriculture. In 2010, water demand in the Batinah was 45 % larger than available renewable water resources (OSS 2012). The continuous abstraction of groundwater has led to a decline in groundwater levels of around 0.3–0.4 m/year (Rajmohan et al. 2007), and today’s levels are several meters below sea level near the coastline. Consequently, seawater intrusion has caused a deterioration of groundwater quality and numerous farms have been abandoned due to the high salinity of groundwater available for irrigation. Between 2006 and 2010, the Batinah region has lost more than 5000 ha of farmland to seawater intrusion (OSS 2012).

472

Arab J Geosci (2016) 9:472

Page 2 of 12

Seawater intrusion is an issue of concern for the majority of coastal aquifers in the world (Kallioras et al. 2013). Seawater intrusion resulting from groundwater abstraction for agricultural irrigation has been reported in many regions including the Mediterranean (Zalidis et al. 2002; López Geta 2003; Gaaloul et al. 2012), Malaysia (Baharuddin et al. 2012), Korea (Lee and Song 2006), the Middle East (Melloul and Goldenberg 1997), Australia (Werner 2009), and California (Hanson 2003). Managing seawater intrusion in coastal aquifers essentially requires modifying the aquifer’s water balance (Werner et al. 2013). This may involve regulating pumping activities in order to reduce the amount of water abstracted. However, this may have considerable socio-economic effects for farmers relying on groundwater abstraction for irrigation. Furthermore, controls on groundwater abstraction require comprehensive monitoring of abstraction rates. In the Batinah, groundwater abstraction wells are generally not metered, and groundwater abstraction is not charged. The introduction of water meters is not welcomed by the farmers in fear of future water charges. The current practice does not give the farmers any incentive to save water, and as a result, the majority of farms use simple flood irrigation instead of more efficient smart irrigation technologies. A promising approach could be the introduction of combined irrigation/ fertilization systems which may lead to reduced water consumption and to substantial savings in fertilizer and labor costs (Zaier et al. 2016, Smart Irrigation System. Design for groundwater saving monitoring at farm scale, Submitted to Sensors). Another option to modify the aquifer’s water balance is increasing the groundwater recharge. In the Batinah, seven main groundwater recharge dams exist with a capacity of 32 Mm3 to capture surface runoff and enhance the replenishment of the aquifers (OSS 2012). Most dams are built to store the water only for a short time period in order to avoid evaporation given the hot climate prevailing in Oman. The stored water is usually released slowly to the wadi channel over a few days after a rainfall event to allow infiltration and aquifer recharge. Abdalla and Al-Rawahi (2013) showed that the Al Khod dam has a strong effect on the replenishment of the aquifer and contributes significantly to the mitigation of seawater intrusion in this area. A first study on coastal salinization in the Batinah found that seawater intrusion already posed a severe problem on agricultural development between 1970 and 1980 (Stanger 1985). Al Barwani and Helmi (2006) evaluated the results of groundwater sampling programs between 1984 and 2005 in the area between Seeb and Suwayq. They found that seawater intruded up to 12 km inland and that the areas with water suitable for irrigation decreased by 7 % in this period. Walther et al. (2012) developed a numerical flow and transport model for Wadi Ma’awil in the Batinah for the simulation of seawater intrusion. In a follow-up study, they found that it may take hundreds of years to remediate seawater intrusion in this area (Walther et al. 2014).

In a framework of Integrated Water Resources Management, a number of studies have addressed the problem of seawater intrusion in the Batinah (Kalbus et al. 2012). Tools were developed for improved irrigation management (Kloss et al. 2012). One optimization tool consisted of a stochastic framework for decision support for optimal planning and operation of water supply in irrigation (Schütze et al. 2012). Integrated assessment, prognosis, planning, and management tools were developed, which take into consideration socio-economic aspects in addition to meteorological, hydrological, and agricultural phenomena, and were applied to selected catchments in the Batinah (Grundmann et al. 2012; Grundmann et al. 2013). This paper compares the potential effects of intervention scenarios addressing seawater intrusion in the region of Suwayq in the Batinah. Five different scenarios are considered including business as usual, introduction of smart irrigation, increased efficiency, construction of recharge dams, and a combination of those. A groundwater flow and transport model is used to simulate the effect of reduced groundwater abstraction rates and enhanced groundwater recharge downstream of dams on seawater intrusion and resulting groundwater salinity. An economic analysis estimates the subsequent effects on the farmers’ profits. The scenarios are ranked based on their present value.

Study area The study area is located around the village of Al Suwayq in the Batinah coastal plain in the Sultanate of Oman (Fig. 1). Two wadi systems extend over the study area: Wadi Bani Ghafir and the Mayhah-Mabrah-Hajir System, which discharge to the north into the Gulf of Oman. Average annual precipitation is 77 mm. Around 8000 farms are situated in the study area (Fig. 2) with a total cultivated area of approximately 104 km2 out of which 51 % is allocated to vegetable crops, followed by 24 % to forage crops and 16 % to date palm trees (MAF 2014). Each farm has one or more groundwater wells to supply irrigation water. The aquifers in the study area are formed by a 16- to 50-mthick Alluvium formation overlaying a 213- to 582-m-thick Upper Fars formation. The Alluvium formation consists of loose weathered limestone and ophiolite gravels with interbedded clays, and the Upper Fars formation consists of dolomites, cemented conglomerates, and chalky limestones. The underlying Middle Fars Formation consists of claystones and is considered an aquitard. The groundwater found in the Alluvium and Upper Fars formation is predominantly of sodium bicarbonate or magnesium bicarbonate type, suggesting young, recently recharged water (MRMWR 2006). The recharge areas of the aquifers are located in the upstream areas of the wadi catchments in the Hajar mountains, and discharge occurs into the Gulf of Oman.

Arab J Geosci (2016) 9:472

Page 3 of 12 472

N

Doha Abu Dhabi

Gul

f of

Al Suwayq

Om

an

Al Suwayq Al Muscat coa Batin sta ah l pl ain

UAE SAUDI ARABIA

Wadi Bani Ghafir

Mayhah-MabrahHajir System

OMAN

Administrative boundary Al Suwayq Model boundary Wadi catchment Elevation (m asl) High : 2986

0

125

250

0

500 km

5 10

Low :0

20 km

Fig. 1 Study area (UAE United Arab Emirates, m asl meters above sea level)

Materials and methods

Available data

Groundwater model

Data on subsurface properties as well as on the inflow and outflow components of the water balance were generally scarce. For constructing the aquifer geometry, borehole logs from only six boreholes were available (MRMWR 2006). The location of these boreholes is shown in Fig. 2 (Boreholes Upper Fars). The borehole logs revealed a thickness of the Alluvium ranging from 16 to 50 m and a thickness of the Upper Fars ranging from 213 m in the western part to 582 m

A numerical groundwater flow and transport model was set up using SEAWAT (Langevin et al. 2008) to simulate the seawater intrusion in the study area. Available data on subsurface properties and water balance components and their incorporation into the groundwater model are described below.

b

Boreholes Upper Fars Boreholes Alluvium Farms Proposed Recharge Dams Model boundary

0 -2 -4 1974

Observation wells (OW)

2

0

5

10

20 km

OW1 OW3 OW5 OW7

OW1corr OW4 OW5corr OW7corr

2010

Legend

4

2006

OW7

6

2002

OW3OW4

1998

OW6

8

1994

OW2

10

1990

OW1

12

1986

OW5

14

1982

Al Suwayq

16

1978

Observed groundwater level (m asl)

a

OW2 OW4corr OW6

Fig. 2 a Farms, boreholes, observation wells (OW), and proposed recharge dams in the model area and b time series of observed and correlated (corr) groundwater levels at the seven observation wells from 1974 to 2010

472

Arab J Geosci (2016) 9:472

Page 4 of 12

in the eastern part of the study area. The logs showed that the Upper Fars consists of alternating layers of conglomerates, cemented gravels, limestones, siltstones, and mudstones. However, no continuous layering could be identified from the six borehole logs. Pumping tests at two of these boreholes in the Upper Fars formation (one at the eastern transect and one at the western transect) yielded estimates of hydraulic conductivities of 2.3 m/day (west) and 18.4 m/day (east). Storage coefficients were estimated to 2.5E−04 (west) and 3.2 E−04 (east) (MRMWR 2006). For the Alluvium, estimates of hydraulic conductivity were available from three boreholes around the study area (Boreholes Alluvium in Fig. 2) and ranged from 0.9 to 15 m/day. Specific yield was given at the western borehole with 2.8E−04. Average annual groundwater inflows from the mountainous regions to the model area were estimated in MRMWR (2006) based on catchment area and width, hydraulic gradient, and transmissivity. For the Mayhah-Mabrah-HajirSystem, estimated average annual groundwater inflow was 40.5 million m3/year and for Wadi Bani Ghafir 21.7 million m3/year. Direct rainfall recharge in the coastal plane was estimated as 22 % of annual rainfall (OSS 2012). Average annual rainfall in the study area sums up to 77 mm/year as calculated from 17 rainfall stations, with two stations dating back to 1974 and the majority of stations commencing in 1983. Thus, average annual direct rainfall recharge was estimated to 17 mm/year. Groundwater abstractions are not metered in the study area; therefore, measurements of abstraction rates were not available. Estimates of total groundwater abstraction for the Batinah region are given in OSS (2012) for the years 1984, 1995, 1997, and 2010. The location and size of farms in the Suwayq study area were available as a GIS layer file. Furthermore, the location of three proposed recharge dams was available (Fig. 2). Records of groundwater level and salinity were limited. Observations of groundwater levels were available from seven observation wells in the study area (Fig. 2). At three of these wells, observations commenced in 1974; the remaining wells had data records starting around 1984. Groundwater levels were recorded as depths below measurement point. To convert them to hydraulic heads, the elevations of the measurement points above reference level need to be known. The data records indicated that these elevations were obtained from a digital elevation model. Results of groundwater salinity surveys from several years between 1984 and 2005 are presented in Al Barwani and Helmi (2006). Salinity data from 1995, 2000, 2005, and 2010 were available for 138 wells in the study area. However, the depths and screen lengths of the sampled wells were not known, and therefore, the depths at which the water samples were taken were unclear. Hence, the data were of limited utility.

Model setting The numerical groundwater model extends over the agricultural areas within the administrative boundaries of Suwayq (Fig. 1). The model area is approx. 20 × 40 km. The model grid was aligned in flow direction and set up with a uniform grid cell width of 500 m. In flow direction, the grid size varies from 1000 m toward the mountains to 200 m near the coast. The thicknesses of the Alluvium and the Upper Fars were interpolated from the borehole data. The Alluvium is represented by three model layers and the Upper Fars by 30 model layers. Hydraulic parameters of both formations were interpolated from available borehole information. Vertical hydraulic conductivity was assumed one tenth of horizontal hydraulic conductivity. At the mountain-facing side, groundwater inflow from the mountainous area is represented as specified-flux boundary condition. The sea side is modeled as constant head boundary and the lateral sides as no-flow boundary as they are aligned with the flow direction. No records of groundwater abstraction were available from the approx. 8000 farms in the study area. Therefore, abstraction rates were estimated based on total groundwater abstraction in the Batinah as given in OSS (2012). Earliest data on groundwater levels were from the year 1974. We therefore defined this year to represent the natural, pre-development conditions and as starting year for pumping activities. Although we know that pumping had already started much earlier (Stanger 1985), we considered this assumption appropriate for our study as there was no other data available and a considerable increase in abstraction is known to have started in the 1970s. For the period between 1974 and 2010, specific pumping rates (m3 of pumped water per m2 of cultivated area per day) were interpolated for each year from the data given in OSS (2012) and multiplied with the cultivated area per model grid cell to obtain a pumping rate per grid cell. This results in spatial variations in pumping rates across the study area depending on the size of cultivated area. The depths of the pumping wells were only known for about 40 wells in the study area where they ranged between 15 and 100 m with an average of 45 m. The wells are therefore assigned to the third model layer which is located at depths between 35 (average layer top) and 60 m (average layer bottom) below surface. Location and size of three proposed recharge dams were known; however, no information on proposed operation schemes was accessible. To simulate the future effect of these recharge dams, a constant average recharge rate is assigned to the downstream area of the dams. An estimate of the average recharge rate (8.22E−4 m/day) was obtained from the findings of Abdalla and Al-Rawahi (2013) who studied the impact of Al Khod recharge dam in the Batinah. This dam is located around 75 km away from our study area at the eastern edge of the Batinah. The subsurface consists of similar rock types,

Arab J Geosci (2016) 9:472

and therefore, we assumed that a similar recharge rate can be expected for the proposed dams in the Suwayq study area. A constant salt concentration of 35 g/L is set as boundary condition at the sea side, and a background concentration of 0.3 g/L is assigned as initial salt concentration in the groundwater and the inflow from the mountainous area. Longitudinal and lateral dispersivities (200 and 20 m) were adopted from Walther et al. (2012) who estimated the values for a neighboring catchment. The diffusion coefficient is set to 1.9E−04 m2/ day (Tanaka 1975). Freshwater density is 1000 kg/m3, and seawater density is 1025 kg/m3. Stress periods and calibration Three stress periods were defined: (i) pre-development state before 1974 (steady state), (ii) post-development state 1974– 2010 using estimated historic pumping rates, and (iii) prediction of different intervention scenarios over the period 2010– 2040 using estimated future pumping rates and recharge rates for the recharge dams. Inflow from the mountains, hydraulic conductivities, and specific pumping rates were calibrated to match simulated and observed groundwater levels for the period 1974–2010. Since only three observation wells had records starting in 1974, the remaining four observation time series were correlated with those three to obtain a full set of seven values for the year 1974 (Fig. 2). Seasonal or longer-term variations of inflow and outflow components were not considered in this study. For example, in the years 1995–1997, annual rainfall was higher than usual, which resulted in increasing groundwater levels. However, no information on resulting changes in inflow from the mountains or pumping rates was available. Therefore, a constant value of inflow from the mountains was used over the entire simulation period and adjusted during calibration. Similarly, the pumping rates were not varied annually, but the overall rate of increase in pumping between 1974 and 2010 was adjusted during calibration. The simulation of intervention scenarios in the third stress period (2010–2040) aimed at predicting the future development of seawater intrusion depending on the introduction of smart irrigation technologies, the construction of recharge dams, and a combination of both as compared to a businessas-usual scenario (Table 1). A detailed description of the different scenarios follows below. For each scenario, the salt concentration of the groundwater at a depth of 35–60 m below surface (layer 3) was retrieved from the model results and classified into one of five salinity classes (OSS 2012): & & & & &

Freshwater (salt concentration ≤1.5 g/L) Low salinity (1.5 g/L < salt concentration ≤ 3 g/L) Moderate salinity (3 g/L < salt concentration ≤ 5 g/L) High salinity (5 g/L < salt concentration ≤ 10 g/L) Very high salinity (salt concentration >10 g/L)

Page 5 of 12 472

The farm area falling into each salinity class was calculated and the percentage of farm area within each salinity class derived.

Intervention scenarios and economic analysis Salinization of groundwater used for irrigation results in a decrease in crop yields and, thus, in farm profitability (Zekri 2008). An economic analysis is carried out to estimate the economic effects of seawater intrusion coupling the numerical model with a cost-benefit analysis model. The analysis is based on the simulation of five intervention scenarios that include policy decisions at farm level and/or at aquifer level. The cost-benefit analysis is undertaken at the region level considering each scenario as a separate project. For all scenarios, we assume that farmers will produce the same types of crops and same crop mix all over the 30-year simulation period. Constant prices are used. A discount rate of 5 % is used to annualize the investment costs and/or to discount for the future value of the money. Gross profit depending on salinity class and farm size is taken from OSS (2012). The analysis is done at project level reflecting costs and benefits at society level only. The economic analysis is simulated for five scenarios considering a period of 30 years. The first scenario is business as usual (BAU), where no policy changes of any type are undertaken. It is a Blaissez faire laissez aller^ scenario where farmers keep the same pumping rates without any governmental intervention to protect the aquifer from over-pumping and salinization. The remaining scenarios assume an exogenous institution intervention where the government interferes to correct for the salinity problem by providing subsidies to improve the irrigation systems at farm level and/or by building recharge dams. The second scenario considers the introduction of smart irrigation (SI) in the form of drip/sprinkler irrigation linked to soil humidity sensors monitored through a central computer in the farm. The objective is the application of adequate volumes of water according to each crop water requirement with a fully automated system. This would avoid subjective decisions by farmers on when and how much water should be applied and the consequent low irrigation efficiency. Zaier et al. (2016, Smart Irrigation System. Design for groundwater saving monitoring at farm scale, Submitted to Sensors) estimated the annual cost of SI at US$623/ha or US$262/feddan or acre. This cost includes the electronic system which upgrades the existing drip/sprinkler system in order to fully automate it. We assume 20 % of water saving with the SI system and no other benefits such as yield increase or labor irrigation cost reduction. Scenario 3 is a variant of scenario 2, where we assume that gross farm profit will increase by 10 % as a consequence of the installation of the SI system. The 10 % increase will result from higher crop yields due to constant soil humidity at the root zone and a reduction of the labor irrigation cost

472

Arab J Geosci (2016) 9:472

Page 6 of 12

Table 1 Overview of the intervention scenarios

Acronym

Scenario

Short description

BAU SI SI + E

Business as usual Smart irrigation Smart irrigation + efficiency

RD

Recharge dam

SI + E + RD

Smart irrigation + efficiency + recharge dams

Same pumping rates as in 2010 Pumping rates reduced by 20 % Pumping rates reduced by 20 % plus an increase in gross farm profit by 10 % due to improved efficiency of fertilization and irrigation Same pumping rates as in 2010 plus an increase in recharge at aquifer level due to dams Pumping rates reduced by 20 % plus an increase in gross farm profit by 10 % due to improved efficiency plus an increase in recharge due to dams

(Raine et al. 1997; Smith et al. 2005). Scenario 3 is called hereafter SI + efficiency (SI + E). Scenario 4 considers a government intervention through the construction of three recharge dams (RDs) to enhance aquifer recharge. The full cost of the dams’ construction is supported by the government, and there is no mechanism in place that allows any cost recovery from farmers as a result of increased availability of groundwater. The cost of the three dams is estimated to be US$23.4 million (actualized from MWR 2000). Finally, scenario 5 (SI + E + RD) is a combination of scenario 3 and scenario 4 where we evaluate the joint action of introducing SI + efficiency and the construction of the three recharge dams. An overview of the scenarios is shown in Table 1. The cost-benefit analysis is based on the incremental cash flow estimated as the profit obtained in each one of the scenarios minus the profit of the BAU scenario (e.g., Campbell and Brown 2007).

Results and discussion

these observations are highly uncertain as the elevations of the measurement points (usually the top of the well casing) have not been measured with high precision. These elevations appear to have been obtained from a digital elevation model. The resulting observed groundwater levels, calculated from the elevation of the measurement point and the depth from measurement point to water level, may therefore be affected by uncertainties of several meters. Deviations may also be a result of an underrepresentation of subsurface heterogeneity in the model. Since only bore logs from six boreholes were available, information on the spatial heterogeneity of aquifer properties was limited and the model was constructed with a relatively homogeneous distribution of aquifer properties. The simulated drawdown for the post-development state 1974–2010 is generally in good agreement with the overall trend of the observed drawdown (Fig. 4). Seasonal variations and changing short-term trends are not represented by the model. For example, in the years 1995–1997, observed

Water balance components and groundwater levels 1974–2010 The calibrated rate of inflow from the mountains is 71 million m3/year. This is within the same range as the calibrated rate of 68 million m3/year by Walther et al. (2012) for their model in a neighboring catchment of similar size and estimations by Gerner et al. (2012). Direct rainfall recharge sums up to 12 million m3/year, resulting in a total freshwater inflow into the study area of 83 million m3/year. For the time period 1974– 2010, the calibrated specific pumping rate increases from zero in 1974 to 4.3 L/m2/day in the year 2010. This results in a total groundwater pumping rate of 164 million m3/year in 2010. The calibrated water balance components are in the same range as given in MWR (2000) for the region between Barka and Al Suwayq. Simulated groundwater levels of the pre-development state (1974) are in fair agreement with the observed values at the seven observation wells (Fig. 3). There are, however, deviations of 2–3 m at some wells. It must be noted that

Fig. 3 Observed vs. simulated hydraulic heads for the year 1974

Arab J Geosci (2016) 9:472

Page 7 of 12 472

-1.00 -2.00 -3.00 -4.00 -5.00 -6.00 1974

observed simulated 1984

1994

0.00

drawdown (m)

-4.00 observed simulated 1984

1994

drawdown (m)

0.00

-6.00

-10.00 1974

observed simulated

1984

1994

1994

2004

OW4

-2.00 -3.00 -4.00 observed simulated 1984

1994

2004

OW6

-2.00 -4.00 -6.00 -8.00

2004

1984

0.00

OW5

-4.00

observed simulated

-1.00

-6.00 1974

2004

-2.00

-8.00

-4.00

-5.00

drawdown (m)

drawdown (m)

-3.00

-6.00 1974

-3.00

0.00

OW3

-2.00

-5.00

-2.00

-6.00 1974

2004

OW2

-1.00

-5.00

-1.00

-10.00 1974

observed simulated 1984

1994

2004

OW7

0.00 drawdown (m)

0.00

OW1 drawdown (m)

drawdown (m)

0.00

-1.00 -2.00 -3.00 -4.00 -5.00 -6.00 1974

observed simulated

1984

1994

2004

Fig. 4 Observed vs. simulated drawdown at the seven observation wells (OW1–OW7) for the period 1974–2010

groundwater levels increased in all wells because of unusually high amounts of rainfall which presumably have led to increased recharge and decreased water consumption. Furthermore, in some of the wells, a different trend is apparent before 1995 and after 1997, with groundwater levels declining at a higher rate after 1997. This is contrary to the

fact that the size of cultivated area in the Batinah decreased after 1997 (OSS 2012). Decreasing cultivated area should lead to decreasing pumping rates and a lower rate of decline in groundwater levels, yet this is not supported by the data. The reason for the stronger decline after 1997 could also lie in a lower recharge as the successive years were

472

Page 8 of 12

rather dry. Further research into recharge processes in the study area would help uncover the underlying processes, which was beyond the scope of this study. Furthermore, metering of abstraction rates would help clarify the reasons for the observed trends in groundwater levels. Our model is not able to reflect the observed short-term variations and changes in trends because recharge is assumed constant and no changes in pumping regime or cultivated area are considered due to a lack of information.

Salinity 1974–2010 Salinity data from 138 wells in the study area were accessed as measurements of electrical conductivity and were converted to concentrations for comparability, using the relation given in Shammas and Jacks (2007). The results show higher salinities in the eastern part of the study area and closer to the shore (Fig. 5). They also show a high degree of heterogeneity. Some wells at short distances away from each other (<300 m) and at the same distance from the shore have differences in salinity of up to 6 g/L. These observations suggest that the samples were taken at different depths; however, no information on sampling depths was available. Other factors potentially influencing the salinity at specific locations are heterogeneities in aquifer properties or heterogeneities in pumping rates. Unfortunately, no detailed information was available on these factors and the reasons for the observed salinity variations could not be clarified. The simulated salinity at the farms in the study area for the year 2010 is shown in Fig. 5 (overlain with the salinity observations). The increase in salinity from the mountains to the shore is fairly homogeneous as a result of the relatively homogeneous distribution of aquifer properties in the model.

Fig. 5 Simulated salinity at the farms in the study area overlain by measured salinity (dots) in 2010. Colors of the dots of the salinity measurements correspond to the colors of the salinity classes of the farms

Arab J Geosci (2016) 9:472

The boundary between freshwater and low salinity water is at a distance between 1.6 and 2.2 km from the shore at the assumed well depth. This is in agreement with the findings of Al Barwani and Helmi (2006) who presented a map of seawater intrusion in the year 2005 for the Suwayq area showing a similar distance. When comparing salinity measurements with simulated salinity in Fig. 5, it seems that the salinity distribution is much more heterogeneous than the model results suggest. However, as mentioned earlier, no information on the well depths was available, and it is likely that actual well depths are different from the depths that we assumed for this comparison. More information on well depths, aquifer heterogeneity, and actual pumping rates would be needed to understand these deviations. Comparing the range of measured salinities with the range of simulated salinities at the locations of the sampled wells, the total range (minimum to maximum) is represented well in the simulation results (Fig. 6). The interquartile range is also similar. The median from the simulations is lower than the median from the measurements for all years compared, but the rate of increase in median is similar. Figure 7 shows the percentage of farm area in each salinity class calculated from the simulation results. For 1974, the percentage of farm area in the very high salinity class is negligible, and groundwater is classified as fresh at 90 % of the farms in the study area (Fig. 7). This is a reasonable outcome as intense pumping has only started in the 1970s, and it can be assumed that farms have only been established in areas where the groundwater was fresh. Ten percent of farm area is affected by increased salinity, which could be a result of the assumed well depths used for the calculation. It is likely that some wells were shallower than assumed in our model and yielded water with lower salinity. The results for 2010 show a decrease in the

Fig. 6 Measured and simulated salt concentrations at 138 wells in the study area. Box plots show minimum, 25th percentile, median, 75th percentile, and maximum

Arab J Geosci (2016) 9:472

Fig. 7 Simulated percentage of farm area per salinity class for the different scenarios; in terms of salinity, the SI and the SI + E scenarios are the same, so only one bar (SI) is presented for both scenarios

percentage of farm area with fresh groundwater to 74 % and an increase in farm area with very high salinity water to 5 % (Fig. 7). This is comparable to the findings in OSS (2012), where 2 % of total cultivated area in the Batinah was classified as having very high salinity water in 2010. Overall, the rate of intrusion and the range of simulated salinities compare with findings from other studies, but the representation of spatial variability remains unsolved. Due to the poor data quality of the available salinity measurements, the observed salinity distribution is uncertain and we refrained from putting more effort into model calibration to achieve a better fit to observed salinities. A better fit would not necessarily mean a better model performance, and therefore, we accepted that the model performance in terms of salinity could not be verified with high confidence. This study aims at comparing various intervention scenarios in terms of their effects on salinity and subsequent aggregated benefits for the farmers in the region. It does not aim at predicting a detailed spatial

Page 9 of 12 472

distribution of salinity or effects on individual farms. At this level of aggregation, the estimated differences between the scenarios can be interpreted when the differences appear large in relation to model fit and data quality. Another limitation of the study is that the model assumes a constant size and distribution of cultivated area over time. In reality, the size of cultivated area has increased between the 1970s and the 1990s. In the model, this was simulated as increase in pumping rates per farm area with the total size and distribution of cultivated area from the year 2010. No historical data on the distribution of farm area was available, and therefore, this approach was taken. However, fewer wells with higher pumping rates produce a different pattern of salinity compared to more wells with lower pumping rates. Furthermore, it is likely that farms would be abandoned when the salinity gets too high and groundwater abstractions at these locations would cease. Affected farmers would either give up farming entirely or move further inland and start abstracting groundwater at a new location. In OSS (2012), it is discussed that many farms in the Batinah were abandoned since the mid1990s, but a relocation of farms further inland is not discussed. Our model setting does not consider stopping pumping at salinized farms but assumes continued abstraction. The introduction of a moving boundary condition to provide for abandoned farms and the effects on the salinity distribution could be investigated in a future continuation of this study. Salinity distribution for the intervention scenarios The simulation of the BAU scenario resulted in an increase in farm area with very high salinity water to 20 % while the farm area with freshwater decreased to 58 % (Fig. 7). The effect of

Fig. 8 a Comparison of farmers’ gross profit in million US$ per year under the five scenarios. b Present value of gross profit per scenario in million US$

472

Arab J Geosci (2016) 9:472

Page 10 of 12

SI technologies in the SI scenario is noticeable particularly for the very high salinity class where the proportion of farm area is about half that of the BAU scenario. The proportion of farm area with freshwater (65 %) is slightly higher compared to BAU. The three planned recharge dams in the RD scenario lead to clear improvement compared to BAU, with 68 % of the farm area having freshwater and 8 % having very high salinity water. The combination of improved irrigation technologies and recharge dams (SI + RD) results in roughly the same salinity distribution as in the year 2010, which means that seawater intrusion could not be reversed with these interventions, but a further deterioration of water quality could be prevented over the projection period of 30 years. Economic analysis The results of the simulations are presented in Fig. 8a. Out of the five scenarios, only the SI + E + RD scenario has a positive trend. Under scenario BAU, the gross profit will decrease from US$55 million per year in the starting year to US$48 million in year 30 with an average decrease rate of 0.45 % per year. The BAU scenario is a baseline scenario to which the remaining scenarios are compared. The introduction of SI will allow saving 20 % of groundwater which in turn will slow down seawater intrusion as shown in Fig. 7 compared to the BAU. However, the SI scenario represents a deterioration of the salinity compared to the base year 2010, though the deterioration is much less than for the BAU. In terms of gross profit, the SI scenario performs worse than the BAU for the first 7 years and then slightly better than the BAU afterward. This is due to the fact that the gains resulting from seawater intrusion and salinity slowdown are lower than the cost of investment in the SI system. The annual cost of introducing the SI is estimated at US$1.6 million. The return to the investment is lower than the investment cost. This is the reason why scenario SI’s gross profit curve shows below the BAU curve during the first years. Scenario SI + E shows the effect of introducing SI plus improving efficiency at farm level by yield increase and reduction of irrigation labor costs. The results show that this scenario is much better than the BAU. The gross profit starts with US$59.9 million and ends with US$54.8 million, a figure higher than the BAU US$48 million. The difference between SI and SI + E reflects the importance of farmers’ adjustment to the new technology and improvements in crop yields which are expected to take place because of the homogeneous soil humidity over a crop cycle. The results of scenario RD are quite illustrative about the benefits from recharge dams’ construction. The gross profit is higher than in case of BAU. It starts from US$55.8 million and reaches US$54.7 million. The RD is thus a superior solution to SI if we assume that the increase in water storage in the aquifer is not accompanied by an increase in pumping. This is actually not guaranteed as experience has shown that in the

past 20 years despite the construction of recharge dams and the introduction of drip/sprinklers, the groundwater deficit has not decreased (Zekri 2009; Zekri and Al Mamary 2014). Scenario SI + E + RD combines the effects of SI, efficiency improvements at farm level, and the construction of recharge dams. This scenario has the highest gross profit. The starting year’s gross profit is US$58.2 million, and the final year’s gross profit is US$61.4 million. This scenario stabilizes farmers’ profit over the 30-year period. From an economical perspective, farmers’ utility is slightly increasing over time (0.2 %/year), which could be interpreted as sustainable. Furthermore, the salinity status as shown in Fig. 7 is the best of all scenarios. However, this scenario does not respect the sustainability criteria, interpreted as the flow of natural resources is not declining over time. In fact, compared to the salinity status of 2010 (see Fig. 7), the situation is worsening as the proportion of farm area with freshwater declines from 74 % in 2010 to 72 % at the end of the year 2040. Figure 8b shows the present value (PV) of each one of the scenarios. The PV is the discounted cash flow of gross profit of each scenario minus the investment costs minus the gross profit of the BAU scenario discounted at 5 % rate. It is the added value of the scenario compared to the option of BAU Bdoing nothing.^ The PVallows classifying the scenarios. The higher the PV, the better is the project or scenario. All the four possible scenarios are feasible and have positive PV. The results show as expected that the best alternative corresponds to scenario SI + E + RD where building the recharge dams is accompanied by the introduction of SI and an improved efficiency at farm level. For this scenario, the PV reaches US$141 million over 30 years. The least profitable scenario is the SI with a PV of UD$14 million. This indicates clearly that relying on irrigation efficiency improvement alone is not interesting from an economic point of view. The current policy is in fact based most on irrigation efficiency improvement via subsidization of drip/sprinklers. The figure shows that if the SI is accompanied by an efficiency improvement at farm level, then the SI + E would be the second best scenario with US$93 million over 30 years. The scenario on recharge dams RD would result on a PV of US$56 million, which is almost half of the SI + E. Thus, from a policy perspective, it is better to combine the recharge dams with the introduction of SI systems which will help improve efficiency at farm level and produce the highest PV. SI alone or RD alone is not the best options. The classical subsidy of the drip/sprinkler is not recommended neither because, as mentioned above, there is no tangible positive impact on salinity control.

Conclusions In this study, we simulated the effect of different intervention scenarios on seawater intrusion in an arid area coastal aquifer.

Arab J Geosci (2016) 9:472

We showed how continued pumping in a business-as-usual scenario leads to a severe deterioration of groundwater quality during the simulation period of 30 years. Management options such as the introduction of SI technologies to reduce water consumption and the construction of groundwater recharge dams showed to significantly alleviate the effects as they lead to an enrichment of the available water resources and therefore counteract further seawater intrusion. However, for our study area of Suwayq, the combined effects of reduced water consumption and increased recharge could only maintain the water quality distribution of the starting year in our simulations and thus could only prevent a further deterioration instead of improving the water quality. Similarly, a combination of SI, higher efficiency, and recharge dams could stabilize the farmers’ profit over the simulation period and produce the highest PV of the studied scenarios, but again, none of the scenarios leads to an increase in farmers’ profit during the simulation period. Although the groundwater model is affected by considerable uncertainties due to the scarcity of available data, this exercise combining groundwater modeling with economic analysis gives useful insights into the general behavior of the system and clear indications of the future effects of decisions made on the use and management of the groundwater resources in the study area. The results of this study indicate that the problem of seawater intrusion needs to be tackled by a comprehensive, integrated strategy as a few simple (yet expensive) measures such as more efficient water use and enhanced recharge by dams may not suffice. Farmers are required to improve their efficiency by at least 10 %, for the seawater intrusion to be halted. Exogenous institutions such as quota setting and penalties on excessive groundwater use will be further investigated.

Acknowledgments The authors would like to thank the Ministry of Regional Municipality and Water Resources for providing rainfall, groundwater, and borehole data for the study site. This work was supported by The Research Council of Oman (Open Research Grant RC/ AGR/ECON/12/01).

References Abdalla OAE, Al-Rawahi AS (2013) Groundwater recharge dams in arid areas as tools for aquifer replenishment and mitigating seawater intrusion: example of Al Khod, Oman. Environ Earth Sci 69: 1951–1962. doi:10.1007/s12665-012-2028-x Al Barwani A, Helmi T (2006) Sea water intrusion in a coastal aquifer: a case study for the area between Seeb and Suwaiq, Sultanate of Oman. J Agric Mar Sci 11:55–69 Baharuddin MFT, Taib S, Hashim R, et al. (2012) Assessment of seawater intrusion to the agricultural sustainability at the coastal area of Carey Island, Selangor, Malaysia. Arab J Geosci 6:3909–3928. doi:10. 1007/s12517-012-0651-1

Page 11 of 12 472 Campbell H, Brown R (2007) Benefit-cost analysis: financial and economic appraisal using spreadsheets. Cambridge University Press, Cambridge, UK Gaaloul N, Pliakas F, Kallioras A, et al. (2012) Simulation of seawater intrusion in coastal aquifers: forty five-years exploitation in an Eastern Coast aquifer in NE Tunisia. Open Hydrol J 6:31–44 Gerner A, Schütze N, Schmitz GH (2012) Portrayal of fuzzy recharge areas for water balance modelling—a case study in Northern Oman. Adv Geosci 31:1–7. doi:10.5194/adgeo-31-1-2012 Grundmann J, Schütze N, Lennartz F (2013) Sustainable management of a coupled groundwater-agriculture hydrosystem using multi-criteria simulation based optimisation. Water Sci Technol J Int Assoc Water Pollut Res 67:689–698. doi:10.2166/wst.2012.602 Grundmann J, Schütze N, Schmitz G, Al-Shaqsi S (2012) Towards an integrated arid zone water management using simulation-based optimisation. Environ Earth Sci 65:1381–1394. doi:10.1007/s12665011-1253-z Hanson RT (2003) Geohydrologic framework of recharge and seawater intrusion in the Pajaro Valley, Santa Cruz and Monterey Counties, California. US Department of the Interior, US Geological Survey Kalbus E, Kalbacher T, Kolditz O, et al. (2012) Integrated water resources management under different hydrological, climatic and socioeconomic conditions. Environ Earth Sci 65:1363–1366. doi:10. 1007/s12665-011-1330-3 Kallioras A, Pliakas F-K, Schuth C, Rausch R (2013) Methods to countermeasure the intrusion of seawater into coastal aquifer systems. In: SK S, Sanghi R (eds) Wastewater Reuse and Management. Springer, Netherlands, pp. 479–490 Kloss S, Pushpalatha R, Kamoyo KJ, Schütze N (2012) Evaluation of crop models for simulating and optimizing deficit irrigation systems in arid and semi-arid countries under climate variability. Water Resour Manag 26:997–1014. doi:10.1007/s11269-011-9906-y Langevin CD, Thorne DT, Dausman AM, et al (2008) SEAWAT version 4: a computer program for simulation of multi-species solute and heat transport. U. S. Geological Survey Techniques and Methods Book 6, Chapter A22 Lee J-Y, Song S-H (2006) Evaluation of groundwater quality in coastal areas: implications for sustainable agriculture. Environ Geol 52: 1231–1242. doi:10.1007/s00254-006-0560-2 López Geta JA, de Dios Gomez J, de la Orden JA, et al (2003) Coastal aquifers intrusion technology, Mediterranean countries. Publicaciones del Instituto Geologico y Minero de Espana, Serie: Hidrogeologia y Aguas Subterraneas No. 8 MAF (2014) Agricultural Census 2013–14. Ministry of Agriculture and Fisheries, Sultanate of Oman Melloul AJ, Goldenberg LC (1997) Monitoring of seawater intrusion in coastal aquifers: basics and local concerns. J Environ Manag 51:73– 86. doi:10.1006/jema.1997.0136 MRMWR (2006) Drilling and aquifer testing in the northern Batinah, Musandam governorate and Northern Al Massarat. Final report for the Northern Batinah. Prepared by Geo-Resources Consultancy for the Ministry of Regional Municipalities and Water Resources, Oman MWR (2000) National Water Resources Master Plan. Prepared by Binnie & Partners and Arcadis Euroconsult for the Ministry of Water Resources, Sultanate of Oman OSS (2012) Oman salinity strategy. Ministry of Agriculture and fisheries. Sultanate of Oman & International Center For Biosaline Agriculture, UAE Raine SR, McClymont DJ, Smith RJ (1997) The development of guidelines for surface irrigation in areas with variable infiltration. In: Proceedings of Australian Society of Sugar Cane Technologists. pp 293–301 Rajmohan N, Al-Futaisi A, Jamrah A (2007) Evaluation of long-term groundwater level data in regular monitoring wells, Barka, Sultanate of Oman. Hydrol Process 21:3367–3379. doi:10.1002/ hyp.6543

472

Page 12 of 12

Schütze N, Kloss S, Lennartz F, et al. (2012) Optimal planning and operation of irrigation systems under water resource constraints in Oman considering climatic uncertainty. Environ Earth Sci 65:1511– 1521. doi:10.1007/s12665-011-1135-4 Shammas MI, Jacks G (2007) Seawater intrusion in the Salalah plain aquifer, Oman. Environ Geol 53:575–587. doi:10.1007/s00254007-0673-2 Smith RJ, Raine SR, Minkevich J (2005) Irrigation application efficiency and deep drainage potential under surface irrigated cotton. Agric Water Manag 71:117–130. doi:10.1016/j.agwat.2004.07.008 Stanger G (1985) Coastal salinization: a case history from Oman. Agric Water Manag 9:269–286. doi:10.1016/0378-3774(85)90038-1 Tanaka K (1975) Measurements of self-diffusion coefficients of water in pure water and in aqueous electrolyte solutions. J Chem Soc Faraday Trans 1 Phys Chem Condens Phases 71:1127–1131. doi:10.1039/ F19757101127 Walther M, Bilke L, Delfs J-O, et al. (2014) Assessing the saltwater remediation potential of a three-dimensional, heterogeneous, coastal aquifer system. Environ Earth Sci:1–11. doi:10.1007/s12665-0143253-2 Walther M, Delfs J-O, Grundmann J, et al. (2012) Saltwater intrusion modeling: verification and application to an agricultural coastal arid

Arab J Geosci (2016) 9:472 region in Oman. J Comput Appl Math 236:4798–4809. doi:10. 1016/j.cam.2012.02.008 Werner AD (2009) A review of seawater intrusion and its management in Australia. Hydrogeol J 18:281–285. doi:10.1007/ s10040-009-0465-8 Werner AD, Bakker M, Post VEA, et al. (2013) Seawater intrusion processes, investigation and management: recent advances and future challenges. Adv Water Resour 51:3–26. doi:10.1016/j.advwatres. 2012.03.004 Zalidis G, Stamatiadis S, Takavakoglou V, et al. (2002) Impacts of agricultural practices on soil and water quality in the Mediterranean region and proposed assessment methodology. Agric Ecosyst Environ 88:137–146. doi:10.1016/S0167-8809(01)00249-3 Zekri S (2008) Using economic incentives and regulations to reduce seawater intrusion in the Batinah coastal area of Oman. Agric Water Manag 95:243–252. doi:10.1016/j.agwat.2007.10.006 Zekri S (2009) Controlling groundwater pumping online. J Environ Manag 90:3581–3588. doi:10.1016/j.jenvman.2009.06.019 Zekri S, Al Mamary S (2014) Regional Initiative on Water Scarcity in Oman. National Report for the FAO Programme BRegional Initiative on Water Scarcity in the Near East.^ Ministry of Agriculture and Fisheries, Sultanate of Oman