1 PREDICTION OF SALTWATER INTRUSION IN JENEPONTO REGENCY INDONESIA

2 USING ANALYTICAL METHOD AND NUMERICAL MODEL

3
1) 1) 1) 1) 1)
4 Abdul Rivai Suleman , Sugiarto Badaruddin , Indra Mutiara , Zulvyah Faisal Muh Taufik Iqbal
2)
5 Muhammad Ihsan

6
1
7 Civil Engineering Department, Politeknik Negeri Ujung Pandang, P.O. Box 90245, South Sulawesi,
8 Indonesia
2
9 Civil Engineering Department, Sekolah Tinggi Teknik Baramuli, Pinrang, Souh Sulawesi, Indonesia
10

11

12 *Corresponding author at: Civil Engineering Department, Politeknik Negeri Ujung Pandang, PO Box
13 90245, South Sulawesi, Indonesia. Email address: rivai.suleman@poliupg.ac.id (Abdul Rivai
14 Suleman)
15

16 Email addresses:
17 Abdul Rivai Suleman: rivai.suleman@poliupg.ac.id
18 Sugiarto Badaruddin: sugibadaruddin@poliupg.ac.id
19 Indra Mutiara: indramutiara@poliupg.ac.id
20 Zulvyah Faisal: zulvyahfaisal@poliupg.ac.id
21 Muh Taufik Iqbal: muhtaufikiqbal@poliupg.ac.id
22 Muhammad Ihsan: muhammadihsan@alumni.ait.asia
23
24
25
26 ABSTRACT
27

28 This paper for the first time evaluates the extent of saltwater intrusion (SWI) in coastal aquifers in two
29 sub-districts in Jeneponto Regency Indonesia employing analytical method in the form of sharp-
30 interface approach and numerical dispersive simulation of finite difference solution developed by
31 SEAWAT, based on field observation and secondary data. In general, the extents of SWI obtained
32 from the analytical solution for four aquifers in the region are larger relative to the results of numerical
33 solutions. Accordingly, the results show that the maximum SWI extent occurs in the aquifer of
34 Binamu2, where the analytical solution and numerical simulation produced SWI extents of 850.4 m
35 and 510 m, respectively. The finding can be explained by taking into account the pushing seaward
36 effect induced by the existence of circulating flow inside the saltwater wedge below the freshwater-
37 saltwater interface which is considered in the numerical solution but excluded in the analytical
38 solution. The result also showed that the SWI conditions in Jeneponto Regency are still in reasonable
39 limits but groundwater exploitation control is highly required in the region to retain this situation.
40

41 Keywords: Saltwater intrusion, numerical modelling, analytical solution, finite

42 difference, freshwater and saltwater interface
43
44 1. INTRODUCTION
45

46 As one of alternatives of water resources, groundwater has shown a potential to be exploited in large
47 quantities, with constraints where the amount of extraction cannot exceed the amount of water
48 entering the aquifer. Consequently, the use of groundwater is mandatory to be in accordance with the
49 yield capacity of aquifers where the exploitation must be regulated by a set of policies in national and
50 regional levels.
51

52 The coastal area is topographically lowland with morphological form of coastal plain. Geologically,
53 the composing rock is generally in the form of alluvial deposits consisting of clay, sand and gravel
54 resulting from the transport and erosion of rocks in river upstream areas. Generally, rocks in the plain
55 are less compact, resulting in the idea of reliable potential of groundwater. Favourable coastal aquifers
56 are generally in the form of confined aquifers, despite unconfined aquifers can be a good source of
57 groundwater especially in coastal areas. The major problems in the coastal area are the diversity of
58 aquifer systems, the position and distribution of seawater both naturally and artificially due to the
59 extraction of groundwater for domestic, agricultural and industrial needs. On coastal aquifers, changes
60 in hydrogeology of coastal areas can induce seawater movement towards land potentially
61 contaminating groundwater in aquifers and this condition is widely known as saltwater intrusion
62 (SWI).
63

64 Historically, the occurrence of SWI is generally caused by excessive pumping or extraction of

65 groundwater on land and this can lead to significant losses in groundwater availability in coastal
66 aquifers around the world (Badaruddin et al., 2015; FAO, 2007). However, the effects of climate
67 change (e.g. sea level rise and a decrease in groundwater recharge quantities) can also lead to the
68 occurrence of SWI (Post, 2005). Therefore, the vulnerability of coastal aquifers to climate change,
69 increased volume of groundwater extraction and sea level rise must be considered integrally in
70 groundwater management.
71

72 The SWI is basically a complex process involving variable-density flow, transport of solution and
73 hydrochemical processes (Werner et al., 2012), which makes assessing groundwater relatively
74 difficult and expensive. As a result, large-scale assessments of coastal aquifer susceptibility to SWI
75 generally only use qualitative methods such as GALDIT (Lobo-Ferreira et al., 2007) and CVI (Ozyurt,
76 2007), which only considers a number of factors that are considered to have an impact on SWI. In
77 addition, these methods are generally lacking in theoretical grounds and subjectively are more focused
 78 on the selection of only one element that is related to the SWI. Recently, an alternative prediction of
79 large-scale SWI has been developed by Werner et al. (2012). This method is based on steady-state
80 conditions of Strack equation (1976) which assumes that the interface of saltwater and freshwater in
81 an aquifer is a sharp-line, so that this method involves physical mechanics of the SWI, even though
82 under ideal conditions. Whereas to get conditions that are close to real conditions in predicting SWI,
83 numerical modeling, which assumes that the contact between saltwater and freshwater in an aquifer is
84 in the form of a mixing dispersive zone, is required.
85

86 In the last decade there has been a rapid population growth throughout the world, including in
87 Indonesia, and this has led to a massive exploitation of groundwater. This phenomenon has caused a
88 negative impact on quantity and quality of groundwater, including decreasing of groundwater levels,
89 increasing fluctuations and decreasing of groundwater quality, as well as SWI in several regions.
90 Thus, a real and integrated effort is needed to minimize these negative impacts, either by the
91 government, the public and the private sector.
92

93 In this study, we applied analytical methods from Werner et al. (2012) and numerical dispersive
94 model in predicting SWI in two sub-districts in Jeneponto Regency. For the first time, analytical
95 methods and 2-dimensional numerical modelling were applied in determining the extent of
96 saltwater intrusion in this regency. Jeneponto Regency is one of the districts in South Sulawesi,
97 Indonesia whose part of boundaries are coastal region with high possibility to experience an SWI
98 due to long time extraction of groundwater for domestic and irrigation purposes in the region
99 (Syamsuddin et al., 2009). Considering the aforementioned conditions it is important to carry out
100 preliminary assessments of SWI condition in this area and is considered necessary to understand
101 the causes of saltwater intrusion which will be beneficial for local decision maker to develop
102 groundwater resources protection policy.
103

104 2. METHODOLOGY
105

106 3.1. Conceptual Model

107

108 In this study, determination of SWI length was carried out only in 4 (four) cross sections in
109 locations as shown in Fig. 1 and only under steady-state conditions. Due to the limitations of
110 available hydrological and hydrogeological data (e.g. aquifer thickness, soil stratigraphy,
111 recharge and specific storage), simplification was carried out on several hydrogeological data but
112 certainly still considered secondary data from previous study (i.e., Syamsuddin et al., 2009).
113 Since there are no detailed boring log data available that can provide a comprehensive
114 description of the stratigraphic conditions of the soil in the study area, it is assumed that the
115 aquifer type in the study site is an unconfined aquifer based on the study of Syamsuddin et al.
116 (2009). Table 1 provides the hydrogeological data used in the SWI analysis of this study.
117

118

119
120 Figure 1. Observation lines of SWI (see the yellow lines).
121

122 Table 1. Hydrogeology data of aquifers employed in numerical models

Aquifers
Parameter
Binamu1 Binamu2 Arungkeke1 Arungkeke2
K (m/d) 5.80 5.80 5.80 5.80
hs (m) 68 68 64 64
GWL hf (m) 3.6 2.0 9.5 8.7
xf (m) 5000 4800 3300 3200
n (-) 0.46 0.46 0.46 0.46
Sy (-) 0.32 0.32 0.32 0.32
L (m) 4 4 4 4
T (m) 0.4 0.4 0.4 0.4
Dm (m2/d) 8.6 x 10-5 8.6 x 10 -5
8.6 x 10 -5
8.6 x 10-5
 (-) 0.025 0.025 0.025 0.025
Wnet (mm/y) 56.70 56.70 56.70 56.70
123
124 3.2. Analytical Solution
125

126 The analytical solution for locating the contact between seawater and freshwater in the fixed
127 flow conditions developed by Strack (1976) is the basis of the method described by Werner et al.
128 (2012) and will be summarized here. Fig. 2 shows the conceptual model for unconfined and
129 confined aquifers.
130

131
132 Figure 2. Description of hydrogeology parameters for (a) unconfined aquifers and (b)
133 confined aquifers, based on the theory of Strack (1976).
134

135 The quantities of water in the aquifer domain as shown in Fig. 2 consists of net groundwater
136 recharge (Wnett [L/T]), which takes into account infiltration, evapotranspiration and pumping
137 distributions, fresh water drained into the sea (q0 [L2/T]), and the lateral flow in aquifers from
138 inland (qb [L2/T]). Hydraulic pressure head (hf [L]) (groundwater table for unconfined aquifer
139 conditions) relates to the depth of the location of the interface line of saltwater and freshwater ( z
140 [L]) explained in the Ghyben-Herzberg relation; z = hf/. Here,  [-] is the dimensionless ratio of
141 density  = (s - f)/f, and s (= 1025 kg/m3) and f (= 1000 kg/m3) is the density of saltwater
142 and freshwater, respectively, resulting = 0.025. The thickness of fresh water is h [L], the bottom
143 of the aquifer is z0 [L] below the average sea level and the hydraulic conductivity is K [L/T].
144

145 In analyzing the SWI for unconfined aquifers, Equations (1) (Fig. 2a) and (2) (Fig. 2a) are used
146 as below:
147

2q0 x  Wnet x 2
hf   (1   ) z0  z0
2
148 Zone 1 (x ≥ xt) (1)
K
149

150 and,
151

   2q0 x  Wnet x
2
152 Zone 2 (x ≤ xt) hf    (2)
1   K
153

154 From Cheng and Ouazar (1999), the farthest position of SWI (xt [L]) is obtained as shown by
155 Equation (3):
156

2
q0  q0  K (1   ) z0 2
157 xt      (3)
Wnet  Wnet  Wnet

158

159

160 For confined aquifers, Equation (4) (Fig. 2b) and (5) (Fig. 2b) are used as below
161

x2
q0 x  Wnet
162 Zone 1 (x ≥ xt) hf  2  z  h0 (4)
0
Kh0 2
163

164 and,
165


166 Zone 1 (x ≤ xt) hf  2q x  W
0 net x2 
K
 z0  h0 (5)

167

168 From Cheng and Ouazar (1999), the maximum length of SWI (xt [L]) is obtained as shown by
169 Equation (6):
170

Kh0 2
171 xt  (6)
2q0
172

173 3.3. Numerical Model

174
175 To validate the results of an analytical calculation and to get a more real picture, numerical
176 modeling is usually required in predicting the SWI under numerical-dispersive solutions. In this
177 case, a two-dimensional modeling is used using SEAWAT version 4 which is used for flow with
178 variations in density and transport of solutions. This program employs a finite difference method
179 with limitation of modelling only for saturated flow. The description of the numerical method
180 and the equations used in SEAWAT can be found in Guo and Langevin (2002) and Langevin et
181 al. (2008).
182

183 The conceptualization of aquifers from the four research locations used in the numerical model is
184 configured as separate aquifers for Binamu1, Binamu2, Arungkeke1 and Arungkeke2 aquifers
185 and modelled in two dimensions and cross sections perpendicular to the coast, as shown in Fig.
186 3. In this study, all representations in the numerical model including the boundary conditions is
187 assumed to be a head controlled condition (Werner and Simmons, 2009) and the energy level is
188 considered constant even though there is the effect of pumping groundwater.
189

190 The domain model is uniformly discretized in carrying out steady-state simulations for the four
191 aquifers, where for the Binamu1 aquifer uses 500 vertical columns with a width of 10 m and 76
192 horizontal layers with a thickness of 1 m, the Binamu2 aquifer uses 480 vertical columns with a
193 width of 10 m and 76 1 m thick horizontal layer. Arungkeke1 aquifer uses 330 vertical columns
194 with a width of 10 m and 78 horizontal layers with a thickness of 1 m, while the Arungkeke2
195 aquifer uses 320 vertical columns and 78 horizontal layers with 1 m thick. This discretization is
196 consistent with the Peclet number which is smaller than 4, as recommended by Voss and Souza
197 (1987) to reduce numerical oscillation. The condition of particular energy head is assumed to be
198 at the groundwater boundary condition on land and the constant concentration condition is
199 assumed to be at the coastal boundary conditions, with a concentration of saltwater of 35 kg/m 3.
200 Preconditioned Conjugate-Gradient 2 (PCG2) and General Conjugate Gradient (GCG) are used
201 as a solution for solution flow and transport equations. The differential scheme is used for
202 advection solutions with Courant numbers of 0.75. The use of Courant numbers smaller or equal
203 to 1 is usually needed to limit the occurrence of numerical disperse in order to achieve more
204 accurate results (Zheng and Bennet, 2002).
205
206
207 Figure 3. Conceptualization of aquifers used in numerical modeling of SWIs for: (a) Binamu1
208 aquifer, (b) Binamu2 aquifer, (c) Arungkeke1 aquifer, and (d) Arungkeke2 aquifer.
209

210 3. RESULTS AND DISCUSSION

211

212 3.1. Groundwater level measurement in the areas.

213

214 Table 2 presents data on water level in four observation points in Binamu and Arungkeke sub-
215 districts in Jeneponto Regency. Based on water level observation data, it is noticed that the
216 groundwater level at the observation site is quite varied which may be caused by heterogeneous
217 conditions of the porous medium. In this study, soil heterogeneity of aquifer was ignored because
218 of financial constraints in acquiring these data.
219

220 Table 2. Groundwater level (GWL) at the observation sites (groundwater level is measured
221 from mean sea level)

Sub- Coordinates
Aquifer code GWL (m)
districts
Latitude Longitude
Binamu1/SDJP273 Binamu 539’25’’ 11943’52’’ 3.6
Binamu2/SDJP54 Binamu 539’33.5’’ 11943’46’’ 2.0
Arungkeke1/AK1/TP1 Arungkeke 539’28.7’’ 11947’42’’ 9.5
Arungkeke2/AK2/TP2 Arungkeke 539’2.4’’ 11948’10’’ 8.7

222
223 3.1.1. SWI results from analytical method
224

225 The hydrogeological parameters listed in Table 3 together with Equations (1), (2), (3), and (6)
226 are used in estimating the amount of groundwater flow discharge into the sea and also the
227 theoretical length of SWI in steady-state conditions at each the aquifers studied are shown in
228 Table 3. These results represent the theoretical conditions of SWI for an extensive duration
229 (steady-state) based on recent hydrological and hydrogeological parameters.
230

231 Table 3. Results of SWI calculations based on analytical solutions.

SWI extents
Sub- Coordinates from
Aquifer code
districts coastline (xT)
Latitude Longitude (m)
Binamu1/SDJP273 Binamu 539’25’’ 11943’52’’ 607.0
Binamu2/SDJP54 Binamu 539’33.5’’ 11943’46’’ 850.4
Arungkeke1/AK1/TP1 Arungkeke 539’28.7’’ 11947’42’’ 234.8
Arungkeke2/AK2/TP2 Arungkeke 539’2.4’’ 11948’10’’ 250.6

232

233 3.1.2. SWI results from numerical model

234

235 For the parameters listed in Table 2, the steady-state conditions for Binamu1, Binamu2,
236 Arungkeke1 and Arungkeke2 aquifers are achieved using an extensive duration of 300 years,
237 until the end of the SWI is in a stable position (no change in position during the duration
238 observation). For the four aquifers observed, the value of xt obtained was smaller than the results
239 obtained from analytical solutions. A larger SWI prediction by analytical solutions (thin line
240 method) is expected to occur (for example from the case of Pool and Carrera, 2011). This is
241 because the intrusion boundaries will be suppressed towards the sea due to flow circulation in th e
242 interface zone between freshwater and saltwater due to mixing between the both which in this
243 case is considered in numerical methods and is conversely ignored in the analytical solutions.
244

245 As explained earlier, the parameter values used in the SEAWAT numerical simulation are shown
246 in Table 2. For the purposes of determining the distance of intrusion and comparison with
247 analytical solutions, the length of SWI (xt) is defined as the intersection of 50% isochlor
248 concentration line and the bottom of the aquifer. However, the 5% and 95% isochlor
249 concentrations lines are still shown as additional information. As seen in Fig. 4, the value of xt
250 obtained for Binamu1, Binamu2, Arungkeke1 and Arungkeke2 aquifers is 390, 510, 160, and
251 190 m, respectively. In contrast to Arungkeke1 and Arungkeke 2, the length of the SWI is almost
252 the same for Binamu1 and Binamu2, although the observation locations are only about hundreds
253 of meters apart, but the xt value obtained from numerical solutions and analytical solutions is
254 notably different. This is correlated with the results of observations of water levels that are
255 notably different between Binamu1 and Binamu2 aquifers (i.e. 1 m) as obtained from field
256 observations, which is most likely due to the exclusion of soil heterogeneity effects on SWI.
257

258
259 Figure X. Numerical simulation results for SWI prediction for (a) Binamu1 aquifer, (b)
260 Binamu2 aquifer, (c) Arungkeke1 aquifer, and (d) Arungkeke2 aquifer, under steady-state
261 condition.
262

263 4. CONCLUSION
264
265 Some conclusions can be drawn from the results. One is that varied water levels in adjacent
266 locations; for example, GWL in Binamu1 and Binamu2 differs to 1 m despite of its proximity for
267 only about hundreds of meters. This is most likely due to the heterogeneity of the soil layer at the
268 study site where the data is yet exactly undetermined. The results of this study also reveal that the
269 maximum SWI occurs in Binamu1 aquifers where analytical solutions and numerical simulations
270 produce SWI lengths of 850.4 m and 510 m, respectively.
271
272 In general, from the results of the study, the prediction results of SWI are considerably different on
273 both analytical and numerical solutions where the value of SWI from the analytical method is
274 greater than the value of SWI from numerical models. This is due to differences in the review of
275 the two methods, where the presence of a mixing zone between saltwater and freshwater calculated
276 in numerical methods which is ignored in analytical solutions, giving a pushing seaward effect
277 (due to the influence of flow circulation in the mixing zone) on SWI boundaries.
278

279 Although there are differences between analytical and numerical solutions, it is obvious that the
280 results of analytical solutions are still functional in the initial assessment of the SWI due to its
281 rapid results. Unlike numerical solutions, it takes longer to prepare the resources needed to perform
282 modeling simulations in order to predict SWI (e.g. software and domain model setup). It can also
283 be concluded that the availability of secondary data in the study area can be considered to be very
284 minimal, especially data on hydrogeological conditions that have not been explored in detail and
285 thoroughly. This is important for further investigation to obtain more accurate and comprehensive
286 results in predicting SWI.

287

288 REFERENCES
289

290 Badaruddin, S., A. D. Werner, et al., 2015. Water table salinization due to seawater intrusion. Water
291 Resources Research.
292 Cheng, A.H.D., Ouazar, D., 1999. Analytical solutions, in: Bear, J., Cheng, A.H.D., Sorek, S.,
293 Ouazar, D., Herrera, I. (Eds.), Seawater intrusion in coastal aquifers – concepts, methods and
294 practices. Kluwer Academic Publishers, Dordrecht, Netherlands.
295 FAO, 1997. Seawater intrusion in coastal aquifers: Guidelines for study, monitoring and control,
296 FAO Water Reports no. 11. Food and Agriculture Organization (FAO) of the United Nations,
297 Rome, Italy.
298 Guo, W., Langevin, C., 2002. User's guide to SEAWAT: A computer program for the simulation of
299 three-dimensional variable-density ground-water flow: USGS Techniques of Water Resources
300 Investigations, Book 6, Chapter A7.
301 Irham, M. N., R. T. Achmad, S. Widodo. 2006. Mapping of saline groundwater on deep aquifer in
302 lower Semarang region. Berkala Fisika, Vol. 9 No. 3.
303 Langevin, C.D., Thorne, D., Dausman, A.M., Sukop, M.C., Guo, W., 2008. SEAWAT Version 4: A
304 computer program for simulation of multi-species solute and heat transport: USGS Techniques
305 and Methods, Book 6, Chapter A22
306 Lobo-Ferreira, J.P., Chachadi, A.G., Diamantino, C., Henriques, M.J., 2007. Assessing aquifer
307 vulnerability to seawater intrusion using the GALDIT method: part 1 - application to the
308 Portuguese Monte Gordo aquifer, In proceedings (ed. J.P. Lobo Ferreira, J.M.P. Viera) Water
309 in Celtic Countries: Quantity, Quality and Climate Variability, IAHS Publication 310,
310 International Association of Hydrological Sciences, Wallingford: 161-171.
311 Ozyurt, G. 2007. Vulnerability of Coastal Areas to Sea Level Rise: A Case Study on Goksu Delta,
312 Masters Thesis, Department of Civil Engineering, Middle East Technical University, Ankara,
313 Turkey.
314 Pool, M. and Carrera, J., 2011. A correction factor to account for mixing in Ghyben-Herzberg and
315 critical pumping rate approximations of seawater intrusion in coastal aquifers. Water Resources
316 Research, 47.
317 Post, V., 2005. Fresh and saline groundwater interaction in coastal aquifers: Is our technology ready
318 for the problems ahead? Hydrogeology Journal 13: 120-123.
319 Strack, O.D.L., 1976. Single-potential solution for regional interface problems in coastal aquifers,
320 Water Resources Research 12: 1165-1174.
321 Syamsuddin, A. Munir, A. Thaha, 2009. Simulation of ground water level fluctuations in the coastal
322 area of Jeneponto. Thesis. Universitas Hasanuddin. Makassar.
323 Werner, A. D., J. D. Ward, et al., 2012. Vulnerability indicators of sea water intrusion. Ground Water
324 50 (1): 48-58.
325 Zheng, C., Bennett, G.D., 2002. Applied contaminant transport modeling, 2nd edition. Wiley
326 Interscience, New York.
327