Introduction

Groundwater is an essential and valuable water resource in many developing countries, especially in cities where there is pressure on public water supply because of insufficient infrastructure. In many areas, groundwater can be used easily and inexpensively, and it normally has good quality with respect to human health and industrial purposes. Population, industrial and economic growth are often significant, with consequent increases in groundwater exploitation to meet demand. Thus, a sustainable use of groundwater resources is required urgently in many cases, taking account of the regional groundwater flow dynamics (Kagabu et al. 2013). Some subsurface environmental problems have been reported that are the result of excessive groundwater pumping such as increasing underground temperature (Yamano et al. 2009; Taniguchi et al. 2009), groundwater drawdown and land subsidence (Foster and Chilton 2003), and groundwater renewal (Yamanaka et al. 2011; Kagabu et al. 2013).

In some large Asian cities, groundwater level drawdowns have been reported (Taniguchi et al. 2008; Yamanaka et al. 2011; Kagabu et al. 2011, 2013). It is also suggested that the degree of groundwater drawdown depends on the development stage of each city (Taniguchi et al. 2008; Onodera et al. 2009; Kagabu et al. 2011). In Indonesia, significant groundwater drawdown, and consequently land subsidence and even groundwater quality changes, have been reported in cities such as Jakarta, Bandung, and Semarang (Wirakusumah and Danaryanto 2004; Taniguchi et al. 2008; Umezawa et al. 2008; Hosono et al. 2011, Kagabu et al. 2011, 2013).

In the Bandung basin, groundwater pumping has been recorded since the early 1900s and found to be increasing (unpublished report, 1991) by Sutrisno, see Table 1). Since 1970, many textile factories have been established in this basin. These factories rely almost entirely on groundwater and have caused significant groundwater drawdowns. In addition, 60% of the total water demand of the Bandung basin is supplied by groundwater (Wangsaatmaja et al. 2006). Consequently, land subsidence has occurred at a rate of 5–75 cm between 2000 and 2008 (Abidin et al. 2008). According to calculations by Taufiq (2010), the groundwater drawdown has contributed to 30–70% of the land subsidence rate.

Table 1 The unpublished reports/papers featured in this report
Full size table

Recent studies have emphasized the impact of subsurface environmental problems, but in this study the focus is on describing the processes that leading to these impacts. Understanding and predicting these processes at basin scale entails developing and integrating knowledge of hydrogeochemical characteristics, aquifer heterogeneity, and field-scale parameterizations, and then simulating a groundwater flow model (Dai et al. 2010, 2014). One of the impacts of excessive groundwater pumping is to cause changes in the groundwater rejuvenation process. Understanding these changes is an important task for effective groundwater management because such changes can cause unexpected problems such as groundwater quality degradation and the disturbance of regional groundwater flow systems (Kagabu et al. 2013).

This study describes the rejuvenation processes of the Bandung basin by using multiple hydrogeochemical techniques (stable isotopes, major ions, and CFC-12 analyses) and estimating a rejuvenation ratio (R) for 14C activity, and then testing this with numerical modeling. The objectives are to systematically examine the groundwater flow system, to trace the rejuvenation processes, and to reveal the groundwater fluxes, thus demonstrating how understanding the rejuvenation phenomenon is important for management of some human activity with respect to groundwater resource protection.

Study area

The Bandung-Soreang groundwater basin, commonly known as the Bandung basin, is located in the center of the western part of Java, an island of Indonesia. This intra-montane basin is surrounded by up to 2,400-m-high volcanic rocks of the late Tertiary and Quaternary. The Citarum River flows in the center of the basin, and is the main river. This basin is one of most developed basins in Indonesia. It has an area of 2,300 km2 and five administrative areas—three regencies (part of Bandung, West Bandung and Sumedang) and two cities (Bandung and Cimahi). The central part of the basin, mostly comprising urban and industrial areas, is a plain measuring about 40 km east–west and about 30 km north–south. The population in Bandung basin is currently more than 7 million people. As regards the regional climate of the Bandung basin, the mean annual temperature is about 23.7 °C, the annual precipitation is 1,500–2,500 mm, and the annual potential evapotranspiration is 1,606 mm (Nurliana 2009). Meanwhile, wet season usually occurs in September–May throughout the region and dry season occurs during June–August—unpublished report (1991) by Iwaco-Waseco (see Table 1).

A geological map of the Bandung basin has been generated by Hutasoit (2009), and is presented in Fig. 1. The hydrostratigraphy of this basin is classified (older to younger order) into the following three formations: Cikapundung Formation (basement rock), Cibeureum Formation (the main aquifer), and Kosambi Formation (aquitard). A description of each formation follows:

  • The Cikapundung Formation is the oldest rock exposed in this basin and consists of conglomerates and compacted breccia, tuff, and andesite lava. The estimated age of this formation is early Pleistocene. The high degree of compaction of this formation differentiates it from the Cibeureum Formation, and forms the basement of the groundwater basin.

  • The Cibeureum Formation lies under the Kosambi Formation and is found in the periphery of the basin, as a fan-shaped distribution of volcanic deposits. It is composed of partially consolidated breccia and tuff deposits, with some lava intrusions, and is late Pleistocene–Holocene age.

  • The Kosambi Formation is the upper surface layer and is distributed in the center of this basin, as a lake deposit. Its lithology consists of unconsolidated claystone, siltstone, and sandstone of Holocene age.

Fig. 1
figure 1

Location map of the sampling points plotted on a geological map of Bandung basin (Hutasoit 2009). Three representative hydrogeological cross sections of the Bandung basin along the A–A′ line, B–B′ line and C–C′ line. The C–C′ line is shown in Figs. 4 and 5

Full size image

The most recent study of the groundwater flow system of the Bandung basin was done by Sunarwan (2014), using descriptions of cuttings, drilling well logs, resistivity logs, major ions, and stable isotopes. Hydrostratigrapically, the Bandung basin can be divided into three units—units of hydrostratigraphy 1 (UHs 1), UHs 2, and UHs 3. UHs 1 is composed of three layers: sandy tuff, clayey sand and silty clay, as a combined aquitard which contains shallow groundwater. UHs 2 (composed of two layers: sandy silt and tuffaceous sandstone) and UHs 3 (composed of one layer, tuffaceous breccia) are distinct lithologically, but it is not clear if they are hydrologically separate; therefore, UHs 2 and UHs 3 were combined as a combined UHs which contains deep groundwater as a semi-confined and confined aquifer. Priowirjanto (1985) and Arianto (2010) also divided the groundwater into two systems: a shallow and a deep groundwater flow system. The shallow groundwater is found in an aquitard with local flow. The deep groundwater is allied in a semi-confined and confined aquifer with an intermediate and regional flow.

For the purposes of this study, the authors also divided the groundwater flow system in the Bandung basin into shallow and deep groundwater flow systems. It was considered that the shallow groundwater is from a combined aquitard (UHs 1) which can be accessed via dug wells, which will be defined as zone S (see section ‘Model description’). This assumption concurs with all previous studies. It was also assumed that the deep groundwater is from a combined UHs and is obtained using drilled wells, which will be defined as zone D (see section ‘Model description’). This assumption follows the approach of Hutasoit (2009), with the deep groundwater obtained from interconnected multi-aquifers belonging to the Cibeureum Formation.

Sampling and methodology

As part of this study, 40 shallow and 65 deep groundwater samples were collected between May and July 2015. The samples were collected from springs, dug wells, production wells, and observation wells. Sampling and field measurements were carried out following standard procedures (Wood 1981). Some in situ tests such as temperature, pH, dissolved oxygen (DO), electrical conductivity (EC), and oxidation–reduction level (ORP) of the water samples were measured in the field with minimal atmospheric contact and by using a portable meter (WM-32EP, TOA DKK Inc).

All the water samples were collected after purging the wells and were filtered using 0.2-μm cellulose acetate filters (Advantec) before storing. The samples for analysis of stable isotopes (δ18O and δ2H) and major ions (Na+, K+, Ca2+, Mg2+, Cl−, \( {\mathrm{SO}}_4^{-} \), \( {\mathrm{HCO}}_3^{-},{\mathrm{NO}}_3^{-} \)) were collected in 100-ml polyethylene bottles for analyses at the chemical laboratory of the Center of Geological Resources, Geological Agency of Indonesia. Major cations were analyzed using Microwave Plasma-Atomic Emission Spectroscopy (4100 MP-AES, Agilent Technologies Inc) and the major anions using ion chromatography (DIONEX ICS 1600, Thermo Fisher Scientific Inc). The analytical error for major ions analyses was typically within 10%. Stable isotopes were measured relative to VSMOW (Vienna Standard Mean Ocean Water) by using an integrated isotope analyzer (Picarro L2120, Picarro Inc). The analytical error for the stable isotopes analyses was ±0.2% for δ18O and ±1.0% for δ2H.

Chlorofluorocarbons (CFCs) are useful tools for tracing and dating groundwater after 1945 (Busenberg and Plummer 1992; Plummer et al. 2000, 2001) because the release of CFCs into the atmosphere started after the 1940s (USGS 2010). Therefore, CFCs were employed as a marker for young groundwater; consequently, if the samples possess some dissolved CFCs, it means a fraction of the water had been recharged since 1940. The samples for CFCs analyses were collected in triplicate and stored in glass bottles sealed with metal-lined caps (Busenberg and Plummer 1992). The CFCs were analyzed using closed-system purge-and-trap gas chromatography and an electron capture detector (GC-2014, Shimadzu Inc) at the Hydrology Laboratory of Kumamoto University, Japan. The analytical uncertainty associated with the CFC analyses was less than 5%. The values were reported only if the concentrations of at least two measurements were within 10% for concentrations higher than 100 ρg/kg as the detection limit (Johnston et al. 1998).

A numerical groundwater flow model was applied to reveal the fluxes in the rejuvenation processes in the Bandung basin. A modular three-dimensional (3D) finite-difference groundwater flow model and structured rectilinear grid operating system was used—the Visual MODFLOW Classic v.2011 (Waterloo Hydrogeologic Inc). The model was constructed for a long-term groundwater flow system, from 1950 to 2015.

Results

Groundwater potential

Measurement of shallow and deep groundwater levels was done by the basin authority: the Office of Energy and Mineral Resources, West Java Province (see unpublished report, 2010, in Table 1). The results show that the shallow groundwater flows correspond with the local topographic pattern (Fig. 2a). Shallow groundwater flows from the periphery to the center of the basin, with levels strongly influenced by the season and the amount of rainfall. The shallow groundwater level showed very little drawdown or significant variations (Fig. 2b); however, the deep groundwater potential for 2010 showed some groundwater depression cones in the textile industry areas such as Cimahi (CMHI), Rancaekek (RCK), and Dayeuhkolot (DHYK; Fig. 2b). The results for the deep groundwater potentials in this study (Fig. 2b) clearly show that the groundwater depressions have become deeper. In the CMHI area, which has the biggest drawdown, groundwater levels dropped from 620 m relative to sea level (sl) in 2010 to 610 m sl in 2015. The same phenomenon also occurred in the RCK and DHYK areas, but with smaller drawdowns.

Fig. 2
figure 2

Historical change of groundwater potential map for shallow and deep groundwater in a 2010 and b 2015. There is no significant difference in shallow groundwater for the two periods; however, there are three groundwater depression areas in deep groundwater (CMHI, RCK, and DHYK area). The analyses of some representative observation wells (1, 6 and 8) are presented in Table 2 and Fig. 6, and for all observation wells presented in Fig. 8

Full size image

Major ions

Table 2 shows the results of the hydrogeochemical analyses. Figure 3a,b shows the distribution maps of water type classified on the basis of major ions using a Stiff diagram. The distribution of shallow groundwater generally shows the water was a Ca-HCO3 and Na-HCO3 type only in the center of this basin, where flows are slower or stagnant. A Ca-Cl type was only found at the southern side of the basin because of proximity to a hot spring. In contrast, the distribution map of deep groundwater shows a different dominant water type, the Na-HCO3 type, but there were some seemingly random occurrences of the Ca-HCO3 type in all the groundwater depression areas.

Table 2 Results for chemical and isotopes analyses of shallow and deep groundwater
Full size table
Fig. 3
figure 3

Distribution maps of water type for a shallow groundwater and b deep groundwater; and distribution maps of CFC-12 concentration for c shallow groundwater and d deep groundwater plotted on a land-use map (Bakorsortanal 2009)

Full size image

Evaluating the redox processes is an essential tool for understanding geochemical evolution in groundwater systems according to Chapelle et al. (1995). Usually the ORP value is used to evaluate the redox potential (Lindberg and Runnells 1984). Figure 4 shows a geological cross-section with parameters in line with the flow path. The south to north cross-section (C′–C cross-section) is the most representative cross-section towards the flow path because it crosses the DHYK area where the magnitude of groundwater drawdown is the smallest. A redox boundary was found in the deep groundwater at 14 km along this section, marked by a change in the redox level (ORP) of <0 mV, coinciding with a complete reaction of oxygen (DO) of 2 mg/l. A change in temperature was found near the northern limit of the section: increasing from 25 ± 1 °C at a fairly constant rate to 28 ± 1 °C just after the boundary. The ORP value was also used to indicate aerobic conditions. That the deep groundwater changes from aerobic to anaerobic was confirmed by the changes from unconfined to confined below the Kosambi Formation.

Fig. 4
figure 4

A geological cross section (C′–C, Fig 1) with selected parameters. The position of the redox boundary is shown by a dotted line. The geological cross section adapted from an unpublished report (2002) by the Office of Energy and Mineral Resources - West Java Province and LPPM–ITB (see Table 1) is shown at the top with observed groundwater potentials (dashed lines) for 2015

Full size image

The deep groundwater indicates more progression in geochemical evolution than the shallow groundwater. The clearest examples are the enrichment of Na+ rather of Ca2+ and the mNa/Ca ratio (Fig. 5). The increasing mNa/Cl ratio, an indicator of strong rock interaction, is also shown clearly. It can be seen on the distribution maps that the Ca-HCO3 type was dominant at the periphery of this basin, and changed to the Na-HCO3 type after reaching the center of the basin. This change could be explained by the cation exchange reaction (Ca2+ ⇔ Na+; Appelo and Postma 2005). The cation exchange reaction in deep groundwater also occurred in the groundwater flow system of the Osaka basin in Thailand (Yamanaka et al. 2011) and the Jakarta basin in Indonesia (Kagabu et al. 2011).

Fig. 5
figure 5

CFC-12 concentration along the cross sections (Fig. 1) showing the three depression areas (Fig. 2b). All geological cross sections adopted from the unpublished report (2002) by the Office of Energy and Mineral Resources - West Java Province and LPPM–ITB (see Table 1) with observed groundwater potentials (dashed lines) for 2015

Full size image

Stable isotopes

Stable isotope ratios of oxygen (δ18O) and hydrogen (δ2H) are ideal tracers for identifying the source of groundwater. The results presented in Table 2 show that the groundwater samples were almost all within the same range, but the deep groundwater had a slightly more homogeneous range (−7.4 to −3.8% δ18O and −54 to −34% δ2H) than the shallow groundwater (−7.4 to −3.8% δ18O and −54 to −34% δ2H). These almost identical ranges of values indicate that the shallow and deep groundwater were recharged at approximately the same elevation. These results are in agreement with analyses of the groundwater source determined by the temperature-depth profile that showed the recharge area to be at the periphery of the basin (Delinom and Suridarma 2010).

CFC-12 concentration

CFC-11, CFC-12, and CFC-113 concentrations were measured for all the samples, but only the CFC-12 concentration was used as a marker for young groundwater in this study because of its stability in the subsurface environment (Plummer et al. 1998). As presented in Table 2, the CFC-12 concentration in the shallow groundwater exhibited a wide range (12.2–540.8 pptv). Moreover, some of the shallow groundwater exceeded the maximum atmospheric concentration (>546 pptv; USGS 2010). For the samples with CFC-12 concentrations greater than 546 pptv, it was not possible to evaluate the exact residence time of the groundwater because the samples were contaminated by local CFC sources. This phenomenon has also been reported in some large cities such as Las Vegas and New York in the USA, and Beijing in China (Carlson et al. 2011).

Figure 3c,d shows the distribution of CFC-12 concentration (pptv) on the land use map. The shallow groundwater with a high CFC-12 concentration indicates a modern (young) groundwater age and that it was recharged after the 1940s. This result is consistent with Tritium (3H) analyses that indicated an apparent age of 12.35–58.38 years (3.13–0.23 TU; Sunarwan 2014). In contrast, the deep groundwater had a low CFC-12 concentration, and most of the deep wells had no CFC-12 or were below the detection limit. This result agrees with 14C analyses that indicated that the apparent age was 1,291–35,000 years—Wahyudin Matahelumual 2008; Satrio et al. 2012; and an unpublished report (1990) by Geyh (see Table 1). However, some of the deep groundwater samples from the groundwater depression areas had high CFC-12 concentrations, indicating that modern (young) groundwater might be recharged to the deeper aquifer.

Discussion

Groundwater flow system of the Bandung basin

This report discusses the groundwater flow system for both the shallow and deep groundwater in the Bandung basin based on the results of all the hydrogeochemical data. According to the results of the stable isotope analyses, all the groundwater is recharged at the periphery of the basin. After recharge, the groundwater flows are divided into two major systems, i.e. shallow and deep groundwater flow systems. This subdivision is consistent with the stratification of the local geology. Comparing the distribution map of water type with the distribution map of CFC-12 concentrations for shallow groundwater, shows that the Ca-HCO3 type with high CFC-12 concentrations occur at the periphery of the basin, while the Na-HCO3 type with low CFC-12 concentrations occurs in the center of the basin (Fig. 3). In addition, for the deep groundwater, the Na-HCO3 type water had low or no CFC-12 concentrations, which indicates a relatively long residence time. Since there were Ca-HCO3 types that appear to be randomly distributed in the groundwater depression areas, it is assumed that the subsurface environment has been disturbed. This assumption will be addressed in the following discussion.

Rejuvenation processes

Some high CFC-12 concentrations were found in some of the deep groundwater samples that were collected in all the groundwater depression areas—for example, in the CHMI area approximately 50% of deep wells (10, 2, 3, 5, 6c, 4, 6b, 7c, 7, and 7b) had CFC-12 concentrations of more than 100 pptv. The samples from all the other depression areas (RCK and DHYK) that also possessed CFC-12 concentrations are presented in Fig. 4. This reveals that shallow, younger groundwater clearly contributes to the rejuvenation of deep groundwater in the Bandung basin. Furthermore, for each groundwater depression area, the magnitude of groundwater drawdown has an impact on groundwater mixing that will be discussed further in a separate study.

The rejuvenation ratio (R) was calculated as an indicator of rejuvenation based on 14C activity (in pMC; percent of ‘modern carbon’) in deep groundwater. At three observation wells, representative of different depression areas (1, 6, and 8; as presented in Fig. 2b) samples had been taken in 2008 and in 2012, as listed in Table 3. The measured pMC activity exhibited an increasing tendency implying that the apparent groundwater age decreased in 2012. This phenomenon is the so-called ‘rejuvenation’ of groundwater age (Stewart et al. 2004; Kagabu et al. 2013). To discuss rejuvenation based on the observed 14C activity, R is determined as an indicator of rejuvenation (Eq. 1).

$$ R=\frac{R_{\mathrm{m}}-{R}_{\mathrm{o}}}{R_{\mathrm{o}}}\times 100 $$
(1)
Table 3 Data of 14C activity in groundwater samples
Full size table

In Eq. (1), R (dimensionless) is a ratio between the difference of 14C activity (pMC) in two different years to concentration at the initial year, and R o and R m (pMC), respectively denote the 14C activity in 2008 (initial year) and 2012 (measured year)

Figure 6 shows that R increases with CFC-12 concentration. The strong relationship is an indication that greater shallow groundwater intrusion provides higher R values—for example, the CMHI area with the largest groundwater drawdown had the highest average CFC-12 concentration (190 pptv) and the highest R (50%). This relationship will be further investigated by the following numerical modeling.

Fig. 6
figure 6

Relationship between R and the average CFC-12 concentration. The CFC-12 concentrations are shown for the three observation areas, and R values are from three observation wells which are representative of the depression areas. The R value is a rejuvenation ratio determined using 14C activity between 2008 and 2012, as listed in Table 3. The circle size shows the magnitude of ‘vertical downward flux’ from modeling (Fig. 10)

Full size image

Groundwater flow modeling

Model description

A 3D model of the groundwater flow system in the Bandung basin was first accomplished by Hutasoit (2009). The model domain covers an area of approximately 4,225 km2. The model is divided into a grid of equally spaced 0.5-km2 cells, with 130 grid lines for both rows and columns, and has 16,900 cells (Fig. 7a). The vertical extent of the Bandung basin is represented by two groundwater flow systems that denoted into six model layers (Fig. 7b) and the thickness of each layer represents the hydrogeological layer thickness. Layers L1, L2 and L3 are set as the shallow groundwater system from a combined aquitard (defined as zone S). Layer L4, L5 and L6 are set as the deep groundwater system from a combined aquifer (defined as zone D). The descriptions of zone S are: layer L1 is made up of volcanic deposits (sandy tuff) on the basin’s slopes, and is part of the upper Cibeureum Formation; layer L2 is a clayey sand layer of the upper Kosambi Formation and spreads only in the center of this basin; and layer L3 is a silty clay layer, and is part of the lower Kosambi Formation. Meanwhile, for zone D: layer L4 is a sandy silt layer in the upper layers of the Cibeureum Formation, and spreads only in the central bottom of the basin; layer L5 is a composed of tuffaceous sandstone of the Cibeureum Formation, and forms the middle aquifer (or UHs 2); layer L6 is composed of tuffaceous breccia of the Cibeureum Formation, and forms the lowest aquifer (or UHs 3). At the bottom of these layers is basement rock of the Cikapundung Formation, which is older and harder, and is set as a no-flow boundary.

Fig. 7
figure 7

a Model grid of the numerical model. The grid comprises 0.5 × 0.5-km2 cells in surficial distribution; b The vertical distribution represents the subsurface condition along the south to north line (a)

Full size image

The Citarum River was adopted as a constant-head boundary because there were very little data about the interaction between river water and groundwater. No-flow boundary conditions were assumed for the bottom and the outer sides of the model. The various recharge rates were assigned to the model grid at the periphery of this basin as recharge area. For the entire model domain, recharge was set at 20–100 mm/year, with an absorption coefficient that changed according to land-use change (Nurliana 2009).

Hydraulic parameters and their calibration

For this study, the groundwater flow model was set up using parameters as presented in Table 4. Hydraulic conductivity (K) and transmissivity (T) were determined initially by Hutasoit (2009) from pumping test results and had been calibrated with the groundwater potential distribution measured in 1950 representing natural groundwater conditions. Furthermore, in this study, those parameters were calibrated by comparing the computed and observed groundwater potential, as described later. The aquifer was assumed to be an isotropic with fixed vertical and horizontal hydraulic conductivity for each layer, and heterogeneity was represented by an equivalent homogeneous aquifer with anisotropic hydraulic conductivity (i.e. K x = K y > K z).

Table 4 Properties of the groundwater flow model
Full size table

In the calibration process, the parameters were adjusted to best fit the long-term fluctuations of groundwater potential by comparing the computed and observed groundwater potential for the long-term period 1994–2015 (Fig. 8a), and simultaneously the calculated pumping volumes were obtained. The initial parameters were from a previous study (Hutasoit 2009) that had been calibrated up to 2016, and the final adjusted parameters were almost the same as the initial ones. The observed groundwater potential values are from ten observation wells with records of groundwater potentials in the deep aquifer (their locations are presented in Fig. 2b). The relationship between the observed groundwater potentials and the estimated values at the end of each year period is in good agreement, thereby validating the model (Fig. 8b). The calculated groundwater pumping volumes were very different to the official records of volumes—unpublished report (2004) from the Office of Energy and Mineral Resources - West Java Province, see Table 1. The calculated volumes were up to 14 times higher than the official volumes in recent years, as presented in Fig. 9. Similar anomalies were found in the Jakarta area, with calculated volumes reaching 12 times the official volumes (Kagabu et al. 2013). These anomalies might be caused by unregistered wells (Braadbaart and Braadbaart 1997) and the fact that many of the pumping volumes have not been reported in official documents. This is one of the biggest problems in groundwater management in Indonesia. This might also have occurred in other Asian cities where there are subsurface problems; therefore, this finding regarding under-reported pumping volumes could provide valuable information for improving monitoring and modeling.

Fig. 8
figure 8

a The long-term fluctuation of groundwater potentials (GWL = groundwater level) between the observed (solid line) and estimated (dashed line) values; b The relationship between the observed groundwater potentials with the estimated values is in good agreement. The observation well locations are presented in Fig. 2b

Full size image
Fig. 9
figure 9

Comparison of groundwater pumping volume between the volume from the official report (block bar) and the calculated volume from the modeling (blank bar)

Full size image

Estimated groundwater flux from the simulation model

Water budget analyses were conducted to calculate the vertical downward flux from zones S to D. The change in this flux was estimated for each year during 1950–2015. The change in vertical flux is affected by the change in groundwater pumping, as shown in Fig. 10. Since the 1970s, the vertical flux has increased significantly, suggesting that the shallow groundwater recharged the deeper aquifer. The flux reached approximately 15% of the total pumping amount during the 2000s, compensating the groundwater pumping in the Bandung basin. In addition, according to Arianto (2010), this vertical flux phenomenon was also indicated by pumping tests at some of the deep drilled wells in the Bandung basin with leakage factor ranging between 0.167 and 1.25 m.

Fig. 10
figure 10

Calculated groundwater pumping volume (solid line), and the vertical downward flux (dotted line) from shallow groundwater (zone S) to deep groundwater (zone D) below the Bandung basin, from a numerical model for period 1950–2015

Full size image

Comparison between observed hydrogeochemical tracer and estimated vertical flux

In the previous section, the shallow groundwater intrusion caused by groundwater pumping was estimated using the simulation model. Figure 6 shows the relationship between R and CFC-12 concentration, and also the estimated ‘vertical downward flux’ at each depression area. The magnitude of this flux is expressed as the diameter of a circle. The biggest flux was at the CMHI area where the biggest drawdown occurred, and the flux is also clearly shown in the other areas. Figure 6 confirms that the vertical downward flux in each area affects the magnitude of the shallow groundwater intrusion, which is also confirmed by CFC-12 concentrations and R values. Thus, the modeling confirmed the results obtained from the hydrogeochemical techniques, i.e. by observed high concentrations and by estimating the R of 14C activity.

A similar calculation was applied to the Jakarta area, and it was found that the vertical flux reached approximately 50% (Kagabu et al. 2013). The differences in flux might be caused by many factors such as the geologic conditions, the magnitude of groundwater drawdown, the amount of groundwater pumping, and the number and distribution of deep wells.

Conclusions

The differences between the shallow and deep groundwater flow systems in Bandung basin were hydrogeochemically characterized. The shallow groundwater is generally a Ca-HCO3 type, with high CFC-12 concentrations; however, a Na-HCO3 type was found in the center of the basin where flow rates are less. Some of the shallow groundwater had CFC-12 concentrations that exceeded the maximum atmospheric concentration, thus indicating that they were contaminated by local CFC sources. In comparison, the deep groundwater was a Na-HCO3 type with low CFC-12 concentrations; in fact, most of the deep groundwater wells had no CFC-12 or were below the detection limits. Stable isotopes analyses revealed that both groundwater systems were recharged from the periphery of the basin; however, the deep groundwater showed more progression in terms of geochemical evolution than the shallow groundwater.

Significant groundwater drawdown has caused age rejuvenation processes in all the groundwater depression areas. These processes were identified by evidence of the observed CFC-12 concentration in the deep groundwater, which can be an indicator of young age. The rejuvenation ratio R was determined using the observed 14C activity at representative observation wells between 2008 and 2012 and showed good correlation with the CFC-12 concentrations, increasing as CFC-12 concentrations rose. This correlation was confirmed by the vertical downward flux obtained from modeling. Furthermore, this ‘vertical downward flux’ was greater in the groundwater depression area and in the shallower part of the deep aquifer, affecting the magnitude of the shallow groundwater intrusion. This increase of flux is caused by excessive groundwater pumping corresponding to industrialization occurring in the Bandung basin. Since the 1970s, this vertical flux has increased significantly and reached approximately 15% of the total pumping amount during the 2000s. The vertical flux model result confirmed the results of other approaches using hydrogeochemical techniques, i.e. by observed high CFC-12 concentrations and by the estimated R from 14C activity.

This study clearly reveals the rejuvenation processes in the Bandung basin by combining multiple hydrogeochemical techniques and numerical modeling. The magnitude of shallow (young) groundwater intrusion into the deeper aquifer is traced systematically by hydrogeochemical markers, the estimation of R, and groundwater flux modeling. The groundwater potential is expected to decline further, and the deeper groundwater is expected to be even more affected by shallow groundwater that is highly polluted by urban and industrial contaminants. Thus there is an urgent need for better monitoring and a reduction of the excessive pumping by either securing alternative water resources or introducing pumping regulations.