Introduction

In the city of Arusha, situated in northern Tanzania, groundwater abstracted through drilled wells and springs is the main source of drinking water supply (GITEC 2011). Most deep wells with large production volume are located in the central part of the study area near the foot of Mt. Meru at an elevation from 1,400 to 1,500 m above sea level (asl) (Ong’or and Long-Cang 2007). According to the Arusha Urban Water Supply and Sanitation Authority (AUWSA) medium-term strategic plan (2015–2020) report, springs contribute 45% of the daily water production, whereas drilled wells and rivers contribute 37 and 18% respectively (AUWSA 2014); however, the proportional contribution from each source varies depending on season of the year. The seasonal variations mainly affect springs and rivers by significant reduction in flows during dry periods but production from wells remains constant (AUWSA 2014). Generally, water production fluctuates seasonally from an average of 35,000 m3/d (dry season) to 60,000 m3/d (rainy season), which is significantly low with respect to the current water demand (93,270 m3/d) in the city (AUWSA 2014).

Despite the lack of reliable information on the extent of groundwater abstraction, there is evidence of groundwater over-pumping in the study area. The decline of water levels (Table 1) and respective yield reduction (Table 2) in wells that have been operational for more than 20 years has been reported (GITEC 2011; Ong’or and Long-Cang 2007). Additionally, an inventory conducted by Pangani Basin Water Office (PBWO) in 2013 revealed more than 400 drilled wells in the study area, most of them unregistered by the responsible water resources management authority. This suggests that groundwater abstraction in the area is not adequately controlled to meet the needs of present and future use (Kashaigili 2010; Van Camp et al. 2014). However, aquifer storage may be affected by a number of factors other than groundwater over-pumping (Custodio 2002). Natural phenomena such as delayed and transient effects of the aquifer system, earth quakes, tectonic movement and climate change, have been reported in many parts of the world with significant effect in aquifer productivity (Custodio 2002; Gorokhovich 2005; Kitagawa et al. 2006; Kløve et al. 2014; Nigate et al. 2017). Due to the complexity and dynamics of hydrogeological processes, knowledge of a particular aquifer system including recharge mechanisms and age of abstracted groundwater is required to inform the cause and extent of the existing problems. Such a situation leaves a number of questions with respect to sustainability of groundwater utilization in Arusha City for present and future water resources development and for avoidance of likely human impacts (drinking water supply) and ecosystem impacts (such as reduced stream flows and springs drying up and subsequent impacts on aquatic life). The current research was conducted to inform some of the existing problems or fill knowledge gaps in the study area. Among the issues addressed include whether groundwater storage depletion is caused by over-pumping or whether the aquifer system in the study area is not actively recharged.

Table 1 Water level trends in the study area (GITEC 2011)
Full size table
Table 2 Well discharge trends over time in the study area (GITEC 2011)
Full size table

In order to address some of the experienced groundwater challenges, the study employed the use of isotopic techniques to investigate the groundwater age and recharge mechanism in the study area. Currently, there are a number of tracers used for assessment of groundwater recharge and mean residence time (Chen et al. 2011). These include chlorofluorocarbons (CFCs), tritium/helium ratio (3H/3He), krypton-85 (85Kr) and carbon-14 (14C) (Douglas et al. 2007; Hoque and Burgess 2012; Sigstedt et al. 2016). CFCs, 3H/3He and 85Kr are mostly used for tracing young groundwater (< 100 years; Mathouchanh and Aeschbach-Hertig 2015; Szabo et al. 1996); however, the current study utilized stable isotopes (hydrogen and oxygen), tritium and 14C to meet the objectives of the current study. The 14C technique uses the decay principle of the activity measured in dissolved inorganic carbon (DIC) to estimate the mean groundwater age of water travelling from the recharge zone to a discharge point along the flow path (Douglas et al. 2007; Hagedorn 2015). These techniques have been undergoing several improvements and are widely used in hydrogeological studies, particularly groundwater recharge and mean age estimation ranging from young (100–1,000 years BP) to old groundwater (1,000 to millions of years (Bakari et al. 2012a, b; Gleeson et al. 2016; Stewart 2012). Despite wide acceptance and global application of isotope techniques, in Tanzania, they have commonly been applied in geology and marine based sediments studies (Muzuka et al. 2010; Muzuka et al. 2004; Öberg et al. 2013). Conventionally, many groundwater assessment studies have been focusing on quality issues and leaving aside aquifer sustainability in terms of potential recharge and groundwater mean residence time (Ghiglieri et al. 2010, 2012; Malago et al. 2017; Van Camp et al. 2014). Lack of such information is likely to threaten sustainability of aquifers under intensive development (Zongyu et al. 2005); thus, this study aimed at establishing groundwater age and recharge mechanism for sustainable groundwater utilization in Arusha City, using 14C, tritium and stable isotope (2H and 18O) techniques. Isotope composition data have been used for interpretation in conjunction with groundwater physical parameters and some hydrogeological information which were established during the borehole drilling and construction stage.

The study area

Location

The study was conducted in Arusha City and Arusha District, which are located on the southern slopes of Mount Meru in northern Tanzania (Fig. 1). The study area is bordered by the three administrative districts of Monduli, Longido, and Meru. The area covers approximately 282 km2 and lies between latitudes 3°15′ and 3°30′ south and longitudes 36°34′ and 36°46′ east (Fig. 1). According to the population and housing census, Arusha City and Arusha District in 2012 had a population of about 416,442 and 323,198 inhabitants respectively (NBS 2013).

Fig. 1
figure 1

Location of the study area

Full size image

Climatic characteristics

The area is characterized by tropical climate with two distinct seasons, dry and wet. The rainfall pattern in Arusha, as part of northern Tanzania, is bimodal with short rains from October to December and long rains from March to May (Kijazi and Reason 2009; Zorita and Tilya 2002). The long rains and short rains vary from approximately 110 to 200 mm and 60 to 100 mm respectively (Kabanda and Jury 1999). The total annual rainfall ranges between 500 and 1,200 mm with mean value of about 842 mm (Kaihura et al. 2001). The temperature typically ranges between 13 and 30 °C with an average annual temperature of about 25 °C (Nonga et al. 2009). The coolest month is July, whereas the warmest is February and the relative humidity varies from 55 to 75% (Anderson et al. 2012).

Geological and hydrogeological settings

The geology of the study area is dominated by volcanic materials of varying ages and recently deposited alluvial sediments (Ghiglieri et al. 2008, 2010; Ong’or and Long-Cang 2007; Wilkinson et al. 1986). Mt. Meru is the main center of volcanic activity in the region. The main features of the volcanic eruption in the area include main cone deposits, mantling ash, lahars, lava flows, pyroclastic materials, tuffs, pumice, agglomerates and volcanic rocks such basalts (Nanyaro et al. 1984; Ong’or and Long-Cang 2007). Some of these volcanic features have been depicted in a hydrogeologic map (Fig. 2). Volcanic rocks are mainly lava flows (basaltic to phonolitic and nephelinitic tuff). These materials, if not fractured or weathered, act as aquitards which favor groundwater movement down the slope (Flint et al. 2001; Wilson and Guan 2004). Their properties normally change with time due to physical and chemical reactions such as weathering and subsequent erosion and re-deposition of materials (Arıkan et al. 2007; Nesbitt and Young 1984).

Fig. 2
figure 2

Hydrogeological map showing location of sampling sites (AA.VV. 1983)

Full size image

The area is also affected by tectonism leading to the development of fractures and faults which act as conduits to groundwater flows in some areas (Ghiglieri et al. 2010). Figure 2 shows the fault system within the main cone deposits of pyroclastic materials with subordinate nephelinitic and phonolitic lavas. The fault lines are assumed to be avenues of huge groundwater flows that manifest through numerous springs that discharge into Themi River (Fig. 2).

Groundwater recharge mainly takes place at high elevation on the slopes of Mt. Meru along fractured formations as well as through infiltration in valleys or depression zones with medium-to-coarse grain sizes (Ghiglieri et al. 2010). Groundwater potentiality in the fractured formation is also supported by a number of springs around the fault zone (Fig. 2), in the northeastern side of the study area. Spring flows from this zone are very high, particularly after or at the end of long rains for example, according to AUWSA daily discharge measurements, 25,698 m3/d was abstracted from the springs in May 2015 for public water supply. This amount is only the portion of groundwater discharged from springs along this fault; the remaining water flows into Themi River which is one of the perennial and reliable water sources in the study area. Overall, most rivers and streams originate from springs located on the slopes of Mt. Meru.

Materials and methods

Field work and groundwater sampling

Two categories of water samples were collected from the study area for different analyses. The first category was sampled from wells with depth ranging from 22 to 200 m below the ground surface. All wells considered in this study were tapping water at different depths depending on screen position and pump location (Figs. S1 and S2 of the electronic supplementary material (ESM)). The second category included samples from springs. Spring water samples were collected close to the point of discharge in order to minimize the effect of atmospheric contamination. Figure 2 shows the locations of sampling sites in the study area. All samples were collected in May 2016, which marked the end of the rainy season in the study area. The rainy season started early April 2016.

Physical parameters such as pH, temperature and electrical conductivity (EC) were measured in situ using a multi-parameter meter (model HI 9828, Hanna Instruments) that was calibrated before use. The readings were taken after the instrument has stabilised. Alkalinity as bicarbonate (HCO3−) was determined in the laboratory immediately after sampling by an acid-titrimetric method using standard sulfuric acid and bromocresol green indicator for end-point detection (Rice et al. 2012)

A total of 25 samples were collected for both stable isotopes (18O and 2H) and radioisotopes analyses (14C and 3H), while 6 samples were from springs and the rest from boreholes. During field work, standard procedures and techniques for groundwater sampling and preservation detailed by Clark and Fritz (1997) were systematically applied prior to laboratory analyses.

High-density linear polyethylene (HDPE) sampling bottles were used to collect and store samples from taps located near the well heads. All boreholes used in this study were installed with electric submersible pumps which provide positive pressure systems with no atmospheric contact. Almost all samples from boreholes were collected after the pump had ran for at least 6 h and in some cases it was found that pumps had been operating for more than a day nonstop for public water supply. This ensured that the water samples collected represent the aquifer formation being sampled and not stagnant water in the wells. All samples were kept in cool box and subsequently stored in a refrigerator (~4 °C) after field work. Samples for stable isotopes analyses were sent to Stable Isotope Facility at University of California Davis, USA, whereas analyses of radioisotopes were carried out at Environmental Isotope Laboratory, University of Waterloo, Canada.

Laboratory analyses

Stable isotopes (δ18O and δ2H) measurement

Simultaneous analysis of δ18O and δ2H isotopes in groundwater samples were carried out at Los Gatos Research, Inc., Mountain View, California, USA, using a Laser Water Isotope Analyzer V2. Sample isotope ratios were standardized using a range of working standards calibrated against Vienna Standard Mean Ocean Water (VSMOW). The isotope values of hydrogen and oxygen are reported in δ-notation relative to VSMOW (Eq. 1). The δ-values are expressed as parts per thousand or permil (‰). The precision for analyzed groundwater samples was ≤0.3‰ for δ 18O and ≤ 0.8‰ for δ2H. The average and standard deviation for an internal check dispersed throughout the run with known isotope ratio values for the calibrated water is given in Table 3.

$$ \delta =\left(\frac{R_{\mathrm{sample}}}{R_{\mathrm{std}}}-1\right)\times \mathrm{1,000} $$
(1)

where Rsample is 18O/16O or 2H/1H ratio of the water sample and Rstd is for the standard in VSMOW.

Table 3 Mean and standard deviation (SD) of water used for calibration
Full size table

Radioisotopes measurement

Tritium (3H)

Tritium was determined by the direct tritium method with a precision of ±8 TU using a PerkinElmer LKB-WALLAC Quantulus 1220–002, USA. The results are expressed as tritium units (TU), whereby 1 TU = 3.221 picocuries(pCi)/L or 1 TU = 0.11919 Bq/L.

Carbon isotopes

The determination of δ13C of dissolved inorganic carbon (DIC) was by mass spectrometry method using a Micromass PRISM-II, UK, 1990. The results are reported in δ values relative to Pee Dee Belemnite (PDB) standard with precision of ±0.2‰. Radiocarbon (14C) activity was determined using accelerator mass spectrometry (National Electrostatics Corporation 1.5SDH-1 Pelletron Accelerator). The results are reported in percent modern carbon (pMC) with precision of ±0.3. According to (Clark and Fritz 1997), the activity of modern carbon represents 95% of the 14C activity in 1950 of the NBS oxalic acid standard equivalent to the activity of the wood grown in 1890 in a fossil CO2-free environment. For the purpose of age dating, the measured 14C activity values were normalized to values corresponding to δ13C –25‰ to account for sample background contamination during the process of graphitization.

Groundwater age dating and 14C correction methods

The groundwater age dating by 14C is governed by the principle of radioactive decay given by Eq. (2). The normalized 14C activity was applied together with initial 14C activity of DIC in the recharge area for each sample to estimate the respective groundwater ages. Groundwater radiocarbon age was expressed as years before present (BP).

$$ A={A}_{\mathrm{o}}{e}^{-\lambda t} $$
(2)

where t is the mean residence time from recharge to discharge point in years, A is the measured 14C activity of DIC in the sample expressed in pMC, while Ao is the initial 14C activity in the recharge, λ is the decay constant which equals ln2/T1/2, and T1/2 is the 14C half life (5730 years).

Radiocarbon ages were determined from the measured 14C activities. The first approach (Table 5) assumed that no dilution occurs other than natural decay of 14C. This assumption is commonly applied in groundwater age interpretation particularly in basaltic formations characterized by lack of organic carbon which may contain minor amounts of calcite in vesicles. This was reported by Bosworth (1989) as cited (Raiber et al. 2015). The initial 14C activity is assumed to be 100 pMC which gives uncorrected ages; however, this assumption ignores any subsequent geochemical reactions from the point of groundwater recharge and along the flow path. Due to the complexity of the geochemical reactions and unidentified sources of carbon in the aquifer system, it is always necessary to correct radiocarbon ages for addressing any possible dilution effects from both known and uncertain sources (Clark and Fritz 1997). There are several existing age correction techniques applied in different environmental conditions depending on available data. This work applied the statistical correction (Vogel 1970) and δ13C mixing method (Pearson and Hanshaw 1970) for 14C age correction. The δ13C mixing method indirectly accounts for carbon chemistry in terms of mixing of different components. Apart from mixing, the process includes an isotope exchange process which is considered as a simple addition of one of the two components, CO2 or solid carbonate, into the mixing (Fontes and Garnier 1979). Fontes and Garnier (1979) concluded that the Pearson method provides good approximation to the initial 14C activity of the total dissolved carbon with respect to their new approach. The initial activity of the total dissolved carbon is estimated based on the 13C content of each species (Clark and Fritz 1997). The correction factor (q) is given by:

$$ q=\frac{\updelta^{13}{\mathrm{C}}_{\mathrm{DIC}}-{\updelta}^{13}{\mathrm{C}}_{\mathrm{carb}}}{\updelta^{13}{\mathrm{C}}_{\mathrm{rech}}-{\updelta}^{13}{\mathrm{C}}_{\mathrm{carb}}} $$
(3)

where the following terms are as defined:

δ13CDIC:

Measured 13C in groundwater

δ13Crech:

δ13C value for DIC in the infiltrating groundwater

δ13Ccarb:

δ13C of the calcite being dissolved

In this work, the values of δ13Crech and δ13Ccarb were opted to be −23 and 0‰ respectively (Clark and Fritz 1997). The statistical correction method as described by Vogel uses initial 14C activity of 85% of the modern carbon (Vogel 1970). However, this approach gives ages which are probably too old and does not account for the occurrence of recent waters in the recharge area (Fontes and Garnier 1979).

Results

Stable isotopes and tritium

Results of stable isotopes of hydrogen and oxygen (Table 4) already plotted and parallel to both East African local meteoric water lines (LMWLs) and the global meteoric water line (GMWL; Fig. 3). The GMWL (δ2H = 8.13δ18O + 10.8) is based on the refined Craig’s line (Clark and Fritz 1997), whereas the LMWL for Dar es Salaam, Tanzania (δ2H = 7.01δ18O + 6.83), and for Entebe, Uganda (δ2H = 7.38δ18O + 10.78),are reported (Rozanski et al. 1996). The best fit line for the analyzed stable isotopes is δ2H = 6.7δ18O + 9.8. Water samples from springs had δ2H values that ranged from −24.8 to −18.5‰ VSMOW, while samples of δ18O ranged from −5.1 to −4.1‰ VSMOW. In samples from wells, δ2H varied from −26.1 to −17.0‰, while δ18O ranged from −5.34 to −4.32‰. The results indicate that spring water is more enriched in 18O and 2H compared to deep groundwater samples. The deuterium excess, d = δ2H - 8δ18O is considered as a measure of the relative proportions of the sample and an index of deviation from the GMWL (Dansgaard 1964). The deuterium excess varied from 13 to 18‰ for the analyzed groundwater samples. All samples were also analyzed for tritium content; however, the results were less than analytical detection limit (6.0 TU) for all samples.

Table 4 Physical-chemical and isotopic characteristics of groundwater in the study area
Full size table
Fig. 3
figure 3

Stable isotope signatures of groundwater in the study area

Full size image

14C and 13C of dissolved inorganic carbon (DIC)

The values of the stable carbon isotope, δ13C, varied from −14.9 to −11.9‰ PDB in spring water, while in well water the values ranged from −15.6 to −8.3‰ PDB. Depending on the prevailing pH condition, δ13C values are higher as water infiltrates through the soil matrix, signifying enrichment (Clark and Fritz 1997). The values of δ13C correlated positively with bicarbonate HCO3− [r = 0.597 (n = 13), p < 0.01] and EC [r = 0.926 (n = 13), p < 0.01] for well samples; however, groundwater sampled from springs showed weak correlation with respect to HCO3− (Fig. 4). The bicarbonate is mainly formed as water dissolves soil carbon dioxide during the infiltration process. Enrichment with δ13C has been observed more in well water samples than springs (Table 4). Low values of EC (347 ± 179 μS/cm) were observed in springs compared to well waters (540 ± 232 μS/cm) with the exception of spring S04 (671 μS/cm) which is located in the city center.

Fig. 4
figure 4

Scatter plot for δ13C with a bicarbonate and b electrical conductivity in water samples

Full size image

Measured 14C activities of water samples varied from modern in spring water (87.00–106.20 pMC) to low values (49.49–92.56 pMC) corresponding to great age in well water samples (Table 4). The high values of 14C activity observed in spring water imply modern groundwater. Samples from wells had relatively low values of 14C activity which indicates longer mean residence times. 14C activities correlated positively (r = 0.49 (n = 12), p < 0.01) with altitude in samples collected from wells (Fig. 5). In addition, samples with higher 14C activities were relatively enriched in both 2H and 18O (Fig. 6).

Fig. 5
figure 5

Scatter plot for 14C activity associated with a altitude and b well depth

Full size image
Fig. 6
figure 6

Relationship between stable isotopes and 14C activity in water samples: a hydrogen, and b oxygen isotopes. The wells inside the ellipse shape are located in alluvial deposits

Full size image

Groundwater age

Radiocarbon dating revealed that groundwater in the study area was recharged in the Quaternary period during late Holocene. The groundwater ages determined based on corrected 14C activity ranged from 1,400 years BP to modern age. All spring water samples fell under modern groundwater; however, samples S10 and S12 were observed to have relatively lower 14C activities compared to others (Table 5), whereas samples from wells of different depths had groundwater ages ranging from 1,400 to 100 years BP. The 14C activities (directly related to estimated groundwater ages) were positively correlated (r = 0.49) with altitude but no clear trends were observed with well depth (Fig. 5). Young groundwater was observed in most wells located at high altitude and vice versa.

Table 5 Groundwater 14C ages rounded to the nearest 102 years
Full size table

Discussion

Moisture source

The stable isotope results gave a regression line with a slope of 6.7, which is close to but less than the one established by Rozanski et al. (1993) for the GMWL, i.e., 8.13 (Fig. 3; Clark and Fritz 1997) and the East African LMWL for Tanzania and Uganda with slope values of 7.01 and 7.38 respectively. Such a value indicates minimum or limited evaporation of infiltrated rain water during the recharge period in the study area. These findings are also supported by high deuterium excess values recorded by most groundwater samples ranging from 13 to 18‰ VSMOW (Table 4). Similar results were reported in a study conducted by Ghiglieri et al. (2012) at the North Eastern part of Mt. Meru where the values of deuterium excess varied from 10 to 17‰ VSMOW. Generally, at global scale the deuterium excess values range from −2 to about 10–15‰ VSMOW (Froehlich et al. 2001) depending on weather conditions. Moreover, other work reported in the literature shows that warm and dry conditions enhance evaporation, which leads to low or even negative deuterium excess values (Katsuyama et al. 2011, 2015; ; Steen-Larsen et al. 2014). High deuterium values observed in the current study indicate dominance at low temperature and low humidity during groundwater recharge (Plummer et al. 2012).

Groundwater recharge

High 14C activities observed in spring water (98.1 ± 7.9 pMC) correspond with recently recharged groundwater in the study area (Fig. 6). The results indicate that the shallow aquifer is actively recharged and responds to seasonal variations. This is supported by spring flows (Fig. 7) whereby high flows are experienced during or immediate after the rainy season and vice versa during dry conditions. The spatial distribution of 14C activities shows weak correlation (r = 0.49) with altitude (Fig. 5). The weak correlation may be due to groundwater mixing from different layers probably recharged at different times. High 14C activity values were observed from wells located at high altitudes, the area which receives more rainfall compared to low altitudes in the southern slopes of Mt. Meru. This suggests that groundwater recharge is actively taking place in areas of high altitude on the slopes of the Mountain; however, no clear trends were established between 14C activity and well depth (Fig. 5) probably indicating that most deep wells are tapping water from the same aquifer but at different depth. Additionally, spring waters are relatively more enriched with respect to hydrogen and oxygen isotopes (Table 4). This implies a higher fractionation effect due to evaporation in shallow groundwater than in deep wells; however, the difference in stable isotopic signatures is somehow narrow which implies that the recharge of both shallow and deep groundwater took place under similar climatic conditions (Hoque and Burgess 2012).

Fig. 7
figure 7

Seasonal variations of spring flows (S04 and S07) characterized by modern groundwater. Source: The data were compiled from daily flow measurements conducted by Arusha Urban Water Supply and Sanitation Authority. The flow is measured by means of a broad-crested weir with a rectangular control station

Full size image

Apart from 14C dating, recent groundwater in springs was also evidenced by seasonal variations of spring-flow data established for two springs (S04 and S07) from 2014 to 2016 (Fig. 7). The flows respond positively during the rainy season by increasing significantly and vice versa during dry months. This indicates that protection of these springs is of utmost importance otherwise the recharge cycle will be interfered by rapid urbanization and subsequently drying up or decrease of discharges; nevertheless, the springs S10 and S12 recorded relatively lower values of 14C activities, i.e., 87.4 and 89 pMC respectively. These activities probably indicate that the two springs originate from stable aquifers which do not respond easily to recharge from local precipitation.

Tritium results indicated low concentrations in all samples (<6.0 TU); however, the detection limit (6 TU) of the method used was not able to delineate exactly the amount in each sample which is expected to be low. In a similar study conducted on the northeastern Mt. Meru slope (Ghiglieri et al. 2012), the maximum and mean values of 2.6 and 1.42 TU respectively were recorded in groundwater samples. The study also indicated tritium content of 2.8 TU in a rainwater sample. The low tritium content from both the previous and current studies suggest that groundwater in the study area is recently recharged.

Groundwater exploitation

The estimated uncorrected groundwater 14C ages for investigated wells ranged from 5,700 to about 600 years BP. Based on the Pearson corrected 14C ages (Table 5), only four wells (W08, W10, W11 and W19) revealed old groundwater (1,400 to 100 years). In terms of geological time scale, the ages seem to be very young; however, it is quite a long period of time for the case of continuous groundwater exploitation and its respective replenishment through vertical recharge from local precipitation. The rest of the sampled wells have modern groundwater, which suggests that the aquifer is actively recharged. The persisting water-level decline (Table 1) is probably a result of an over-pumping practice as the groundwater development is not adequately controlled in the study area. Nevertheless, groundwater ages have been measured and reported worldwide ranging from months to millions of years (Sturchio et al. 2004). Examples include (Bretzler et al. 2011) who reported groundwater radiocarbon age in the main Ethiopian rift systems ranging from 800 to 5,000 years BP at 80 to 260 m depths. In south-east Tanzania (Bakari et al. 2012a) reported 14C age of groundwater was found to be ranging from 13,000 years BP in deep confined aquifers to 1,300 years BP in shallow unconfined aquifers. Based on the current and previous studies, most groundwaters in the Rift valley systems were recharged in late Pleistocene and Holocene in the Quaternary period.

The age of groundwater is mainly affected by a number of factors, including aquifer type and geology of the area (Bretzler et al. 2011; Sukhija et al. 1996; Wassenaar et al. 1991). In the current study, young and modern groundwater (high 14C activity values) were observed in wells located at high altitude on the slopes of Mt. Meru, which is the potential recharge zone of the study area (Ong’or and Long-Cang 2007). Other wells observed to have young water are located in areas dominated by alluvial deposits in the northwest (Figs. 2 and 6). The oldest groundwater (1,400 years BP) was observed in a well located at low altitude (W19) in the central part of the study area. This zone is mainly dominated by the basalts formation, ranging from slightly weathered to fresh rocks (Fig. S1 of the ESM). Some of these wells are tapping water from different geological layers because of well-screen positions (Figs S1 and S2 of the ESM). The position of screens at different depths within a single well could mean that the samples collected represented mixed groundwater, i.e., young and old waters together. The mixed groundwater components could only be identified by applying a combination of tracers suitable for dating young groundwater (less than 50 years) such as 3H/3He or 85Kr (Corcho Alvarado et al. 2007; Mazor 1993; Sültenfuβ et al. 2011). The use of 14C tracer facilitated only identification of groundwater component with age greater than 100 years, whereas other components below this limit were categorized under modern groundwater. Generally, all samples considered in this research represent groundwater recharged in late Holocene.

Decrease in water levels and subsequent decline in well yields in the research area have been reported (Ong’or and Long-Cang 2007). This is in line with estimated groundwater ages of up to 1,400 years BP, as such a long period of time is required to replenish the continuously abstracted groundwater through production wells. Additionally, potential groundwater recharge due to local precipitation is expected to be reduced because most production wells are located within built-up residential areas and more development is still taking place, expanding towards potential recharge areas. The development is likely to be interfering with groundwater recharge mechanisms, particularly the infiltration rate of precipitation (Han et al. 2017; Rose and Peters 2001; Zomlot et al. 2017). Based on estimated 14C ages, the aquifer storage depletion in the study area is a result of groundwater over-pumping; however, future investigation on other factors likely to affect recharge mechanisms need to be undertaken. These may include effects due to climate variability, earthquakes and tectonic movement, as the region is also affected by these natural phenomena.

Conclusions

The use of carbon-14 and stable isotope techniques for groundwater investigations has revealed abstraction of relatively old groundwater from Arusha aquifers. Groundwater ages estimated in the study area ranged from 1,400 to 100 years BP; a long period of time like this is required to replenish the continuously abstracted groundwater through production wells. Generally, all groundwaters considered in this study were recharged in the late Holocene; however, no clear trends were established between 14C activity and well depth, probably indicating that most wells are tapping water from the same aquifer but at different depths or layers depending on screen positions. The oldest groundwater (1,400 years BP) was observed in a well located at low altitude in the central part of the study area. This zone is mainly dominated by basalts formation, ranging from slightly weathered to fresh rocks.

The 14C results revealed young and modern groundwater in samples collected from shallow aquifers i.e., spring water and wells located in areas dominated by alluvial deposits at high altitudes. The presence of modern groundwater suggests that shallow aquifers are actively recharged and respond positively to seasonal variations. Additionally, spring flow data showed positive response during the rainy season by increasing flows significantly and vice versa during dry months, indicating that water sources including recharge zones in the study area require a protection strategy to avoid drying up impact as well as decrease of discharge. Potential groundwater recharge due to local precipitation is expected to be reduced because most production wells and springs are located within built-up residential areas and more development is still taking place, expanding towards the recharge areas. The development is likely to be interfering with groundwater recharge mechanisms, particularly the infiltration rate of precipitation. Due to continuous pumping of groundwater for public water supply in the study area and high mean residence times, as revealed by 14C results, the existing groundwater abstraction rate in the study area will not sustain the future demand. However, future investigation on other factors likely to affect recharge mechanisms needs to be undertaken. These may include effects due to climate variability, earthquakes and tectonic movement, as the region is also affected by these natural phenomena.