Groundwater evolutions in the semi-arid aquifers of Dodoma City, Tanzania using stable isotopes of water and hydrogeological facies

Growing human populations coupled with fast socio-economic development in Tanzania have greatly increased the need for fresh water. While this trend has been widely observed in many national cities and towns recently (Kashaigili 2010), the situation in Dodoma City is worse owing to the sudden influx of people into the area already known for its low rains, limited surface water (SW) sources, and high dependence on groundwater. Yet, to account for shortages or poor-quality SW sources, the city has since its inception heavily relied on groundwater (Rwebugisa 2008; Taylor et al. 2012). However, the groundwater resources in the fast-growing and recently declared national capital city are often poorly understood and managed, further adding to the city's water stresses (Kashaigili 2010; Taylor et al. 2012). Regarding groundwater provenance, there is either no knowledge or only partial understanding of the system, mostly limited to the Makutupora wellfield (MWF), which is the main source of water supply to the city. This leaves the groundwater situation in much of the rest of the city largely unknown, where numerous boreholes (BHs) and dug wells (DWs) exist and continue to be installed.

Further, recent climate studies reveal that local evaporation has increased and there are clear worsening drought intensity and severity. Similarly, rainfall, the predominant source of groundwater recharge in the area has also significantly decreased (Chang'a 2010; Shemsanga et al. 2016; Luhunga et al. 2018). These adverse shifts in climate trends have led to SW bodies drying out faster and/or becoming highly mineralised rendering them unsuitable for use for much of the year (Shemsanga et al. 2018). Hitherto, studies clearly show that, as rainfall and SW bodies become less reliable, groundwater continues to bear the increased burden of fresh water supplies (Calow et al. 2010; MacDonald et al. 2012). Thus, more pressure is exuberated on the relative drought and pollution-tolerant groundwater system (Calow et al. 2010; MacDonald et al. 2012) to meet the increasing water demands for much of the year.

Yet, in response to the shortage of freshwater sources, and the fact that local water supply is mostly biased towards the city centre and posh areas, about 1,369 BHs and 3,428 DWs have anonymously been sunk in low-income neighbourhoods and city suburbs (Shemsanga et al. 2018). Thus, these unregistered wells potentially drain a substantial amount of water from the system which is only partially understood. However, the unconfined aquifers to which the DWs occur are understood to be interconnected to the deep aquifers (DAs) (Shindo 1990); hence, the need for holistic management as required by the national water policy and integrated water resources management approaches to which Tanzania water resources management strategies abide with URT (2002).

Further, to ensure adequate water supplies, the government recently drilled and rehabilitated more than 52 BHs within the city including in Nzuguni, Ihumwa, MWF, and Mtumba. This is happening while there are strong signals of periodical declining groundwater levels in the MWF and local rains, the main source of recharge has slightly decreased or become less predictable (Taylor et al. 2012; Luhunga et al. 2018). These dynamics make it crucial to assess how the hydrogeological system operates as a whole and survives the harsh climatic conditions including in what way the blue water distributes into the limited SW sources and groundwater, both shallow aquifers (SAs) and deep (DGs). The partitioning of rainwater into the various water pockets and mixing of different water masses thereof could best be done via multifaceted approaches combining SIs and chemical species of water (El Mountassir & Bahir 2023; Gupta et al. 2023; Moussaoui et al. 2023; Su et al. 2023; Zhang & Dong 2023). SIs (δ¹⁸O and δ²H) occur naturally in the environment and are especially useful for understanding hydrological circulations, the origin of water masses and their interactions (Pandey et al. 2023).

Within the City, limited SIs of water (δ²H and δ¹⁸O) studies have been done. These already identified Chenene Hills as the prominent recharge site for the MWF (Dincer 1980; Aly 2000). Further, Shindo (1990) estimated that about 10% of annual rains are responsible for recharging the MWF. These studies, however, mostly focused on the MWF, leaving the situation in much of the city's catchment, which is already heavily pumped, largely unknown. The studies also did not sufficiently address the interactions between different water masses in much of the city. Nevertheless, central to the sustainability of the system requires clear knowledge of how the system operates as a whole to sustain the populace against low rains, high evaporative forces, and more-or-less continuous abstractions from multiple points (Taylor et al. 2012). Effective management of such a complex hydrological system, therefore, calls for a detailed understanding of the entire catchment, hydrogeological characteristics, recharge fluxes, and interactions of various water pockets (Cook & Herizeg 2000; El Mountassir & Bahir 2023).

Due to their inertness, SIs have broadly been used to map the origin and residence time of water masses (Sethy et al. 2022; El Mountassir & Bahir 2023). The suitability of SIs for hydrological studies, including as ideal tracers for recharge areas and water flow patterns, is to account for their signature in rain (the main recharge source) which is systematically affected by altitude, latitude, air temperature and ocean proximity (George et al., 2006). Moreover, SIs are sensitive to physical forces namely mixing, evaporation (E), and salinisation. Thus, as water moves between reservoirs, its SI signature changes in quantifiable manners helping to trace its provenance and interactions thereof (Clark & Fritz, 2013; Su et al. 2023). These qualities are vital for (i) mapping recharge even if it took place afar and (ii) Crafting an effective water management strategy for the area (Clark & Fritz, 2013).

It is for these reasons that SIs of the water have widely been used to assess groundwater dynamics and interactions with SW and provide insights into the mixing of water masses (Gupta et al. 2023; Moussaoui et al. 2023; Zhang & Dong 2023). Further, SIs provide a long-term understanding of SW–groundwater dynamics and complement the traditional hydrogeochemistry in delineating the provenance and movements of both solutes and water (Kendall & McDonnell 2012; Biddau et al. 2023; Pandey et al. 2023). For instance, aquifers recharged by rains are distinguished from those recharged by lakes and/or rivers based on differences in SIs owing to evaporation effects (in Lakes) or river meandering through different altitudes and latitudes (Clark & Fritz, 2013). Furthermore, SIs also provide past groundwater behaviours and help predict its future situations (Clark & Fritz, 2013). Similarly, when combined with dating of groundwater SIs have been used to trace temporal and spatial water circulations in various reservoirs and have been very useful in establishing sustainable groundwater management protocol (Biddau et al. 2023; Gupta et al. 2023; Moussaoui et al. 2023). Thus, blended with traditional methods, SIs are considered less resource intensive, make groundwater studies more complete, and often simplify the understanding of complex hydrogeological systems and hence their effective management (Biddau et al. 2023).

This paper, therefore, assesses the origin and behaviour of various water resources and possible mixing between SW (rainfall, river, and dams), groundwater (shallow and deep), and environmental factors (geology and climatic). This information is vital in reaching a sustainable exploitation and management of water resources including such vital details as the protection of recharge areas and deducing the origin of the water that is being pumped from the city.

Standard procedures for water sampling, preservation, transportation, and chemical analyses were conducted (Greenberg et al. 2005). Water samples were collected from rainfall (10), DGs (76), SGs (23), rivers (2), and dams (4), two from each Hombolo dam (HD) and Matumbulu dam (MD) between December 2021 and June 2023. The rain events samples were collected from MWF, the city centre, Hombolo Suburbs, Chenene Hills, Dodoma Meteorological Station, and Ihumwa suburb. Tightly connected to 1 L plastic bottles were 60 cm diameter plastic funnels that were placed at 1.5 m above ground to collect rainwater. All sampling bottles were covered with aluminium foils and emptied immediately after rain events. BH and SGs sampling were done in triplicate using a bailer sampler and water pumps, respectively. Immediately after filtering using 0.45 μm syringe filters, water samples intended for hydrogeochemical investigation were collected in 0.5 L air-tight high-density polyethene (HDPE) bottles washed in the laboratory before being rinsed thrice by the sampled water. SIs samples were collected in air-tight vials and stored below 4 °C before their analyses at the State Key Laboratory of Estuarine and Coastal Research of East China Normal University based on standard procedures (Clark & Fritz, 2013). The analytical precision for δ²H and δ¹⁸O were +0.80‰ and +0.10‰, respectively.

From the existing rainfall SIs from across the country, an LMWL was established to be ²H = 7.43δ¹⁸O + 9.77 and the correlation between δ¹⁸O vs. δ²H was 0.92, a remarkably close agreement (Figure 3). The relationship between δ¹⁸O vs. δ²H from SW, DGs and SGs were, respectively, ²H = 5.84δ¹⁸O + 2.42., ²H = 5.097δ¹⁸O −1.2898X and ²H = 6.69δ¹⁸O + 8.78. These equations show that DGs have the lowest slope followed by SW and finally SGs (Figure 3). The different slope patterns signify recharge source diversity with DGs most probably recharged by heavy precipitations while SGs get more periodical renewals.

Figure 3(a) and 3(b) show the vertical profile of δ¹⁸O and δ²H where, in both cases, there are increasing depletion SIs signatures with increasing depths. Further, Figure 5(a) and 5(b) show seasonal variations of SIs where higher rains are mostly associated with heavy SIs enrichment in DGs. These two graphs offer vital insights into water mass mixing and interactions in the city.

Figure 6 shows the spatial distribution of δ¹⁸O and δ²H where among DGs, MWF has the highest enriched SIs values followed by Hombolo and the city centre sub-systems. Further, it is shown that δ¹⁸O and δ²H range from −5.61 to −2.1‰ and −33.62 & −6.98 ‰ and have SD of 0.84 and 6.38 (Table 1, Figure 6). These findings show that (i) there is a sudden enrichment of groundwater around HD compared with the most immediate surroundings. (ii) There are micro spatial variations in SIs abundances from the DGs, especially along the city centre hills and the surrounding areas of Ihumwa and Mtumba.

Table 2 shows that HD and MD contribute 57.52 and 22.24% to respective SGs while HD adds up to 16.98% to the underlying DA system. Further, the Makutupora aquifer feeds Hombolo DG for about 75.36% while Makutupora DGs received about 9.56% from the overlying SAs (Table 2).

The major cations and anions are Na⁺ and (⁠ & Cl⁻), respectively (Figure 7). Figure 7 reveals the main water types are NaHCO₃, mainly in MWF while NaCl types mostly occur in the city centre and the flood plain from the city towards Hombolo through Nzuguni, Kitelela, and Mahomanyika suburbs.

Figure 8(a) and 8(b) show the characteristic distribution of Cl⁻ and δ¹⁸O relative to potential associations among various water masses. Notice the close association between (LKR, springs, DGs, and SGs) and several (DGs and SGs) (Figure 8(a). The close association between (LKR, springs, DGs, and SGs) and the (DGs and SGs) is also shown in Figure 9(a) and 9(b). These figures are used to show associations and/or common originality between different water masses, which was very much pronounced in the city.

Figure 9 gives relations between Cl⁻, δ²H, and Na⁺ among water sources regarding their SIs enrichment and possible origin. Figure 9(a) shows that except for four points, most SGs have low Cl⁻ and Na⁺ ions and that generally, there are significantly more Cl⁻ than Na⁺ ions in most DG samples. Similarly, Figure 10(b) shows higher Cl⁻ ions in DGs than in SGs, except for the same four cases. Otherwise, rivers are more depleted in SIs than springs, DGs, SGs and dam waters, in that order. Notice the strong overlapping in SIs between SGs and DGs in Figure 9(b). The results show that MD is highly enriched but not as high as HD and the latter is by far more mineralised than the former. Not to be missed in the trend are the highly mineralised SGs close to HD, which also tend to have relatively enriched SIs.

Figure 10 shows the relationships between δ¹⁸O with various physicochemical facies. Displayed in all but Figure 10(f), M & N, are increasing δ¹⁸O enrichment with mineralisations. Notice the corresponding depletion of δ¹⁸O as the environment becomes more alkaline (Figure 11(f)). Further, Figure 9 shows a close correlation between and Cl⁻ where high is nearly always associated with corresponding high Cl⁻ suggesting a possible common source, possibly anthropogenic.

Figure 11(a) shows the spatial distribution of DO among DGs while 10B illustrates characteristic DO concentrations in various water pockets. It is observed that DGs are mostly anoxic followed by SGs, dams, rivers, and rain samples, in that order. The highest, lowest, and average DO values among the DGs are 5.4, 2.32, and 0 mg/L, respectively. Notice the characteristic upward departure of dam samples while spring occurs close to DGs and LKR is very much close to LMWs (Figure 11(b)). It is shown that a few SGs are more anoxic than others and occur among the DGs (Figure 11(b)). Finally, it is shown that river samples plot close to rain samples that have relatively higher DO concentrations (Figure 11(b)).

In this study, δ¹⁸O and δ²H, results from DGs, SGs, springs, dams, LKR, and rains are highly variable (Table 1). The rainwater SIs show that its compositions were fairly depleted (δ¹⁸O −5.52‰) and (δ²H −32.99‰) (Table 1). However, compared to previous studies in the area, these values are slightly more depleted. Shindo (1990) reported average SIs values of −4.31 and −20.01, respectively. While variations of SIs are common for tropical climates (Dansgaard 1964), it is also attributed to the fact that the study area has highly variable temporal and spatial air masses (Shindo 1990). Shindo (1990) identified that the southern part of the Inter Tropical convergence zone, ITCZ passing in the area at the onset of the rainy season had δ¹⁸O of between −8 and −4. These abundances are significantly different in the northern part of the ITCZ passing in February with a value of −2 to 0, respectively. Thus, the relatively higher values observed in the current study could be attributed to the fact that the average rainwater SIs by Shindo (1990) did not include rainwater samples from northern ITCZ, which have lower SIs values (Shindo 1990).

Generally, except for a few cases, the majority of rainwater aligned on the left lower end of the LMWL. These values, nonetheless, are still within the normal range for the Indian Ocean region (from −38 to −6‰, and from −6 to −2‰, respectively) (Povinec et al. 2012). Compared to deep groundwater samples, the δ²H and δ¹⁸O ranged from −33.62 to −22.1, and −5.61 to −4.13 ‰, respectively. The significant difference in rainwater and groundwater values and characteristic low slope of δ²H vs. δ¹⁸O relative to the GMWL (Figure 3), defined by Craig (1961) as δ²H = 8.13 δ¹⁸O + 10.8 could suggest an isotopic fractionation as groundwater percolates and circulates through the various aquifer media (Schiavo et al. 2009; Povinec et al. 2012; Pandey et al. 2023).

Worth noting, recent years have witnessed major shifts in climatic regimes in terms of rainfall longevity and intensities and increasing temperature and evaporation all of which affect SIs fractionation (Bakari et al. 2012). Further records show that there has been an increase in about 1.13 °C over half a century (Shemsanga et al. 2016) which could significantly increase SIs enrichment. The same study also found a decreasing rainfall of about 55.9 mm over the past nine decades. Further, both temperature and evaporation are highest during wet spells of the year (Chang'a et al. 2010; Luhunga et al. 2018) suggesting a higher likelihood of SIs enrichment. Thus, the worsening climate regimes in terms of decreasing rainfall and increasing temperature and evaporation, threaten the future availability of groundwater (Calow et al. 2010; MacDonald et al. 2012). This may further explain why HD, MD, and some SGs are particularly enriched with heavy SIs relative to local rains. For semi-arid locations, however, it is also common to find that water is enriched in the unsaturated zone during the recharge process (Shindo 1990; Bakari et al. 2012). Thus, these results support previous studies in the MWF that past recharge happened during a much cooler climate (Dincer 1980; Shindo 1990).

A scattergram of δ¹⁸O/δ²H shows that except for a few cases, the majority of samples from DGs, SGs, springs, and rivers distribute close to and along LMWL while HD and MD divert below and to the right of the line (Figure 4). Like for most DGs, all springs align below LMWL and are remarkably close to most DGs. From Figure 3, groundwater samples can be grouped into three classes: (i) from SGs that largely plot above LMWL, (ii) from DGs and springs that largely align below the LMWL, and (iii) from overlapping samples from SGs and DGs plotting below and above the LMWL, respectively (Figure 3). Thus, SGs are more enriched in heavy SIs than most DGs (Figure 3). Except for a few DWs that are enriched by evaporation and plotting below LMWL, DGs are generally more depleted in heavy SIs than the DWs and local rains. Globally, the spatial distributions in precipitation isotopes show that more isotopically depleted rains are common in the cooler regions and/or months of the year while more enriched values dominate warmer seasons and hotter ecosystems (Clark & Fritz, 2013). In areas closer to the equator, the precipitation amounts mainly drive the annual differences in the isotopic compositions of rainfall, with isotopically depleted rains during higher rainfall (Clark & Fritz, 2013). Thus, precipitation events (the main sources of recharge globally) have specific isotopic fingerprints that can be traced and mapped as water flows through an ecosystem (Pandey et al. 2023; Su et al. 2023). The current results show that with DGs, only heavy rains that are accustomed to more depletion in heavy SIs affect the most recharge or DAs are recharged from afar, and the waters become trans-located to the area and in the process adjusting the SIs signatures (Shindo 1990; Bakari et al. 2012). However, a few BHs had relatively more enriched waters than the average and these were remarkably similar to SGs (Figure 4). This could be attributed more to the effect of the mixing of water due to pumpage and recharge as the area is characterised by increasing SIs depletion with increasing depth (Figure 4) where vertical changes in groundwater isotopic content are observed (Figure 4). Further, small, localised enrichments are also possible through such processes as solute concentrations (through transpiration and/or evaporation), and/or mineral dissolutions (Lee 2015; Pandey et al. 2023). Figure 10 supports this observation where there are more enriched SIs with increasing mineralisation.

LMWL is represented by ²H = 7.43 δ¹⁸O + 9.77 where δ¹⁸O and δ²H have a strong relationship (R² = 0.92). It is worth mentioning that this slope is slightly lower than the GMWL, ²H = 10.8 δ¹⁸O + 8.13, R² = 1 (Figure 3) which is generally not uncommon in East African Equatorial regions due to the prevailing rains originating from highly variable temporal and spatial air masses (Shindo 1990; Bakari et al. 2012). Confining samples from MWF alone, Onodera et al. (1995) found the relationship between δ²H and δ¹⁸O to be of the order of δ²H = 7.8 δ¹⁸O + 16.6. This equation is significantly different from the equations represented in Figure 3, which is attributed to the relatively wider aerial coverage of the current study hence more hydrogeological diversity, including in areas out of MWF. In this study, however, attitudinal and rainout effects (ROI) that are common with SIs (Bakari et al. 2012) were not pictured. This could partly be attributed to the limited aerial and altitudinal coverage of the study area with complex sources of moisture (Figure 3) (Shindo 1990). However, compared to a study held in Dar-es-Salaam, current SIs for rainfall, groundwater, and rivers are comparatively highly depleted due to both altitudinal and ROI effects (Bakari et al. 2012).

The movement of water molecules within a specific catchment can be traced by analysing the specific abundances of SI (δ¹⁸O and δ²H), which is particularly valuable for data-scarce aquifers (Pandey et al. 2023; Su et al. 2023). In the semi-arid aquifers of Dodoma, the use of SI as tracers provides a powerful means to study various aspects of the groundwater system, including the sources of groundwater recharge, interconnections between aquifers, the interactions between SW and groundwater, and the effects of evaporation or decreasing precipitation and increasing local temperatures and evaporation (Pandey et al. 2023).

A comprehensive investigation of SIs offers a valuable tool to acquire knowledge on the regional groundwater circulation patterns and geochemical processes within the aquifer system, especially for data-scarce aquifers like the one in Dodoma City (Pandey et al. 2023). This approach is particularly valuable in regions with limited hydrogeological data, as the stable isotope signatures can provide insights into the complex dynamics of the groundwater system that may not be readily apparent from other data sources or where not enough data are unavailable.

A plot of δ²H vs Cl⁻ is used to explain recharge fluxes and possible interactions of water sources of varying degrees of SIs enrichments (Sklash & Mwangi 1991; Lee 2015). Likewise, the evaporation and mingling of two distinct groundwater sources spanning along a linear trend can be distinguished by a δ²H vs. Cl⁻ trend line (Schmerge 2001; Lee 2015). From Figure 9B, it can be demonstrated that local dams have the most enriched SIs values while springs and rivers have the least. Further, it is shown that DG samples align right below SGs suggesting that on average, the former is more depleted in heavy SIs than the latter (Sklash & Mwangi 1991). Since both δ²H and Cl⁻ increase with evaporation, the following sets of explanations can be drawn (i) compared to SGs, DGs were recharged with water that had undergone minimal evaporation. (ii) SGs are locally more affected by evaporation. (iii) The interactions between different water masses bring about the observed mixed trends and overlaps among water masses. (iv) Except for the extreme cases with DWs, the majority of DGs have higher Cl⁻ than SGs for the obvious reason of sustained longevity with the rock matrices and the associated dissolution of chloride-bearing minerals (Shindo 1990). (v) The characteristic plotting of HD and MD SIs abundances suggest that evaporation is not the prime reason for high Cl⁻ concentration as both dams occur in more-or-less similar settings but the former has highly saline waters than the latter. This may suggest that the salinisation of HD is of geological origin (Shindo 1990). (vi) Salinity does not play a key role in SIs enrichment as some DWs had high salinity but were not as highly enriched as HD waters. Thus, evaporation is the major mechanism for SIs enrichment locally, which is in line with other global studies (Bakari et al. 2012; Povinec et al. 2012; Khaska et al. 2013; Lee 2015).

Further, DWs close to MD and HD offer the best examples of water masses interaction where from the perspectives of specific SIs signatures, and through two components mixing model equation, it is shown that about 22.2 and 57.5% of the water in DWs originate from the highly enriched MD and HD, respectively (Table 2). Further, it is shown that 17.0% of the DGs in the Hombolo catchment originate from the highly enriched HD. Thus, the sedimentation of the dam and its declining water levels in recent years (Mahenge 2019; Gayo 2021) will ultimately impact both SGs and DGs alike as the dam is a vital recharge source. Thus, measures must be taken now to protect the systems by limiting adverse human activities from the catchment slopes where the bulk of the sediments originate. Similarly, the runoff diversion for agricultural activities in the upper catchment should also be controlled as fresh runoff is highly needed for the dilution of the dam and to sustain its levels (Shemsanga et al. 2018). Finally, the contributions of the Makutupora aquifer to the Hombolo groundwater subsystem were found to be remarkably high (75.5%) hence the need for co-conserving the two as they are highly interdependent (Table 2). This would bring the catchment closer to the nationally highly advocated principles of integrated water resources management (URT 2002).

In the bid to avoid the use of high saline HD waters, and the threat of dry wells in dry months but benefit from high yields of the dam surroundings; local people have dug several SWs close to HD. However, chemical analysis of the DWS suggests high levels of HD-SG interactions and the chemistry of the water in one such sampled well was highly mineralised and nearly that of HD (Table 2). SIs from the well also support the interaction as its water has highly enriched δ¹⁸O and δ²H compared to SWs from afar and HD water is represented for about 57.52% of the mixture. This is yet another proof of actively interacting water masses, this time involving the dam and SGs.

Figure 7 shows the characteristic distribution of SIs among DG water where several key messages are highlighted: (i) BHs around the HD have relatively more enriched SIs. This suggests that part of their water might have originated from the highly enriched HD. (ii) MWF has the highest enriched values among the deep groundwater. It is worth highlighting that MWF is the major discharge site for the city, which is habitually associated with higher mineralised water (Tularam & Krishna 2009; Povinec et al. 2012). In the current study, the high correlation between mineralisation and SIs abundances has been well observed (Figure 10). (iii) The prominent recharge sites of Chenene Hills (NE) have the most depleted values, very near to those of rainwater (Figure 4). This is in line with general hydrogeological studies that recharge sites are often associated with fresh waters (Tularam & Krishna 2009; Pandey et al. 2023; Su et al. 2023) and in the local area, more depleted waters.( iv) The city centre has average SIs signatures, nonetheless more enriched values than areas around the Chenene Hills but not as the case at the MWF. This could be explained by the fact that micro-scale recharges might be happening in the many hills surrounding the city and those in Ihumwa and Mtumba suburbs. These provide freshwater into the pumped system and hence the observed average SIs. Thus, except for a few cases, BHs closest to Chenene hills, as is the case for all other BHs are more depleted than those in the MWF, and to some extent, the city centre and Hombolo suburb (Figure 6). This is partly justified by the fact that MWF, the city centre and Hombolo are among the heaviest pumped areas (Shemsanga et al. 2018) which tend to be more mineralised and enriched SIs waters (Figure 10).

Unlike previous studies, however, cases of DGs with less depleted heavy SIs are also observed in the mixture. This could be due to the following: (i) the influence of pumping in mixing waters of different SIs signatures (Figure 4). (ii) Potential mixing of SGs and DGs (Figures 8, 9 and 11). (iii) Potential mixing between SWs and DGs (Figure 12). Further, notice that all springs are relatively more depleted in heavy SIs than BHs and span very much closer to SIs abundances of local rain. This could probably suggest that springs are more directly influenced by local rains and that, unlike DG water, springs receive recent recharge from local rains. A previous study found that the flow in all springs significantly decrease during day months suggesting that they are mostly replenished by rainwater (Shemsanga et al. 2018). Similarly, rivers are more depleted in heavy SIs than BHs but plotting below LMWL and GMWL which suggests that they are prominently of local meteoric origin as opposed to baseflow contribution from the groundwater or the extent of groundwater contribution is insignificant compared to meteoric. Aligning below these lines, however, may suggest that the rivers have undergone some levels of evaporation-related enrichment as they meander along the catchment (Figure 3). Comparable results were reached by Shindo (1990).

Similarly, there was a high association between the amounts of monthly rainfall with SIs signatures in the DAs mainly due to dilution. From Figure 4, it is observed that higher rains are largely associated with more enriched DGs. This suggests that when strong rains lead to recharge, the aquifer becomes more enriched in heavy SI. Conversely, when weaker rains lead to little or no recharge the DA is more depleted with increasing depths (Figure 4). Thus, with no or less recharge, pumping taps relatively deeper depth, which is more depleted (Figure 4). These findings generally agree with recent studies showing that groundwater level in the aquifer is largely a function of rainfall intensity where only extreme rains trigger recharge (Taylor et al. 2012). Notice that both HD and MD have extremely elevated levels of SIs enrichments. However, while HD water is highly mineralised (Table 1), MD is less so, and its values are nearly comparable to that of rain. Thus, the enrichment of SIs, in both cases, is likely due to higher evaporation than anthropogenic and environmental/geological factors (Bakari et al. 2012).

Compared to local rains, SGs are more enriched in heavy SI. Yet, some SG samples align on the same slope as LKR and LMWL while LKR plots are very close to rain samples (Figure 3). This trend could be due to: (i) active SW groundwater interactions. (ii) SG waters mostly are of fresh local meteoric origin percolations. (iii) LKR is mostly of a complete local meteoric origin with a minimal degree of enrichment (Figures 8, 9 and 11). The first case is best explained by what was noted with DWs close to MD and HD, in which, the wells located closest to dams had highly enriched values and diverted below the LMWL and to the right compared to other DWs. These characteristics suggest the movement of water from the dam into the SGs (encircled in red, Figure 3). Thus, the rightward departure of HD and MD from the LMWL does not mean that the waters are from different sources but rather, have undergone significant enrichments. This is further supported by the seasonal variations of the SIs signatures where the values are closer to the LMWL during the middle of the rainy season and furthest in October when it is about to rain (Figure 4). Thus, the rightward departure from the LMWL suggests non-equilibrium evaporation (Shindo 1990). However, SIs signatures of the wells are not as enriched as those in the dam hinting a possibility that the SA water is mixing with the dam water to some degree and/or the SAs also easily benefit from local rains. It is likely, therefore, that the shallow semi-arid craton aquifer of Dodoma is mainly recharged by both meteoric water (MW) and SW; however, further investigations are needed to justify this hypothesis.

It is shown from Figure 3 that among SW samples, LKR plots to the left of LMWL and GMWL while for dam values, except for one sample from MD which was collected at the beginning of the rain season, plot below both lines and to the far right (circled in blue). The characteristic departure from the GMWL portrays a considerable evaporation (Clark & Fritz, 2013). Worth noting, LKR plots along the scatter span of the majority of local rains, a strong implication that it is more nourished by rainfall than groundwater. This suggests that much as the river is seasonal, the declining rainfall magnitude that nourishes the river will likely affect the local ecosystem.

Except for a few cases, it is noted that the majority of DGs and SGs align below and above the LMWL and GMWL, respectively. This trend could be attributed to the fact that SGs are mostly freshly and easily replenished by precipitation, which shows a closer resemblance to the isotopic signatures or may simply imply that SGs are not as depleted as the DGs (Shindo 1990). However, cases of some DGs and SGs plotting above and below LMWL and GMWL were also recorded (Figure 3). The characteristic SGs plotting below and to the right of the LMWL and GMWL are most likely due to localised non-equilibrium enrichment by evaporation (Shindo 1990) or the effect of salinity that has been associated with isotopic enrichment of water molecules (Clark & Fritz, 2013).

Similarly, the plotting of groundwater to the right and parallel to the LMWL indicates that such water underwent a re-evaporation after initial condensation before the rain reached the recharge zones hence, causing further kinetic fractionation on the rains (Clark & Fritz, 2013). Otherwise, three schools of thought could explain the characteristic plotting of the DGs above LMWL and GMWL as follows. (i) There is a potential effect of mixing between SGs and DGs (Figure 12). (ii) There is a potential effect of pumpage that mixes deeper groundwater that, in the study area, is more depleted in heavy SIs (Figure 3). (iii) The samples plotting away from LMWL could be of paleo-water origin which often plots far from the line (Yurtsever & Araguas 1993). Previous studies had dated groundwater in MWF in the order of thousands of years (Shindo 1990). Compared to local rains, however, the majority of SGs and DGs align to the right and are generally more enriched. This suggests that local groundwater is characterised by some evaporation-related fractionation (Clark & Fritz, 2013).

In the bid to assess the validity of these hypotheses, two components mixing between DGs and SGs at MWF and Hombolo sub-catchment were conducted (Equation (3)). The results show that SGs contribute to about 9.6% of the DGs (Table 2). The most likely mechanism for the SGs mixing with DGs is through a well-pronounced network of fault systems in the area when the SA is saturated with fresh rains (Rwebugisa 2008). These faults are also regarded as the direct avenue through which the aquifer quickly responds to precipitation and pollutants alike. Previous studies found the recharge lag to be of the order of 1–2 months whereas, for the complete confined system and the fact that the area experiences low rains (550 mm per year), it would take longer for the groundwater level to respond after rain events (Shindo 1990). These results contradict earlier findings that SGs in MWF do mix with DGs (Rwebugisa 2008). Close observation of BHs SIs abundances within and outside MWF however clearly shows that some points have more depleted values suggesting some kind of mixing water masses. While addressing the issue of high nitrates in MWF, Clark & Fritz (2013) also suggested the possibility of direct runoff entering the aquifer system via the fault system and the characteristic rapid response of the aquifer to rains and pollutants.

The results show that local groundwater samples are distinctively classified based on salinity, pollution, and longevity to which the water has interacted with rock matrices. The results show that the dominant anions and cations are Na⁺ and ⁠, respectively. The ionic ratios of major dissolved constituents show that cations and anions follow the order > Cl⁻ > > and Na⁺ > Ca²⁺ > Mg²⁺ > K⁺ order and have respective average concentrations of 252.0, 160.3, 74.4, 61.8 and 148.3, 60.6, 31.6, 15.1 mg/L. Piper diagram further shows that the major water types are (I) NaHCO₃ mainly occurs in the MWF and represents old waters (Bakari et al. 2012) and (II) NaCl is mostly in and around the city suburbs and in the flood plains from the city leading to HD through Nzuguni, Kitelela, and Mahomanyika suburbs (Figure 7). Hitherto, pockets of other highly variable water types occurring in the mixture are also not uncommon, which represents the complexity of the hydrogeochemical formations from which the water draws its signatures (Figure 8, Figure 12). This trend was also reported by Rwebugisa (2008).

Figure 10 shows more enriched SIs with increasing mineralisation where the majority of graphs have positive slopes suggesting more enriched heavy isotopes are encountered right-wards. Also notice the close similarity between Cl⁻ and distributions suggesting that the two ions are likely to be of the same origin, principally anthropogenic. Previous studies in the area associated the increasing with human activities related to sewage, livestock wastes, and agricultural inputs and it is likely that some Cl⁻ are also from these sources (Dincer 1980; Shindo 1990; Aly 2000; Rwebugisa 2008).

Except for two cases (where SGs had unusually high EC of 19,876 and 21,000 μS/cm, respectively), the majority of SGs are not as heavily mineralised as DGs with an average EC of 943.14 and 1262.6 μS/cm, respectively. The high mineralisation of DGs likely originates in the weathered zone matrices and fractures when water molecules pass and interact with them (Shindo 1990).

Figure 9 offers an initial perspective of water interactions. The close proximity of (DGs and SG), (river, DGs, and SGs) (springs, river, SGS, and DGS) samples indicate the possible hydraulic interactions or similar origin of the water masses (Panno et al. 2006). Figure 10(a) further shows the relationship between Na⁺ and Cl⁻ and helps to identify groundwater origin based on their chemical compositions (Schmerge 2001; Li et al. 2008). The results show that the ratio of Na⁺ and Cl⁻ for DGs and SGs has R² of 0.65 and 0.94, respectively. It is worth noting that the natural sources of ions are rock–water interactions, atmospheric deposits, and human-related sources namely agricultural inputs and human wastes are vital (Panno et al. 2006). However, concentrations of Na⁺ were comparatively higher than Cl⁻ in most SGs (Figure 9(a). While high Na⁺ irrespective of Cl⁻ and ions concentration suggests characteristic rock–water interactions with Na-clays and Na-feldspars present as silt-and sand-size fragments in the soil zone and aquifers, the reverse could mean more significant anthropogenic pollutants from agricultural inputs and human and livestock wastes (Panno et al. 2006). A close correlation between and Cl⁻ is also demonstrated in the current results (Figure 10). High concentrations of these two ions in water have often been associated with human wastes (McQuillan 2004; Malla et al. 2015). However, the fact that the DGs have a higher concentration than SGs shows that DGs have had longer residence time with the rock matrices and that this is the major geochemical mechanism. Otherwise, SGs would have responded faster to and SGs would have higher values than the DGs, which is not the case.

The δ¹⁸O of the DGs ranges from −5.61 to −2.1 ‰, and has a characteristic high correlation with physicochemical parameters namely DO, EC, pH, ⁠, ⁠, ⁠, ⁠, and Cl⁻ (Figure 9). The anoxic condition in the DGs in MWF suggests comparatively longer residence time (Figure 10(a) and 10(b)). These results support previous findings that some water from the MWF water is of the order of thousands of years, ∼3,794–5,203 years (Shindo 1990) which could be mixing with younger waters of recent replenishments from precipitations. These traits suggest the water has had longer contact with the rock matrices leading to a higher rate of mineralisation than SGs (Dincer 1980; Shindo 1990; Aly 2000). Figure 11(b) shows that except for a few overlaps, DGs have the least DO followed by the springs, SGs, dams, rivers, and rainfall samples; in that order. Further, it is shown that river waters plot close to the rains and have higher DO. This suggests the rivers are not nourished by baseflows but are mostly of meteoric origin. In terms of DO spatial distribution, the most anoxic water occurs in MWF followed by Hombolo suburbs (Figure 11(a)). While the anoxic conditions could mostly be related to the mineralisation and presence of paleo waters (Shindo 1990), the case in the Hombolo subsystem is most likely due to contamination where some local BHs had very high nitrate- values of up to 223.7 ppm (Table 1). These findings support other studies suggesting the reduction conditions of organic carbon oxidation and nitrification decreased DO (Ghiglieri et al. 2010; Okiongbo & Douglas 2014). Further, high-nitrate water in Hombolo suburb could suggest more DO is involved in the nitrification process (McQuillan 2004). Finally, the relatively higher DO in and around the city centre may suggest that the groundwater is frequently replenished by fresh meteoric, oxygen-rich recharge waters (Chen & Liu 2003). For instance, such areas could be located in relatively permeable overburdens leading to a rapid recharge process, and/or the aquifer system could be unconfined, which allows free passage of fresh MW into the system (McQuillan 2004; Ghiglieri et al. 2010).

Hydrogeochemical and SIs of water were used to assess the origin of water sources, how the complex and semi-arid hydrogeological system in Dodoma city operates as a whole and in what ways the various water components interact with one another. The results show that except for HD and MD, most groundwater and SW sources closely span along or close to the LMWL and very much near the GMWL. Statistically, the δ¹⁸O abundances for DGs, SGs, Springs, Dams, LKR, and Rains ranged from −5.61 to −0.21, −7.45 to. −0.2.99, −5.12 to −4.98, −0.49–6.32, and −6.02–5.05 to and had averages of −4.13, 1.04, −5.06, 4.06, −5.54, and −5.52, respectively. This suggests that local rains play key roles in sustaining both surface and groundwater replenishments. However, DG samples are largely more depleted in heavy SIs than SGs suggesting only heavy rains effect their ultimate recharge. Thus, the decreasing rainfall magnitudes and longevity, in the city, will most likely affect the availability and sustainability of local water resources and dynamics, particularly deed aquifers, and further strain the potable water supply for increasing needs.

Surface–groundwater interactions were observed in DWs close to MD and HD where it was established that 22.2 and 57.5% of the waters originate from highly enriched MD and HD, respectively. Similarly, 17.0% of Hombolo suburb's DGs originate from HD, thus, the recent sedimentation and declining dam levels may ultimately impact the local sustainability of groundwater. Thus, integrated measures must be taken to limit the adverse human activities around the dam catchment. While the contribution of Makutupora SA to DA was small (9.6%), it was found that 75% of groundwater in the Hombolo suburb originates from the Makutupora aquifer, supporting earlier hypotheses that the two systems are indeed, highly interconnected, and must, therefore, be co-managed. This is the first time; however, the actual interactions between the two systems have been quantified.

Generally, there is an increased enrichment of heavy SIs with increasing mineralisation. This suggests older waters, which have had longer interactions with rock matrices, are more enriched with SIs than fresher and younger waters. Two major groundwater types were dominant namely NaHCO₃, mainly in MWF and representing old water, and NaCl type mostly occurring in and around the city suburbs and in the flood plains from the city leading to the HD. Pollution of the groundwater system was also observed where higher Nitrate- of up to 223.7 ppm in Hombolo suburbs were found, similar trends were observed in other parts of the city. Generally, SGs were more polluted than DG, river water, and spring water in that order. This suggests that the current groundwater management regime may not be working well and requires to be revisited and improved. It is recommended the entire catchment, surface, and deep groundwater, be protected as the two systems are interconnected and certain pockets within the city catchment are already highly polluted.

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.