Geothermal gradients are in the range of 21.1°C–30.3°C/km (East of basin), 25.1°C–32.9°C/km (Center of basin), and 27.3°C–53.5°C/km (South of basin) (Figure 4). The geothermal gradient distribution is also highly variable locally across the basin with a strong positive anomaly (40°C–60°C/km) near the towns of Dobyn and Shonzhy (Figure 5). The geothermal gradients are moderately elevated (∼29.8°C/km) in the central part of the basin where wells are currently producing geothermal waters for balneology purposes at sanatorium Keremagash (well 1-TP), and greenhouse and fish-farming (wells 5539, 1-RT, 2-TP; Figures 2, 5). Data from wells 3-G and 6-G support slightly lower geothermal gradients in the eastern part of the basin is indicated by.

The observed variations in geothermal gradient (Figures 4, 5) are consistent with abrupt variations in crustal thickness and/or localized vertical flow of heated water along faults and fracture systems. Prior interpretations alternatively did not capture this high variability, and argued for linear alike depth-temperature gradients within a relatively uniform intracratonic sag basin (Figure 3). Potential evidence for this variability includes temperature anomalies along the Shonzhy fault trend between the towns of Shonzhy and Dobyn (Figure 5). The displacement geometry of the Shonzhy fault (Figure 3) is debatable. Most maps show it as a normal fault with the formation of an inferred graben on the South side (Figures 2, 3). However, Grützner et al. (2019) have classified the Shonzhy as an active south-dipping reverse fault characterized by about 20 m of recent vertical displacement, and an unknown amount of lateral displacement (Figure 9C in Grützner et al., 2019). The overall context of Grützner et al. (2019) thrust and strike-slip faults in the Ily Basin, argues the term “south dipping reverse fault”, and imply the fault is a part of a steeply dipping flower structured thrust fault system, where some faults are reactivated as reverse faults by compression. Yet another possibility is that the Shonzhy fault is a more discrete strike-slip fault oriented parallel to those documented in the mountains to the south and the north (Figure 2). Its location >20 km north of the North Ketmen Fault (Figure 2), and thick sedimentary cover south of it (Grützner et al., 2019), suggests possible subsidence caused by strike-slip transtentional forces, genetically related to transpression that contributed to flanking mountains, which would also result in an array of en echelon faults and together create a network of grabens, step overs, and/or relay ramps that formed pull-apart depressions (Faulds and Hinz, 2015).

The proximity of the temperature anomaly to the Shonzy fault (Figure 5) argues for a genetic relationship between the two. Potential scenarios include:

i) The normal-fault scenario inferred from Kalitov 2016 (Figure 3) implies extensional stresses generating a graben or similar depression from crustal stretching, which further implies local crustal thinning. This would force higher temperature gradients beneath the tectonically stretched block (graben). Similarly, a pull-apart depression caused by transtensional forces would cause similar crustal thinning and an analogous temperature anomaly.

Either of the above scenarios implies extensive networks of faulting and fracturing which would promote upward migration of deeper sourced hot water and its potential accumulation in highly permeable Upper Cretaceous strata in the Shonzhy-Dobyn area. This would imply that although temperature measurements are correct, the anomaly shown on the map (Figure 5) does not represent the actual gradient and is overestimated. Additional structural information coupled with geothermometry (sensu Ferguson et al., 2009) is needed to prove or rule out this scenario. Similar water chemistry of Upper Cretaceous aquifers across the basin argues against challenges to the mixing of their waters with deeper sourced brines, the evidence of thermal springs along the fault line (Figure 1) suggests that the Shonzhy fault(s) might be at least locally open to the surface. Importantly, the localized nature of Shonzhy–Dobyn anomaly suggests the potential presence of more, still undiscovered anomalies in the basin. Section 5.1 provides a detailed review and supporting evidence for intra-basinal strike-slip tectonic setting including the formation of a transtentional pull-apart basin and associated crustal thinning, faults, and fractures.

The data presented in this study challenge and improve prior interpretations for the origin of geothermal gradients and waters in the Zharkent basin because of added observations, some of which include:

i) Historically, the occurrence of geothermal waters in the Zharkent Basin are interpreted to be associated with the basin-scale increased geothermal gradients (Boguslavsky et al., 1999; Jóhannesson et al., 2019) and/or upflow along interconnected fault systems (Mukhamedzhanov, 1989; Kalitov, 2016). However, the source of heat and the plate-tectonic drivers of basin growth, faults, and fluid flow patterns remain speculative.

Below we provide and discuss alternative interpretations for the source of heat and water in the context of presented results, recent literature, and modern tectonic principles.

Recent interpretations of the tectonic history of south-eastern Kazakhstan suggest that the area is in a structural domain dominated by compression and right-lateral shear which is still active today (Kober et al., 2013; Grützner et al., 2019). These processes result in simultaneous mountain building South and North of the Yili basin (eastern extension of the Ily basin in China, Figures 1, 2), and the formation of basinal accommodation for sediments between mountain chains. The post- Late Cretaceous depositional tectonic history summarized below suggests that accommodation for sediments is by basin subsidence, most likely provided by transtentional forces (“Cenozoic phase of east–west extension”; Kober et al., 2013) and formation of a localized pull-apart basin.

Prior thermochronology and tectonostratigraphic studies indicate five stages of Late Cretaceous and Cenozoic tectonic activity in the Ily Basin area (Han et al., 2004; Burtman, 2012; Jia et al., 2020) that significantly shaped the basin. These serve as a foundation for additional interpretations and include:

i) Late Cretaceous-early Paleocene: rapid and localized uplift of the Narat and Borohoro ranges, i.e. the Chinese extensions of the Ketmen Ridge and Dzhungarian Alatau, respectively. Compressional movements reactivated preexisting basin-bounding, south-dipping thrust faults, and were driven by far-field deformation of the Central Asian lithosphere (Jolivet et al., 2010; Jia et al., 2020).

The North Tian Shan orogen trends toward pure-shear crustal shortening, which is absorbed by N-S thrust faults and N-W and N-E sinistral and dextral strike-slip faults (Wu et al., 2021). Moreover, the Zharkent Basin is simultaneously subject to overall N-S shortening and E-W extension (Grützner et al., 2019). Ongoing shear strain is expressed by numerous examples of surface expressions including dextral (Figure 1 in Grützner et al., 2019) and sinistral movements (Figures 5, 6, 9C in Grützner et al., 2019) as well as up to 500 m long, sinistral offset of streams observed on satellite images along the northern flanks of Ketmen Ridge (Figures 1, 2), between Nazgur Cavern (43°28′54.0″ N and 80°15′89.0″ E) and Sukam Village (43°29′12.10″ N and 80°10′48.52″ E) and elsewhere on the flanks of the basin. Intensive shearing observed in basin bounding ranges and ridges extends into neighboring intermontane basins in the North Tien Shan (e.g. E-W elongated depressions in Kyrgyzstan and China; Figure 1), that are defined by E-W-striking flower structures and associated thrust faults trending parallel to the mountain fronts (Kober et al., 2013; Cording et al., 2014; Grützner et al., 2019).

Simple-shear ‘bookshelf’ faulting expresses in the area between the Junggar Basin and the Kazakh Platform, within the area of geothermal interest. The clockwise rotation of the Tarim Basin relative to the stable Kazakh Plate caused a regional N-S shortening that decreases from west to east (Grützner et al., 2019). This is reflected by the left-lateral slip rate from 6.2 ± 1.8 mm/yr range in the Ili Basin to no more than 2.0 mm/yr in the Southern Dzhungarian Alatau fault (Yu et al., 2021).

We note that although thrust faults have locally formed minor (<1 km wide) micro-foreland basins (e.g. Charyn Canyon; Kober et al., 2013; their Figure 11), the geological maps (Figure 2), cross-sections (Figure 3; Kalitov, 2016), gravimetry survey (Gravimetric map of the USSR, 1971), and our field observations show a larger superimposed trend for all exposed sedimentary strata dipping towards the basin center. This implies that flower structure thrust faults illustrated by Jolivet et al. (2010) and Jia et al. (2020) are at most secondary elements in generating the accommodation for an up to 3 km thick Cenozoic sediment pile accumulated between opposing Dzhungarian Alatau and Ketmen ridge thrusts (Figures 1–3, 11). Additionally, the basin is progressively deepening towards its center, which contradicts the generation of a foreland basin by peripheral thrust loading, whereby a wedge-shaped basin geometry is expected with the thickest sediments near basin-marginal thrust loads (Figure 3). The intracratonic sag basin (Lucas et al., 1997) implies no genetic relationship with high basin bounding mountain (Figures 2, 3) which seems very unlikely. Thus, the co-occurrence of a major depocenter in the basin center between the bounding mountain belts (Figures 2, 3), but still far away from the influence of flanking flower structures, requires an alternative interpretation that favors common cause, but not cause and effect between these two structural trends.

The main strike-slip, thrust, and other undifferentiated faults overlain on a gravity map (Figure 9) show a correlation of trends. The NE-oriented gravity anomaly isolines align with the orientation of the strike-slip faults, implying that the faults play a role in the formation of the transtensional Zharkent Basin, which provided accommodation space for sediments. Crustal thinning can be inferred from large negative gravity anomalies (Figure 9), induced by the attenuation of cooler dense crust and the introduction of a hotter mantle near the surface that heated the plate and the sedimentary cover. Negative anomalies are consistent with the advent of transtensional structural trends observed and inferred within the basin. The major strike-slip faults and associated mega-flower structures are traced on the surface along the edges of Zharkent Basin (Figure 9; Grützner et al., 2019). However, most likely, there is a significant number of local strike-slip and associated en-echelon faults of Cenozoic age, which are covered by Quaternary deposits (Figure 2) and are typically difficult to localize in the subsurface (Muraoka et al., 2010).

We propose that the Zharkent Basin is regionally a transtensional pull-apart basin filled by up to 3 km of sediments (Figures 3, 9) This is consistent with substantial evidence of shallower basement and transpressional movements in Ily Basin bounding ranges that took place throughout the Cenozoic (Kober et al., 2013; Grützner et al., 2019) as additional components of this regional shearing trend. We further suggest that the proposed tectonic attenuation has significantly contributed to basin-scale crustal thinning, which in return has caused elevated geothermal gradients of up to 55.4°C/km (Figures 4, 5). The variations in temperature gradients (Figures 4, 5) suggests that crustal thinning is uneven across the basin. This phenomena is typical for pull-apart basins, such as the Salton Sea where higher temperatures were attributed to superimposed subregions of extensional to transtensional strain (Faulds and Hinz, 2015). These authors highlighted that grabens and step overs or relay ramps, typical of larger extensional and transtensional systems, are the most favorable intrabasinal areas for locally increased heat. Local heat anomalies like those observed in the Zharkent Basin are thus expected in regional transtensional systems.

Considering the sparsity of well data and the absence of seismic data, a constraint of discrete tectonic elements within the Zharkent Basin is challenging. However, the proposed pull-apart model is consistent with available structural, stratigraphic, and geophysical observations, and offers a conceptual model for assessing intrabasinal grabens and relay ramps as targets for elevated heat flow. The pull-apart interpretation offers a platform for revisiting interpretations of previously mapped faults (Figure 3). The evidence of Shonzhy-Dobyn temperature anomaly characterized by a geothermal gradient as high as 55°C/km (Figures 4, 5), and the argument that it is most likely one of perhaps many intrabasinal grabens (sensu Alexeiev et al., 2009); Section 4.1), suggests the potential for more similar thermal anomalies in the basin which have not been drilled yet. Finding such anomalies in areas where the Upper Cretaceous sandstone is buried deeper (in the Shonzhy-Dobyn area it is ∼50°C at a depth of only 600–700 m) and potentially reaching temperatures of >120°C would mean the discovery of a geothermal play suitable for direct electricity production (Lee, 2001; Dickson and Fanelli, 2003).

Prior workers argue that Upper Cretaceous aquifers in the Zharkent basin are charged by meteoric water sourced from the Ketmen Ridge and Dzhungarian Alatau Mountain (Figure 1) drainage system (Mukhamedzhanov, 1989; Kalitov, 2016) and partially via deep groundwater flow through fractured rocks (Kim et al., 1971). These processes are known as mountain-front recharge (MFR) and mountain-block recharge (MBR), respectively. These interpretations (Kim et al., 1971; Mukhamedzhanov, 1989; Kalitov, 2016) are based on basin configuration (e.g. enclosed basin) and analogous hydrogeological systems. We use new evidence from hydrogeochemistry and water isotopes to test these proposed recharge mechanisms.

The geothermal waters produced from the Upper Cretaceous aquifer in the Zharkent Basin are characterized by extremely low salinities ranging from about 0.3 to 0.8 g/L (Table 4) when compared to the more saline waters more typically observed at their production depths (Mcintosh, 2021) and standards for using it for agriculture and geothermal productions (Yapiyev et al., 2021). Results from a single shallow geothermal well (AE-4) indicates slightly brackish water (∼4.5 g/L), perhaps because of the localized influence of saline surface processes on water chemistry. The very low TDS values and the presence of Si, HCO_3, and Na ions (Table 4) indicate that rock-water interactions are dominated by the dissolution of silicate-rich rocks such as rhyolites and feldspars which are abundant in Paleozoic rocks and derived Mesozoic fragments in the Zharkent Basin (Figures 2, 3). The source of Na and Si is commonly attributed to sodium feldspar and/or rhyolite hydrolysis by hydrogen ions donated by bicarbonate dissociation (Sharp and Tranter, 2017; Absametov et al., 2023). Such silicate dissolution is also documented elsewhere in hydrothermal systems (Shoedarto et al., 2022; Xing et al., 2022). Shoedarto et al. (2022) showed silicate weathering dominates water-rock interaction in hot springs in Southern Bandung, West Java, Indonesia. Chemical dissolution of silicate rocks is positively correlated with temperature (Gaillardet et al., 1999) and is usually low in cold climate regions (Anderson, 2007) compared to higher dissolution in the deep subsurface (Kharaka and Hanor, 2003). The high Si content, high temperature, low salinity, and predominantly quartz lithology of the Upper Cretaceous aquifer (Umralin, 2022; unpublished BSc report) suggests that the solutes in water are likely derived from silicate dissolution in deeper strata. While mixing diagrams (Figure 7) point to the potential presence of evaporates, their actual quantity is probably minimal as evidenced by low TDS (Table 4) and two to three times lower concentrations of Cl compared to Na (Table 4).

The progressive depletion in heavier isotopes with depth (Figure 8) suggests that the primary source of water in the Upper Cretaceous aquifers is the cryosphere (snow, glaciers, and permafrost) of the Tien Shan Mountains (Hoelzle et al., 2019) with snowmelt by far being the largest contributor (Kraaijenbrink et al., 2021). Snowmelt is the most isotopically depleted known water source (Beria et al., 2018), which suggests that winter precipitation in the mountains is the primary source for studied Upper Cretaceous geothermal waters (Figure 8; Table 4). The median values of δ¹⁸O in winter precipitation (December-January-February) in Central Asia are reported to be around -15 to -21‰ (Saidaliyeva et al., 2021), similar to that measured here in geothermal wells of around -14‰ (Figure 8). Summer (June-July-August) precipitation values, by contrast, are isotopically much heavier, with median values of around—5‰ (Saidaliyeva et al., 2021). The data from the global dataset of stable water isotopes in precipitation (Bowen et al., 2005) shows a higher contribution of cold-season precipitation to geothermal groundwater recharge (Figure 8). Recently, Hellwig et al. (2018) reported δ¹⁸O values of water from springs collected in eastern Ily (Altyn-Emel location) to be in the range of -12.8 to -13.6‰ which is very close to our measurements (e.g. -13.04‰ for AE-3; Figure 8). These isotope values contrast with much more enriched values more typical of rainfall signatures (-8.6 to –3.09‰, Hellwig et al., 2018).

The deviation of the isotopic composition of water in geothermal wells right of the LMWL (Figure 8), also known as the ‘oxygen isotope shift’ (Kharaka and Hanor, 2003) is commonly observed in geothermal systems (Kong et al., 2020; Mcintosh, 2021). This enrichment of δ¹⁸O is commonly caused by high-temperature exchange of formation-water with silicate minerals. This interpretation is also supported by our water chemistry data presented herein (Table 4; Section 5.2.1).

Water sample KD-9 (Figure 8), collected from a stream draining from the nival zone in Ketmen Ridge (Figure 2), is more isotopically enriched relative to subsurface samples (the springs, the artesian well shallow GW, deep and shallower geothermal wells (Figure 8; Table 3). This further demonstrates the larger influence of meltwater recharge with deeper aquifers via deeper flow paths from the mountain block. Recently (Mcintosh, 2021), used stable water isotopes to show that maximum meteoric water circulation depths vary from 1 to 5 km in North America. The deeper circulation is expected to occur in areas of greater topographic relief, like that of the Zharkent area. Jasechko et al. (2017) suggested that globally water at wells deeper than 250 m is mostly (42%–85%) ‘fossil’ groundwater, recharged by precipitation more than ∼12 000 years ago. Though, the previous preliminary results of groundwater residence time in the Zharkent basin, based on the absence of tritium detection (Kozhagulova et al., 2022) should be treated as a potential over-interpretation as the samples were analyzed without prior enrichment. A very low salinity (Figure 10) and isotopically depleted signal in deep geothermal wells (Figure 8), cumulatively points to long-duration deep groundwater flow paths from the mountain cryosphere to the Cretaceous aquifer.

Despite having relatively low variability in salinity across the basin, natural flowrates vary up to 3 times, even for the wells that are drilled in close vicinity to each other (e.g. wells 1-RT and 5539 producing 24.5 and 54.5 L/s respectively (Mukhamedzhanov, 1989). This clearly suggests variable reservoir properties controlled by depositional heterogeneity, which are a subject of ongoing research and beyond the scope of this study.

Cenozoic aquifer lithologies are characterized by poor permeability (Figure 10) and low flow rates (Kalitov, 2016), and are not used for geothermal production. However, the integration of their water chemistry (Chakabayev, 1960 and this study) promotes a better understanding of basin recharge.

Vertical salinity variation of Cenozoic formation waters is significant (Figure 10). Water samples taken from Pliocene aquifers have low ion content (around 1 g/L), and show an abrupt rise in Miocene aquifers (with TDS values up to 19 g/L), implying that underlying sedimentary units would be characterized by even higher salinity. However, this is not the case as water gets less saline in underlying Paleogene aquifers, and becomes virtually fresh in deeper buried Upper Cretaceous aquifers (Table 4; Figure 10). These abrupt salinity shifts point to the presence of aquitards between Miocene and Paleogene, and Paleogene and Cretaceous aquifers, respectively (Figure 10). The chemical profile (Figure 10) suggests that dissolved salts (from secondary evaporates such as halite and gypsum) are mostly likely percolating from the modern-day surface (i.e. Ily River floodplains and evaporate ponds) to accumulate at the bottom of the Miocene (depth=2068 m, Na-SO₄ water type, TDS ∼19 g/L; Figure 10). In contrast, the underlying less saline Paleogene formation waters (depths at 2500–2700 m, Na-SO₄ water type, TDS ∼5.6 g/L; Figure 10), and essentially fresh waters in the Cretaceous layers (Na-HCO₃ water type, TDS ∼0.9 g/L) implies that both the Paleogene and Cretaceous aquifers are recharged primarily by lateral groundwater flow (Figure 11), and are not in strong connectivity with the overlying more saline layers. Higher salinities in the Paleogene than in the Upper Cretaceous strata suggest a mixing of laterally charged fresh waters and saline waters that over geological time scale have percolated or diffused through the proposed aquitard 1 (Figure 10). The water chemistry of well 1G based on Chakabayev (1960) is in line with recent data (Kalitov, 2016; Table 1) as well as TDS of the recently acquired water samples presented in this study (Table 4; Figure 10). The similarity of Upper Cretaceous waters obtained from well 1G and more recent deep wells including 5539 (PI-1), 1-RT (PI-2), 2-TP (PI-3), 1-TP (PI-4 and PI-5) implies that our interpretation of historical water chemistry data presented above and proposed recharge mechanisms are likely valid across the basin.

In summary, Miocene and Pliocene strata are recharged from percolation down from the overlying basin floor, and Cretaceous waters are recharged laterally from winter precipitation in the Tien Shan Mountains. Paleogene strata are likely a mixture of the two.