Geothermal reservoir characteristics, genesis, and potential in the Sanjiaochenggou area, Jinchang Depression

1 Introduction

With the continued growth in global energy demand and the ongoing advancement of carbon peaking and carbon neutrality targets, the development of clean, low-carbon, and sustainably exploitable renewable energy has become a major priority in the global energy transition (Dickson and Fanelli, 2013; Teng et al., 2025). Geothermal energy, a renewable energy source derived from the Earth’s internal heat, offers several advantages, including abundant reserves, stable and continuous supply, high utilization efficiency, and limited susceptibility to climatic and seasonal variations (Barbier, 2002; Li et al., 2025). It therefore has considerable potential for applications such as district heating, domestic hot-water supply, greenhouse agriculture, low-temperature industrial heating, and geothermal health and wellness services (Lund and Toth, 2021; Liu et al., 2025). With continuous advances in drilling technologies, geophysical exploration, and resource assessment methods, global geothermal development has gradually expanded from conventional high-temperature geothermal power generation toward the direct use and integrated development of medium- and low-temperature geothermal resources (Goldstein et al., 2011; Rybach, 2003). Geothermal space heating, agricultural applications, and low-temperature industrial heat utilization have consequently become major areas of growth in the geothermal sector. This trend is particularly important in the arid and semi-arid regions of northern and northwestern China, where prolonged winter heating periods result in concentrated energy consumption (Hou et al., 2018). Conventional coal-fired heating systems not only consume substantial amounts of energy but also generate considerable carbon emissions and environmental pollution. Driven by China’s carbon peaking and carbon neutrality strategy and its clean-heating policies for northern regions, the effective development and utilization of medium- and low-temperature geothermal resources are of great practical significance for optimizing regional energy structures, promoting energy conservation and emission reduction, and facilitating green and low-carbon development.

Compared with volcanic–magmatic high-temperature geothermal systems, medium- and low-temperature geothermal resources generally occur at lower temperatures and are therefore less suitable for high-efficiency electricity generation. Nevertheless, owing to their widespread distribution, relatively low development risks, and high efficiency in direct utilization, these resources are particularly suitable for building heating, domestic hot-water supply, geothermal health and wellness services, greenhouse agriculture, aquaculture, and low-temperature industrial heating (Hamor et al., 2023). In recent years, a number of sedimentary-basin-hosted medium- and low-temperature geothermal resources have been discovered and developed in the North China Plain, the Guanzhong Basin, the Hexi Corridor, and the Songliao Basin, providing an important energy supply for regional heating and integrated geothermal utilization (Zhu et al., 2015; Zhang et al., 2024). As an important category of medium- and low-temperature geothermal resources, sedimentary-basin geothermal systems generally develop within relatively tectonically stable sedimentary basins (Yuan et al., 2026; Li et al., 2022). Their geothermal reservoirs are primarily hosted in porous–fractured clastic rocks, including sandstone, sandy conglomerate, and siltstone (Astolfi et al., 2014). These systems are typically characterized by deep burial, concealed distribution, stratiform reservoir development, relatively good reservoir continuity, and pronounced confined conditions (Fan et al., 2020). In the absence of heat supplied by shallow magmatic activity, their thermal energy is derived mainly from heat conduction from the deep crust and heat production through the radioactive decay of crustal elements. Consequently, the formation and evolution of these geothermal systems are jointly controlled by the regional heat-flow regime, lithospheric structure, sedimentary-cover thickness, and deep groundwater circulation (Huang et al., 2025; Kolawole and Evenick, 2023). With the continuous accumulation of global heat-flow data, borehole temperature measurements, and studies on the thermal structures of sedimentary basins, the understanding of sedimentary-basin geothermal systems has gradually shifted from a primary emphasis on geothermal gradients toward an integrated analysis of heat sources, geothermal reservoirs, groundwater circulation, and basin evolution. Based on approximately 152,000 geothermal-gradient measurements worldwide, Kolawole and Evenick established a global database of geothermal gradients in sedimentary basins (Kolawole and Evenick, 2023). Their results demonstrated that the thermal structures of sedimentary basins exhibit pronounced regional variations and that geothermal gradients in different basin types are controlled not only by crustal heat flow but also by lithospheric thickness, sedimentary-cover thickness, basin type, and tectonic evolution. In particular, thick sedimentary successions can produce a pronounced thermal-blanketing effect, which reduces heat dissipation through shallow strata and increases temperatures within the basin, thereby creating favorable conditions for the formation of medium- and low-temperature geothermal resources (Kolawole and Evenick, 2023). Meanwhile, Qiu et al. (Qiu et al., 2021) systematically reviewed the thermal structures of representative sedimentary basins in upper Yangtze and north China and indicated that differences in tectonic setting, sedimentary filling history, and reservoir properties have resulted in distinct heat-transfer regimes among different basins. For example, the Songliao Basin is primarily characterized by a low-permeability, relatively homogeneous conductive heat-transfer regime, whereas the North China Basin exhibits a coupled conductive–groundwater-convective thermal structure. In areas containing well-developed faults, geothermal systems are commonly controlled jointly by fault-related heat transport and groundwater circulation. These findings indicate that sedimentary-basin geothermal systems are essentially the products of interactions among the regional heat-flow regime, sedimentation, groundwater circulation, and reservoir properties (Huang et al., 2025). Their formation mechanisms are therefore characterized by pronounced multivariable controls, making it increasingly difficult to meet the requirements of modern geothermal exploration and resource-development assessment by relying on a single type of geological dataset.

With the continued expansion of sedimentary-basin geothermal resource development, research in this field has evolved from the early identification of geothermal resources based solely on geothermal anomalies or heat-flow values into a comprehensive multidisciplinary framework integrating geology, hydrogeology, geophysics, drilling, well logging, hydrochemistry, isotope geochemistry, and numerical modeling (Morgan, 2015). In recent years, it has been widely recognized that sedimentary-basin geothermal systems are characterized by highly concealed occurrence, deeply buried reservoirs, and complex controlling factors. Consequently, the subsurface reservoir structure and resource potential cannot be accurately characterized using a single exploration method. The establishment of an integrated interpretation framework based on multisource information has therefore become an important direction in the exploration and assessment of sedimentary-basin geothermal resources (Teng et al., 2025; Barbier, 2002; Li et al., 2025). Regional geological and hydrogeological surveys remain fundamental components of geothermal exploration. Their primary objectives are to determine the tectonic evolution of the basin, sedimentary systems, stratigraphic distribution, and groundwater recharge–runoff–discharge conditions, thereby providing the geological context required to interpret the formation mechanisms of geothermal systems (Allen and Allen, 2013). On this basis, geophysical techniques, including wide-field electromagnetic methods (WFEM), controlled-source audio-frequency magnetotellurics (CSAMT), magnetotellurics (MT), and integrated gravity–magnetic interpretation, have been extensively applied to the identification of concealed geothermal reservoirs and the investigation of deep geological structures (Spichak and Manzella, 2009; Huenges, 2010). By exploiting contrasts in the electrical resistivity of subsurface materials, these methods can be used to delineate fault structures, low-resistivity water-bearing formations, and the spatial distribution of geothermal reservoirs. When jointly interpreted and cross-validated with drilling data, they can substantially improve the accuracy of deep geothermal reservoir prediction (Wang et al., 2025).

Although substantial progress has been made in recent years in understanding the thermal structures, groundwater circulation, hydrochemical evolution, and resource assessment of medium- and low-temperature geothermal systems in sedimentary basins, the formation mechanisms and resource potential of geothermal reservoirs in tectonically stable inland sedimentary basins with limited exploration remain poorly understood (Muffler and Cataldi, 1977). On the one hand, existing studies have largely focused on large and well-developed geothermal fields, whereas medium- and low-temperature geothermal systems in the sedimentary basins of the arid regions of northwestern China have received comparatively limited attention (Huang et al., 2025; Qiu et al., 2021). In particular, systematic investigations into the coupled relationships among geothermal reservoir structure, hydrodynamic conditions, hydrochemical evolution, and resource assessment remain insufficient (Pandey, 2018). On the other hand, during the early stages of geothermal exploration, the limited number of boreholes commonly results in considerable uncertainty in key parameters, including reservoir boundaries, effective thickness, porosity, and groundwater recharge conditions. Consequently, the assessment of geothermal resources and development potential based on a single well remains highly challenging (Steeples, 1984; Hajto and Anna, 2018). Moreover, with the continued expansion of geothermal resource development, issues such as reinjection efficiency, thermal breakthrough, corrosion and scaling, and geothermal wastewater disposal have increasingly become major constraints on the sustainable exploitation of geothermal resources (Lin et al., 2013). Resource-volume estimation alone is therefore no longer sufficient to meet the requirements of geothermal development and utilization. Instead, geothermal reservoir architecture, hydrodynamic characteristics, fluid geochemical evolution, and development-related constraints should be evaluated within an integrated framework to establish a more comprehensive conceptual model of the geothermal system and a more robust resource assessment methodology (Li et al., 2025; Moeck and Beardsmore, 2026).

The Sanjiaochenggou area of the Jinchang Depression is located within the Huangnan sub-sag of the Jinchang Depression, Chaoshui Basin, which forms part of the northern basin system of the Hexi Corridor. It represents a typical inland clastic sedimentary basin. Geothermal geological investigations in this area have long remained relatively limited, with previous studies focusing primarily on regional geological surveys and petroleum exploration. Although temperature measurements from a small number of deep wells have indicated the presence of geothermal anomalies at depth, several key issues, including the stratigraphic position and spatial architecture of the geothermal reservoirs, their water-bearing capacity, the origin of the geothermal fluids, and the geothermal resource potential, have not yet been systematically investigated. The JCDR-1 combined exploration–production well provides the first direct deep-borehole constraint on the stratigraphic architecture, temperature distribution, hydraulic response, and geothermal-fluid characteristics of the Sanjiaochenggou area, thereby allowing the previous geophysical predictions and regional hydrogeological interpretations to be tested at the borehole scale.

Accordingly, this study addresses three research questions. First, to what extent can the reservoir and stratigraphic interfaces inferred from WFEM and CSAMT be validated and refined using deep drilling and well-logging data? Second, what combination of reservoir-caprock structure, temperature distribution, hydraulic response, hydrochemistry, and isotopic evidence explains the formation and evolution of the geothermal system? Third, how do single-well control, geophysically inferred reservoir extent, and analytical assumptions affect the estimated geothermal resources and preliminary production scenarios? To address these questions, hydrogeological, geophysical, drilling, well-logging, temperature, pumping-test, hydrochemical, and isotope data were jointly evaluated. The principal advances are the quantitative comparison of pre-drilling geophysical predictions with the JCDR-1 drilling results, the establishment of a borehole-constrained conceptual model for a predominantly conduction-heated confined clastic geothermal system, and the separation of controlled and inferred resource areas within deterministic uncertainty scenarios. These results provide site-specific understanding of a poorly explored geothermal system in the eastern Hexi Corridor and a transferable framework for assessing comparable inland sedimentary-basin geothermal systems where deep-borehole control remains limited.

2 Engineering background

2.1 Study area

As shown in Figure 1, the Sanjiaochenggou geothermal field is located in the northeastern part of Jinchuan District, Jinchang City, Gansu Province, China. Geographically, it lies within the Jinchang Depression along the southern margin of the Chaoshui Basin in the eastern Hexi Corridor. The study area extends from Xin’an Village–Xiyouziwa in the west to Xujiagou Village in the east, and from Xiaojingzi in the north to Sanjiaochenggou in the south, covering an area of approximately 47.02 km². It is bounded by longitudes 102°14′06″–102°18′38″ E and latitudes 38°34′13″–38°39′23″ N. The terrain is generally open and gently slopes from southwest to northeast, with ground elevations ranging from 1,390 to 1,440 m above sea level. Geomorphologically, the area represents a typical piedmont alluvial–proluvial plain in the arid region of the Hexi Corridor.

The key engineering information and basic geothermal reservoir parameters of the geothermal system in the study area are summarized in Table 1. The representative geothermal well, JCDR-1, was drilled to a final depth of 2,003.5 m and penetrated multiple sedimentary sequences of Quaternary, Neogene, Cretaceous, and Jurassic age. The measured bottom-hole temperature was 61.7 °C, while the wellhead water temperature and daily production rate were 47 °C and 2,016.85 m³/day, respectively. These results indicate that a medium- and low-temperature geothermal resource of appreciable scale occurs in the Sanjiaochenggou area. The geothermal water is mainly hosted in sandstone, sandy conglomerate, and sandy mudstone units extending from the lower Neogene to the Cretaceous strata, whereas the overlying mudstone and fine-grained argillaceous deposits form a relatively stable caprock. This stratigraphic assemblage gives the study area the fundamental characteristics of a stratiform porous geothermal reservoir hosted in a sedimentary basin.

TABLE 1. Key engineering and geothermal reservoir parameters of the Sanjiaochenggou geothermal field.

2.2 Geological and hydrogeological conditions

As shown in Figure 2, the study area is located within the Jinchang Depression of the Chaoshui Basin. The ground surface is predominantly covered by unconsolidated Quaternary sediments, whereas clastic sedimentary sequences of Neogene, Cretaceous, and Jurassic age occur at greater depths. The regional stratigraphy is dominated by sandstone, sandy conglomerate, mudstone, and sandy mudstone, exhibiting a typical basin-fill succession characterized by alternating coarse-grained clastic and argillaceous layers. The study area has a typical arid continental climate, characterized by limited precipitation and intense evaporation. The long-term mean annual precipitation and evaporation are approximately 129.4 and 1,635.1 mm, respectively, indicating that evaporation greatly exceeds precipitation. Surface runoff is poorly developed. The Jinchuan River constitutes the principal regional surface-water system; however, its downstream flow is discontinuous during most periods because of reservoir regulation and water withdrawal for irrigation. Groundwater in the study area mainly occurs as pore water within unconsolidated Quaternary deposits and as pore–fracture water within deep clastic rocks, with the latter being more closely associated with the geothermal resources.

Overall, the Sanjiaochenggou geothermal field exhibits the fundamental geological setting of a medium- and low-temperature geothermal system hosted in a sedimentary basin. Thick clastic sedimentary sequences provide potential reservoir space, fine-grained argillaceous deposits favor the formation of relatively impermeable confining layers, and conductive heat from depth supplies the thermal energy required to heat the geothermal fluids. These geological and hydrogeological conditions provide the basis for subsequent geothermal reservoir identification, drilling verification, and geothermal resource assessment.

3 Field tests and methods

In this study, an integrated methodology combining hydrogeological investigation, geophysical exploration, geothermal drilling, comprehensive well logging, pumping tests, and hydrochemical analyses was employed to characterize the geothermal reservoir architecture, temperature field, hydrodynamic conditions, and geochemical properties of the geothermal fluids in the Sanjiaochenggou geothermal field.

As shown in Figure 3, surface hydrogeological surveys, wide-field electromagnetic surveys, and controlled-source audio-frequency magnetotelluric sounding were first conducted to delineate the sedimentary architecture of the basin, the distribution of caprock units, and potential geothermal reservoir targets. Based on these results, the JCDR-1 geothermal well was deployed, and drilling, comprehensive well logging, and pumping tests were subsequently performed to verify and refine the geophysical interpretations. Finally, analyses of major ions, trace constituents, stable isotopes, and water-soluble gases were integrated to determine the hydrochemical characteristics of the geothermal fluids and constrain their origin and evolution, thereby providing a scientific basis for geothermal resource development and assessment.

3.1 Hydrogeological investigation and geophysical exploration

The hydrogeological investigation covered the entire study area and focused on documenting stratigraphic lithology, geomorphological units, structural features, groundwater discharge points, production and domestic wells, groundwater-table depths, water temperatures, spring discharges, and water-quality characteristics. A total of 101 survey sites were identified, including 36 production and domestic wells, one spring, and 64 geological and geomorphological observation sites. The survey routes extended for approximately 85.5 km and covered the major geomorphological units, stratigraphic boundaries, and potential hydrogeological anomalies within the study area. On the basis of these investigations, the spatial relationships among shallow groundwater, surface-water systems, and deep geothermal water were preliminarily established, providing fundamental constraints for the subsequent layout of geophysical survey profiles and selection of the drilling location.

Geophysical exploration was conducted using wide-field electromagnetic sounding (WFEM) and controlled-source audio-frequency magnetotelluric sounding (CSAMT). The WFEM survey used a grounded horizontal electric-dipole source driven by a high-power pseudo-random multifrequency transmitter and broadband electric-field receivers. The frequency range was 0.015625–8,192 Hz, and the horizontal electric-field component along the survey profiles was measured. Two WFEM profiles, with a combined length of 6 km, were established using 32 measurement stations at 200 m intervals. The transmitting-electrode separation was 1.1 km, and the transmitter–receiver separation was approximately 12 km. The CSAMT survey used a grounded horizontal electric-dipole transmitter and broadband electromagnetic receivers to measure the inline electric-field component and transverse magnetic-field component over a frequency range of 1–7,680 Hz. Three CSAMT profiles, with a total length of 20.4 km, were surveyed using 109 stations at intervals of 200 m. Cagniard apparent resistivity and impedance phase were calculated from the measured electromagnetic-field components.

Data quality was evaluated using repeated measurements. Two repeat stations were used for the WFEM survey, corresponding to 6.7% of the production stations. The mean relative errors of the two repeated measurements were 2.56% and 2.16%, respectively, and the overall repeatability error was 2.37%. Five repeat stations were used for the CSAMT survey, corresponding to 4.8% of the production stations. The survey-wide root-mean-square repeatability errors were 4.01% for apparent resistivity and 3.20% for impedance phase. During preprocessing, frequency points with low signal-to-noise ratios or discontinuous responses were removed, anomalous spikes and power-line interference were rejected, repeated signal cycles were stacked, and the accepted frequency-domain responses were robustly averaged.

The WFEM and CSAMT datasets were inverted profile by profile using two-dimensional smoothness-constrained inversion. The WFEM apparent-resistivity data were inverted using a damped Gauss–Newton algorithm, whereas the CSAMT apparent-resistivity and impedance-phase data were jointly inverted using a nonlinear conjugate-gradient algorithm (Routh and Oldenburg, 1999; Rodi and Mackie, 2001). A homogeneous half-space with a resistivity equal to the median apparent resistivity of each profile was used as the initial and reference model. An error floor of 5% was assigned to the WFEM apparent-resistivity data, while error floors of 5% for apparent resistivity and 2° for phase were adopted for the CSAMT data. The objective function combined the normalized data misfit with first-order horizontal and vertical model-smoothness constraints. The regularization trade-off was selected using an L-curve criterion and progressively reduced during inversion. Station elevations were incorporated into the finite-element mesh to account for topography. Inversion was terminated when the normalized RMS misfit decreased to ≤2.0 or when its relative reduction was <1% for two consecutive iterations, with a maximum of 20 iterations. To evaluate equivalence and non-uniqueness, the inversions were repeated using initial-model resistivities of 0.5, 1.0, and 2.0 times the profile median and regularization weights of 0.5, 1.0, and 2.0 times the reference value. Only electrical boundaries that remained stable under these tests and were supported by overlapping profiles or borehole stratigraphy were retained in the geological interpretation.

3.2 Geothermal drilling and testing

The JCDR-1 well is located in the north-central part of the study area and constitutes the key borehole used to characterize the deep geological structure of the Sanjiaochenggou geothermal field. The well was constructed as a vertical borehole using a three-section drilling design. The first section extended from 0 to 51.69 m and had a borehole diameter of 444.5 mm; the second section extended from 51.69 to 401.52 m and had a diameter of 346.1 mm; and the third section extended from 401.52 to 2,003.50 m and had a diameter of 241.3 mm. The third section mainly penetrated Neogene, Cretaceous, and Jurassic strata and therefore represented the principal interval for identifying the caprock and geothermal reservoirs. Drill cuttings were logged throughout the entire borehole, and core samples were collected below a depth of 400 m. Drill cuttings were generally sampled at 5 m intervals, with denser sampling conducted in key intervals below the top of the Cretaceous strata. Core samples were primarily used to determine porosity, permeability, specific heat capacity, thermal conductivity, density, and grain-size composition, as well as to identify potential geothermal reservoir intervals.

Comprehensive well logging included eight types of measurements: borehole temperature, apparent resistivity, natural gamma, acoustic logging, caliper logging, spontaneous potential, deep and shallow lateral resistivity, and borehole deviation. Borehole temperature measurements were conducted 24 and 48 h after the cessation of mud circulation to approximate static formation-temperature conditions as closely as possible. The geothermal gradient was calculated based on the temperature difference between the borehole entrance and the maximum logging depth (Equation 1):where G is the geothermal gradient (°C/100 m), T d is the temperature at the depth of logging, and T 0 is the temperature at the borehole entrance, d is the depth of logging. The average geothermal gradient was obtained from the two temperature-logging measurements. Lithological interpretation was based primarily on the natural gamma log and was supplemented by resistivity, spontaneous-potential, and acoustic transit-time data. Aquifer intervals were identified mainly on the basis of low resistivity, relatively low natural gamma values, variations in borehole temperature, and characteristic lithological assemblages. The integrated interpretation of drilling and well-logging data enabled the information obtained from a single well to be translated into a comprehensive understanding of geothermal reservoir stratigraphy, caprock thickness, and the structure of the subsurface temperature field.

G = T d − T 0 d × 100 (1)

3.3 Pumping test

To evaluate the production capacity of the JCDR-1 well and the hydrodynamic response of the geothermal reservoir, a series of well-development procedures were conducted after well completion, including clean-water flushing, chemical cleaning, high-pressure jetting, water-air mixed flushing, air-compressor-assisted development, and pump-surging development. After the well water became clear and free of sediment, the well was allowed to recover to a static water level of 49.63 m below ground surface.

Three constant-rate pumping stages were conducted during separate test periods. Stage S1 was conducted from 16:00 on 13 November to 16:00 on 19 November 2023, Stage S2 from 15:30 on 18 December to 15:30 on 23 December 2023, and Stage S3 from 10:00 on 11 January to 08:00 on 15 January 2024. The corresponding pumping durations were 144, 120, and 94 h, respectively. Because the stages were conducted separately on different dates, they were treated as three independent constant-rate pumping stages rather than as a standard continuous step-drawdown test.

During each stage, discharge was measured using a triangular weir, the dynamic water level was monitored using an electronic water-level meter, and the wellhead water temperature was recorded simultaneously using a thermometer. Measurements were taken at 1, 2, 3, 4, 6, 8, 10, 15, 20, 25, 30, 35, 40, 50, 60, 80, 100, and 120 min after pumping commenced and at 30-min intervals thereafter.

Hydraulic parameters were estimated using the Dupuit steady-flow method, the water-level recovery method, and graphical matching to the Theis confined-aquifer solution. For the Dupuit method, hydraulic conductivity was calculated as Equation 2:

K = Q 2 π s w M ln R r w (2)

The radius of influence was estimated as Equation 3:

R = 10 s w K (3)

Transmissivity was obtained from Equation 4:where K is hydraulic conductivity (m/d), T is transmissivity (m²/d), Q is the stabilized discharge (m³/d), s w is drawdown (m), M is the cumulative effective aquifer thickness (m), r w is the pumping-well radius (m), and R is the calculated radius of influence (m).

T = K M (4)

For the water-level recovery method, transmissivity was calculated as Equation 5:where s 1 ′ and s 2 ′ are residual drawdowns at recovery times t 1 ′ and t 2 ′, respectively. The calculation used the 10–100 min recovery interval following Stage S1.

T = 2.3 Q 4 π s 1 ′ − s 2 ′ log 10 t 2 ′ t 1 ′ (5)

For the Theis analysis, the recorded time-drawdown data were graphically matched to (Equations 6, 7): where s is drawdown (m), S is the dimensionless elastic storage coefficient, t is pumping time, r is radial distance, and W u is the Theis well function.

s = Q 4 π T W u (6)

u = r 2 S 4 T t (7)

A cumulative effective aquifer thickness of M = 428.05 m and a pumping-well radius of r w = 0.0889 m were used. The analyses assumed an equivalent homogeneous and isotropic confined aquifer, an equivalent fully penetrating well, and approximately constant discharge during each pumping stage.

3.4 Hydrochemical analysis

Hydrochemical analyses were conducted to identify the hydrochemical facies of the geothermal fluids and investigate water–rock interactions, recharge sources, and groundwater circulation depths. Samples of shallow groundwater, spring water, and geothermal water from the JCDR-1 well were collected, with the geothermal water samples from JCDR-1 designated SY-009, SY-010, and SY-011. The analytical program included major ions, total dissolved solids, total hardness, pH, trace elements, specific constituents, stable isotopes, radioactive isotopes, and dissolved gases.

The major hydrochemical parameters included Na⁺, K⁺, Ca²⁺, Mg²⁺, Cl⁻, SO₄ ^2-, HCO₃ ⁻, CO₃ ^2-, F⁻, total dissolved solids (TDS), total hardness, and pH. The specific constituents analyzed included strontium, metasilicic acid, metaboric acid, bromide, iodide, and lithium. Isotopic analyses included δ²H, δ¹⁸O, δ³⁴S, ³H, and ¹⁴C. Among these, δ²H and δ¹⁸O were used to determine the recharge source of the geothermal water and assess the influence of evaporation; δ³⁴S was used to identify the sources of sulfate and constrain the dissolution of sulfur-bearing minerals; and ³H and ¹⁴C were used to constrain the residence characteristics and renewal conditions of the geothermal fluids. A radiocarbon-derived age estimate was calculated using the relationship (Equation 8):where t is the radiocarbon-derived age estimate, T 1 / 2 is the ¹⁴C half-life, taken as 5,730 years, A 0 is the assumed initial ¹⁴C specific activity of modern carbon, taken as 813.6 Bq / g, and A is the measured ¹⁴C specific activity of the groundwater sample. No correction was applied for carbonate dissolution, dead-carbon contribution, or water-rock interaction. Therefore, the calculated value was not interpreted as an absolute groundwater residence time, but it can be used to support the inference of a relatively old geothermal-water component and renewal. The dissolved-gas analyses included H₂S, CO₂, O₂, N₂, CO, CH₄, Ar, and He, providing supplementary constraints on the deep fluid environment and the origins of the dissolved gases.

t = T 1 2 ln ⁡ 2 ln A 0 A (8)

Major-ion concentrations were converted from mg/L to milliequivalents per liter before ionic-ratio calculations. Hydrochemical evolution was evaluated using the Na⁺/Cl⁻ and Ca²⁺/Mg²⁺ equivalent ratios, the proportion of Cl⁻ in the total major-anion charge, and the excess of divalent cations, expressed as Ca 2 + + Mg 2 + − HCO 3 + SO 4 2. Cation exchange was additionally evaluated using the chloro-alkaline index (Equation 9):where all ionic concentrations are expressed in milliequivalents per liter. Positive CAI-I values indicate the possible removal of Na⁺ and K⁺ from the water and release of Ca²⁺ and Mg²⁺ from the aquifer matrix through reverse cation exchange, whereas values near zero indicate limited exchange or predominance of a Na–Cl saline end member.

CAI − I = Cl − − Na + + K + Cl − (9)

The approximate recharge elevation was evaluated using the regional stable-isotope altitude effect (Equation 10):where H is the estimated recharge elevation (m), h is the elevation of the JCDR-1 sampling point (1,385.63 m), δ G is the δ¹⁸O value of the geothermal water (−10.8‰), δ P is the representative δ¹⁸O value of low-elevation precipitation in the Hexi Corridor and adjacent arid region (−10.0‰), and K is the regional δ¹⁸O altitude gradient. An annual mean δ¹⁸O altitude gradient of −0.26‰/100 m reported for precipitation on the northern slope of the Qilian Mountains was adopted (Ma et al., 2018). Because precipitation isotopes vary spatially and temporally, the calculated value was used as an approximate elevation constraint rather than as an exact recharge altitude.

H = h + δ G − δ P K × 100 (10)

4 Field test results

4.1 Distribution of potential geothermal reservoirs

The subsurface structure of the study area was jointly constrained by controlled-source audio-frequency magnetotelluric (CSAMT) and wide-field electromagnetic data. As shown in Figure 4, the CSAMT profiles were arranged along northwest- and northeast-trending directions across the study area, covering the Xiasifen area where the JCDR-1 well is located and the surrounding region. This survey configuration provided effective constraints on the lateral continuity of the sedimentary sequences and the spatial variations in local depressions.

All WFEM and CSAMT profile inversions converged within 12–18 iterations and satisfied the normalized RMS acceptance criterion of ≤2.0. The alternative starting-model and regularization tests consistently recovered the principal shallow high-resistivity, intermediate low-resistivity, and deeper medium-to high-resistivity layers. Model variations increased near the profile ends and at the maximum investigation depth. Therefore, the controlled reservoir boundary was delineated using electrical interfaces that remained stable under the sensitivity tests and were supported by overlapping WFEM and CSAMT profiles or by the JCDR-1 stratigraphy. Less-constrained peripheral and deep extensions were classified as inferred. The joint interpretation consequently reduced, but did not eliminate, the equivalence and non-uniqueness inherent in individual electromagnetic inversions.

The CSAMT results are presented in Figure 5. CSAMT Profile I is located in the western part of the study area and is approximately 5.8 km long. Within the uppermost ∼450 m, clustered high-resistivity anomalies are observed, with resistivity values generally ranging from 200 to 300 Ω m and locally exceeding 300 Ω m. These anomalies mainly correspond to unconsolidated Quaternary deposits composed of gravel, cobbles, sand, and silt. At depths of approximately 450–1,400 m, elliptical to stratiform low-resistivity anomalies occur, with resistivity values predominantly ranging from 20 to 50 Ω m. These anomalies are interpreted as Neogene mudstone, argillaceous sandstone, and fine conglomerate assemblages. Below a depth of 1,400 m, the resistivity increases to approximately 40–90 Ω m, corresponding mainly to Cretaceous–Jurassic sequences of mudstone, sandy mudstone, and sandstone (Figure 5a).

CSAMT Profile II crosses the Xiasifen Village area and is the profile most closely associated with the JCDR-1 well. This profile also exhibits a three-layer electrical structure characterized by a shallow high-resistivity layer, an intermediate low-resistivity layer, and a deep medium-to high-resistivity layer. The interval above 400 m is interpreted as a high-resistivity Quaternary layer. The continuous low-resistivity layer at depths of 400–1,400 m corresponds to Neogene mudstone, argillaceous sandstone, and fine conglomerate, whereas the increase in resistivity below 1,400 m reflects a Cretaceous–Jurassic clastic rock assemblage. Compared with Profile I, the low-resistivity layer along Profile II displays greater lateral continuity, indicating that the Neogene succession near the well site has a relatively uniform thickness and provides favorable caprock and localized groundwater storage conditions (Figure 5b). JCDR-1 is located near CSAMT station 22, approximately 2.2 km from the northwestern end of Profile II. Its projected position and borehole-confirmed stratigraphy are superimposed on Figure 5b to permit direct comparison with the resistivity structure.

CSAMT Profile III is located in the central part of the study area and crosses the Jin–A Expressway and the vicinity of Xiasifen Village. A continuous low-resistivity anomaly is developed at depths of approximately 400–1,600 m in the central part of the profile, with resistivity values mostly ranging from 10 to 50 Ω m. This low-resistivity layer is slightly thicker than those observed along Profiles I and II, suggesting the possible presence of a relatively subsided or sediment-thickened zone in the central part of the study area. Below 1,600 m, the resistivity increases, reflecting relatively compact Cretaceous–Jurassic clastic strata (Figure 5c). Collectively, the three CSAMT profiles indicate that the deep strata in the study area are generally distributed in a stratiform geometry, with no pronounced electrical anomalies indicative of major cross-cutting faults along the surveyed profiles.

As shown in Figure 6, the wide-field electromagnetic results provided further constraints on the deep stratigraphic boundaries and the positions of potential geothermal reservoirs. Both Profile II (Figure 6a) and Profile III (Figure 6b) exhibit a vertical electrical sequence characterized by high, low, medium to high, and low/high-resistivity zones. Along Profile II, the Quaternary, Neogene, and Cretaceous strata are approximately 337–350 m, 750–800 m, and 620–700 m thick, respectively. Along Profile III, the corresponding thicknesses are approximately 350–437 m, 800–850 m, and 620–700 m, respectively. The stratigraphic boundaries along both profiles show only minor variations in relief. However, the Neogene succession is locally thicker along Profile III, indicating a certain degree of sedimentary thickening in the central part of the study area.

The inversion-quality and sensitivity analyses described above evaluate the internal stability of the geophysical models, whereas external validation was conducted using the drilling and well-logging results from JCDR-1. For each interface, the midpoint of the pre-drilling geophysically predicted depth range was used as the representative predicted depth. The absolute discrepancy was calculated as Equation 11:and the relative discrepancy was calculated as Equation 12:where z geo is the representative geophysically predicted depth and z well is the corresponding depth identified in JCDR-1. As summarized in Table 2, the base of the Quaternary was predicted at 320–340 m, with a representative depth of 330 m, compared with the drilled depth of 340 m. The resulting absolute and relative discrepancies were 10 m and 2.9%, respectively. The base of the Neogene was predicted at 1,000–1,150 m, with a representative depth of 1,075 m, compared with the drilled depth of 1,110 m, giving discrepancies of 35 m and 3.2%. The base of the Cretaceous was predicted at 1800–1900 m, with a representative depth of 1850 m, whereas drilling encountered this boundary at 1764 m. The corresponding absolute and relative discrepancies were 86 m and 4.9%. Thus, the three principal stratigraphic interfaces were identified with relative depth discrepancies of less than 5%. A larger discrepancy was observed for the upper boundary of the principal geothermal reservoir. The pre-drilling geophysical interpretation placed this boundary at approximately 1,000 m, whereas drilling, lithological logging, and well logging identified the beginning of the principal water-bearing sandstone intervals at 850.8 m. The geophysical interpretation therefore overestimated the reservoir-top depth by 149.2 m, corresponding to a relative discrepancy of 17.5%.

D abs = z geo − z well (11)

D rel = z geo − z well z well × 100 % (12)

TABLE 2. Quantitative comparison between the pre-drilling geophysical predictions and the stratigraphic and reservoir boundaries identified in JCDR-1.

The discrepancy between the geophysical and drilling-derived reservoir-top depths should be interpreted as a difference between an electrical boundary and a hydrogeological boundary rather than as a systematic error in the geophysical depth model. The three major stratigraphic interfaces identified from the geophysical surveys show relative discrepancies of less than 5%, whereas the upper boundary of the principal water-bearing reservoir exhibits a larger discrepancy of 17.5%. This contrast suggests that the geophysical methods were more effective in resolving major stratigraphic contacts than in locating the hydraulic onset of individual permeable sandstone intervals.

Several factors may account for this discrepancy. First, the Neogene succession above and around the reservoir top is lithologically heterogeneous, consisting of interbedded mudstone, sandy mudstone, sandstone, and fine conglomerate. Therefore, the transition from the caprock to the water-bearing reservoir is not necessarily associated with a sharp resistivity contrast. Instead, the geophysical response represents an integrated electrical response of multiple lithologies. In particular, sandstone interbeds within the predominantly argillaceous Neogene succession may modify the local resistivity response and cause the interpreted electrical boundary to differ from the first occurrence of hydraulically productive sandstone identified by drilling.

Second, resistivity is not uniquely controlled by lithology in this geothermal system. The JCDR-1 geothermal water is highly mineralized and characterized by a Cl–Na brine with TDS values of approximately 38.6–44.8 g/L. Saline pore fluids can substantially reduce formation resistivity and may partly overlap with the low-resistivity response expected from clay-rich caprock. Consequently, a low-resistivity zone cannot be uniquely equated with either impermeable mudstone or a specific water-bearing sandstone layer. This resistivity equivalence introduces an inherent uncertainty when the geophysical objective is to identify a hydraulic boundary rather than a major lithological boundary.

Third, the 200 m station spacing and the decreasing resolution with depth limit the ability of the WFEM and CSAMT surveys to resolve relatively thin or gradational reservoir-top transitions. The two-dimensional smoothness-constrained inversion further favors laterally and vertically smooth resistivity distributions. Such regularization is necessary to obtain stable models but can broaden sharp geological transitions and shift the apparent position of a diffuse electrical interface. The sensitivity tests using different starting models and regularization weights indicate that the principal electrical architecture is stable; therefore, the 149.2 m discrepancy is unlikely to result solely from numerical non-convergence or inversion failure. Nevertheless, inversion non-uniqueness and smoothing remain important contributors to boundary-position uncertainty.

Finally, JCDR-1 is located near, but not directly on, the CSAMT Profile II, and some lateral variation in Neogene thickness is observed between the geophysical profiles. Although the regional stratigraphic architecture appears laterally continuous, local lithological variations between the profile and borehole cannot be completely excluded. Thus, the observed discrepancy probably reflects the combined effects of lithological heterogeneity, resistivity equivalence, finite spatial resolution, inversion regularization, and lateral projection rather than a single source of error.

Accordingly, the geophysical results should be regarded primarily as a prediction of the regional electrical and stratigraphic architecture, rather than a direct measurement of the hydraulic reservoir boundary. The JCDR-1 borehole provides an essential calibration point that refines the upper reservoir boundary from approximately 1,000 m–850.8 m. This result highlights an important limitation of electromagnetic exploration in sedimentary-basin geothermal systems: geophysical methods can effectively delineate large-scale caprock–reservoir architecture, but the identification of the exact top of hydraulically productive intervals requires calibration with drilling, well logging, and hydrogeological data. For exploration in comparable sedimentary basins, electromagnetic interpretation should therefore be integrated with lithological, logging, and hydrogeological constraints, and geophysically inferred reservoir boundaries should be expressed with an explicit uncertainty range rather than treated as deterministic interfaces.

Overall, the geophysical investigation indicates that the geothermal reservoir system in the Sanjiaochenggou area is characterized by an unconsolidated Quaternary cover, beneath which fine-grained argillaceous deposits in the upper Neogene form a relatively continuous caprock. Sandstone, sandy conglomerate, and sandy mudstone units extending from the lower Neogene to the Cretaceous constitute the principal porous–fractured geothermal reservoirs. The study area can therefore be classified as a stratiform medium- and low-temperature geothermal system hosted in a sedimentary basin.

4.2 Characteristics of the geothermal reservoir temperature field

Borehole temperature logging was conducted in the JCDR-1 well after a 48 h thermal recovery period following the cessation of drilling-fluid circulation, thereby minimizing the disturbance of the borehole temperature caused by the circulating fluid. The well had a total depth of 2,003.5 m, and temperature measurements were obtained to a depth of 1,992 m. As shown in Figure 7, the borehole temperature generally increased with depth, reaching 26.5, 41.7, and 61.7 °C at depths of 0, 1,000, and 1,992 m, respectively. This temperature profile indicates a stable conductive geothermal regime at depth within the study area.

To quantitatively compare the rates of temperature increase within different depth intervals, interval geothermal gradients were calculated as the ratio of the temperature difference to the corresponding depth difference. The 26.5 °C value represents the measured wellhead temperature at the surface. From 0 to 1,000 m, the temperature increased from 26.5 °C to 41.7 °C, corresponding to an interval geothermal gradient of approximately 1.52 °C/100 m. From 1,000 to 1,992 m, the temperature further increased to 61.7 °C, and the interval geothermal gradient increased to approximately 2.02 °C/100 m. In particular, the local geothermal gradient within the 1,800–1,992 m interval reached 2.5 °C/100 m. The whole-well geothermal gradient calculated from the temperature difference between the borehole entrance (26.5 °C) and the maximum logging depth (61.7 °C at 1992 m) is approximately 1.77 °C/100 m.

Overall, the temperature–depth profile of the JCDR-1 well exhibits a vertical pattern characterized by a normal temperature increase in the shallow section, a relatively gentle temperature gradient within the geothermal reservoir interval, and a locally accelerated temperature increase at greater depths. The principal geothermal reservoir is deeply buried and overlain by argillaceous strata, which favors heat retention. Meanwhile, the relatively high temperatures and geothermal gradients observed at depth indicate that the reservoir is continuously supplied with conductive heat from the surrounding rocks.

4.3 Hydraulic characteristics of the geothermal reservoir

The test conditions and stabilized pumping responses are summarized in Table 3. Stage S1 produced the maximum short-term stabilized discharge of 2,016.58 m³/d at a drawdown of 103.80 m and remained stable for 48 h. The corresponding wellhead temperature and specific capacity were 47.0 °C and 0.225 L/(s·m), respectively. Across the three stages, specific capacity ranged from 0.164 to 0.225 L/(s·m), and no abrupt decrease in discharge was observed during the respective stabilized periods.

TABLE 3. Design and stabilized results of the three constant-rate pumping stages.

As shown in Figures 8a,b, stabilized discharge increased with drawdown. However, because the three stages were conducted separately on different dates rather than sequentially as a standard continuous step-drawdown test, the Q-s relationship is presented only as a descriptive comparison among the pumping stages. Although the Jacob relationship s = B Q + C Q 2 was considered, the available data were not used to separate linear aquifer loss from nonlinear well loss or to calculate well efficiency. The shape of the Q-s relationship alone was therefore not used as evidence of confined-aquifer conditions.

The Dupuit, water-level recovery, and Theis methods yielded transmissivity values of 19.69, 21.97, and 23.40 m²/d, respectively. Using the cumulative effective aquifer thickness of 428.05 m, the corresponding hydraulic conductivity values were 0.046, 0.051, and 0.055 m/d, respectively. The representative mean transmissivity and hydraulic conductivity were 21.69 m²/d and approximately 0.051 m/d, respectively. The representative elastic storage coefficient derived from the two higher-rate pumping stages was 3.44 × 10⁻³.

The method-specific transmissivity estimates ranged from 19.69 to 23.40 m²/d and differed from their representative mean by approximately −9.2% to +7.9%. This range represents method-related variability rather than a formal statistical confidence interval. The observed time-drawdown data and the graphical Theis type-curve matches are presented in Figures 8c,d. Because the hydraulic parameters were derived from single-well data, they represent equivalent single-well properties under the stated analytical assumptions.

Following completion of the pumping test, groundwater-level dynamics in the JCDR-1 well were monitored from January to July 2024. As shown in Figure 9, the depth to water level gradually decreased from 49.63 to 49.35 m, while the corresponding hydraulic head increased from 1,336.01 to 1,336.29 m, representing a cumulative recovery of 0.28 m. No pronounced seasonal fluctuations were observed during the monitoring period, indicating that the geothermal reservoir pressure exhibited a gradual recovery after the pumping test and that no persistent decline in water level occurred over the short term.

In summary, the geothermal reservoir encountered by the JCDR-1 well is characterized by pronounced confined conditions, relatively stable transmissive capacity, and recoverable water levels following pumping.

4.4 Hydrochemical characteristics of the geothermal fluids

Three geothermal water samples (SY009–SY011) were collected from the JCDR-1 well for analyses of major ions, specific constituents, stable isotopes, radioactive isotopes, and dissolved gases. The analytical results were compared with those obtained for shallow Quaternary groundwater, spring water, and bedrock fracture water.

As shown in Table 4, the geothermal water from the JCDR-1 well had pH values ranging from 6.83 to 7.16 and was therefore generally neutral. The total dissolved solids (TDS) concentration ranged from 38.573 to 44.800 g/L, with an average value of 40.991 g/L, indicating pronounced hypersaline characteristics. Na⁺ was the dominant cation, accounting for 75.1%–75.8% of the total milliequivalent concentration, whereas Cl⁻ was the dominant anion, accounting for 92.0%–92.8%. The TDS concentrations of all three samples exceeded 35 g/L, the commonly used boundary between very saline water and brine (Osborn et al., 2013). Accordingly, the JCDR-1 fluid is classified as a Cl–Na-type geothermal brine. Relatively high concentrations of Sr, metasilicic acid, and metaboric acid further indicate prolonged interaction between the deep fluid and the reservoir rocks.

TABLE 4. Major hydrochemical parameters of the geothermal fluids from the JCDR-1 well.

As shown in Figure 10a, the mean TDS concentration of the geothermal water was approximately 49.6 times that of the shallow Quaternary groundwater, while the mean Na⁺ and Cl⁻ concentrations were approximately 130.6 and 196.8 times higher, respectively. In contrast, the mean HCO₃ ⁻ concentration in the geothermal water was lower than that in the Quaternary groundwater. These differences indicate that the deep geothermal water and shallow groundwater evolved under distinctly different hydrochemical conditions.

As illustrated in Figure 10b, the cation composition of the Quaternary groundwater was jointly dominated by Ca²⁺, Mg²⁺, and Na⁺ + K⁺, whereas its anion composition comprised SO₄ ^2-, Cl⁻, and HCO₃ ⁻. By comparison, Na⁺ + K⁺ and Cl⁻ accounted for approximately 75.4% and 92.5% of the total cation and anion milliequivalent concentrations, respectively, in the geothermal brine. The Na⁺/Cl⁻ equivalent ratios of SY009, SY010, and SY011 were 0.88, 1.00, and 0.77, respectively, while Cl⁻ accounted for 92.1%–92.9% of the total major-anion charge. These results confirm a pronounced Na–Cl saline end-member signature but do not uniquely identify halite dissolution as the sole controlling process.

The CAI-I values of SY009, SY010, and SY011 were 0.114, approximately −0.001, and 0.221, respectively. The positive values in SY009 and SY011, together with Na⁺/Cl⁻ ratios below unity, indicate that reverse cation exchange may have removed Na⁺ from solution while releasing Ca²⁺ and Mg²⁺ from the aquifer matrix. The Ca²⁺/Mg²⁺ equivalent ratios ranged from 1.48 to 1.54, and Ca 2 + + Mg 2 + − HCO 3 + SO 4 2 ranged from approximately 121.5–125.9 meq/L. The geothermal samples also contained 2.027–2.211 g/L Ca²⁺, 0.825–0.873 g/L Mg²⁺, and 2.060–2.605 g/L SO₄ ^2-. The combined ion composition is consistent with contributions from chloride-bearing evaporite dissolution, gypsum or anhydrite dissolution, carbonate dissolution, and reverse cation exchange. The high TDS, strong Cl⁻ dominance, low HCO₃ ⁻ concentrations, old isotopic characteristics, and clear separation from the shallow-groundwater hydrochemical field additionally support mixing with or progressive evolution toward a saline formation-water component. These processes cannot be quantitatively apportioned using the present dataset; therefore, halite dissolution is treated as one contributing mechanism rather than a unique explanation.

As shown in Figure 11, the distribution of different groundwater types in the study area on the Piper diagram further confirms this hydrochemical distinction. The geothermal water samples from the JCDR-1 well are concentrated within the Cl–Na facies field, whereas the Quaternary groundwater samples are mainly distributed within the HCO₃ + SO₄–Ca + Mg facies field and its transitional zones. The absence of a continuous transition in ionic composition between the geothermal water and shallow groundwater indicates a weak hydraulic connection between the two systems and suggests that the deep geothermal water circulates and is stored within a relatively independent hydrogeological environment. Dissolved-gas analyses showed that N₂ was the dominant component of the geothermal water, accounting for approximately 82.62 vol%, followed by O₂ at 15.44 vol% and CO₂ at 0.54 vol%. H₂S, SO₂, and C₂H₆ were not detected.

As shown in Figure 12, the local meteoric water line derived from precipitation monitoring in the Jinchang area is δ 2 H = 7.92 δ 18 O + 1.63. The JCDR-1 geothermal water has δ¹⁸O and δ²H values of −10.8‰ and −87‰, respectively. The corresponding values are −11.6‰ and −84‰ for the spring-water sample and −11.7‰ and −87‰ for the bedrock-fracture-water sample. All three samples plot near the local meteoric water line within the isotopically depleted region, and the geothermal water shows no pronounced oxygen-isotope shift. These results support a predominantly meteoric origin and indicate that oxygen-isotope exchange between water and rock was limited under the medium- and low-temperature reservoir conditions.

Using the JCDR-1 sampling elevation of 1,385.63 m, the geothermal-water δ¹⁸O value of −10.8‰, a representative low-elevation precipitation value of −10.0‰, and the regional δ¹⁸O altitude gradient of −0.26‰/100 m, the recharge elevation was estimated as H = 1385.63 + − 10.8 − − 10.0 − 0.26 × 100 ≈ 1693 m, giving approximately 1,693 m. This elevation is approximately 307 m above the JCDR-1 well site and is consistent with recharge in elevated piedmont or basin-margin areas. The estimate is used as an approximate elevation constraint because precipitation isotopes vary seasonally and spatially and the geothermal water contains an old, slowly renewed groundwater component.

The geothermal water from the JCDR-1 well had a δ³⁴S value of 12.8‰, a ³H content of less than 1.0 TU, and a¹⁴C specific activity of 1.18 ± 0.33 Bq / g. The δ³⁴S value provides an additional constraint on the source and evolution of dissolved sulfate. Together with the relatively high SO₄ ^2- concentration (2.060–2.605 g/L) and elevated Ca²⁺ concentration (2.027–2.211 g/L), the δ³⁴S result is consistent with a contribution from sulfur-bearing sedimentary or evaporitic minerals, including the possible dissolution of gypsum or anhydrite. However, because local sulfur-isotope compositions of potential end members were not independently characterized, δ³⁴S alone cannot uniquely distinguish among different sulfate sources. Therefore, it is used here as supplementary evidence for prolonged water–rock interaction and hydrochemical evolution rather than as a definitive tracer of a specific mineral source.

The low tritium concentration indicates a limited contribution of modern recharge and is consistent with the presence of an old groundwater component. Using the radioactive-decay relationship described in Section 3.4, the radiocarbon-derived age range is approximately 5.20 × 10 4 ∼5.67 × 10 4 yr. This numerical range should not be regarded as an absolute groundwater residence time. The calculation does not include corrections for carbonate dissolution, dead-carbon contribution, or long-term water-rock interaction, which may alter the ¹⁴C activity and introduce additional systematic uncertainty. Therefore, the ¹⁴C-derived age estimate is used primarily as supporting evidence for the presence of a relatively old groundwater component and slow renewal, together with the low ³H content and hydrochemical characteristics.

Overall, the geothermal fluid from the JCDR-1 well is characterized by high mineralization, a Cl–Na hydrochemical facies, pronounced enrichment in Na⁺ and Cl⁻, and limited mixing with modern recharge water. The combined hydrochemical and isotopic evidence indicates a relatively long subsurface flow path and a comparatively closed deep geothermal reservoir environment.

5 Discussion and assessment of geothermal resource exploitation

5.1 Genetic mechanism of the geothermal system

The geophysical, drilling, temperature-logging, pumping-test, and hydrochemical results collectively indicate that the Sanjiaochenggou geothermal field is a medium-to low-temperature confined geothermal system dominated by stratiform clastic-rock reservoirs.

The strata in the study area are generally stable and laterally continuous, and the electrical boundaries of the Quaternary, Neogene, Cretaceous, and Jurassic sequences are clearly traceable. No major concealed cross-cutting faults were identified within the geophysical survey area. The available geophysical and borehole-temperature evidence supports a predominantly conductive thermal regime within the investigated area. The reservoir-caprock assemblage, conductive heating from depth, and lateral groundwater flow are considered important factors controlling the geothermal system. The JCDR-1 well revealed that the upper Neogene strata above 850.8 m are dominated by mudstone and sandy mudstone, forming a relatively low-permeability regional caprock, whereas sandstone and sandy conglomerate of the lower Neogene and Cretaceous strata between 850.8 and 1,764.0 m constitute a confined pore–fracture geothermal reservoir. This mud-rich caprock–coarse clastic reservoir assemblage provides favorable conditions for the storage of deep geothermal water and the retention of heat.

The temperature profile showed a generally steady increase with depth. Stable-isotope results indicate that the geothermal water is predominantly of meteoric origin, while the isotope-altitude calculation gives an approximate recharge elevation of 1,693 m, about 307 m above the JCDR-1 well site. This elevation is consistent with recharge in elevated piedmont and basin-margin areas rather than direct infiltration on the basin plain. This interpretation is also supported by the regional hydrogeological setting. Precipitation increases with elevation, Jurassic, Cretaceous, and Neogene clastic strata crop out along the basin margins, and fractured and coarse-grained units provide potential infiltration pathways. The regional topographic and hydraulic gradients are generally directed from the elevated southern and southwestern margins toward the northeastern basin interior. Similar δ²H and δ¹⁸O compositions in the geothermal water, spring water, and bedrock-fracture water provide additional evidence of hydraulic connection with basin-margin groundwater. Regional hydrogeological studies also indicate that mountain-derived precipitation and snowmelt constitute important sources of groundwater recharge in the Hexi Corridor (Osborn et al., 2013). Taken together, these observations support meteoric precipitation and seasonally generated snowmelt infiltrating mainly in elevated piedmont and basin-margin zones, probably including the Longshou Mountain margin, followed by slow lateral migration through Neogene and Cretaceous clastic aquifers. However, the stable-isotope data do not uniquely identify a specific recharge locality or flow path; this pathway therefore represents the interpretation most consistent with the available isotope, topographic, geological, and hydrogeological evidence. During prolonged circulation and conductive heating by the surrounding rocks, the recharge water gradually evolves into medium- and low-temperature geothermal water.

The deep fluid is a Cl–Na-type geothermal brine with a mean TDS concentration of 40.99 g/L. Its Cl⁻-dominated composition and near-unity Na⁺/Cl⁻ equivalent ratios indicate an important contribution from a Na–Cl saline end member, but the variation in Na⁺/Cl⁻ and CAI-I values shows that its evolution cannot be explained by halite dissolution alone. Dissolution of chloride-bearing evaporites, gypsum or anhydrite, and carbonate minerals, together with reverse cation exchange and mixing with or evolution toward saline formation water, provides a more consistent explanation of the measured Na⁺, Cl⁻, Ca²⁺, Mg²⁺, SO₄ ^2-, and HCO₃ ⁻ concentrations. The low tritium content, low ¹⁴C activity, high mineralization, and clear hydrochemical separation from the shallow Quaternary groundwater further indicate a long subsurface flow path, prolonged water–rock interaction, and slow renewal within the deep clastic geothermal reservoir. The three analytical methods yielded transmissivity estimates of 19.69–23.40 m²/d, with a representative mean of 21.69 m²/d. The interpretation of confined conditions is supported primarily by the thick argillaceous caprock, the reservoir-caprock structure identified through drilling and well logging, the hydraulic response during pumping, and the post-pumping water-level recovery. The Q-s relationship is used only as a descriptive comparison among the three pumping stages and is not regarded as independent evidence of confined-aquifer conditions. The long-term recharge capacity remains to be verified through production-reinjection monitoring and multiwell testing.

Based on the above evidence, the conceptual genetic model of the Sanjiaochenggou geothermal field is illustrated in Figure 13. Meteoric precipitation and seasonally generated snowmelt are interpreted to infiltrate mainly in elevated piedmont and basin-margin zones, with an approximate isotope-derived recharge elevation of 1,693 m. The recharge water probably migrates from the southern and southwestern basin margins toward the depression interior through laterally continuous clastic aquifers. Under the confinement provided by the argillaceous caprock of the upper Neogene, the water is stored within sandstone and sandy-conglomerate reservoirs of the lower Neogene and Cretaceous strata. Conductive heat transfer from depth progressively raises the fluid temperature, forming confined geothermal water characterized by medium to low temperature and high mineralization. The recharge location and lateral-flow pathway shown in Figure 13 represent the conceptual interpretation best supported by the current data rather than a uniquely demonstrated groundwater-flow path.

5.2 Geothermal resource assessment, uncertainty, and development potential

Five different quantities are evaluated in this section: heat in place, recovery-factor-based recoverable heat, maximum short-term stabilized discharge, analytical long-term single-well production, and preliminary field-scale analytical production. Heat in place and recovery-factor-based recoverable heat are energy quantities expressed in joules, whereas the remaining quantities describe observed or analytically estimated geothermal-fluid production rates over different spatial and temporal scales.

5.2.1 Heat in place and recovery-factor-based recoverable heat

The geothermal reservoir in the study area consists of clastic aquifers in the lower Neogene and Cretaceous strata. Because these two reservoir units are laterally continuous and vertically superimposed, they were simplified as a stratiform confined geothermal reservoir with a total thickness of 428.05 m for resource estimation. Thickness-weighted effective petrophysical parameters were adopted, and the volumetric heat-in-place method was used for the calculation. In this method, the heat stored within the geothermal reservoir is divided into two components: heat stored in the reservoir rock and heat stored in the geothermal fluid. The geothermal-fluid storage comprises both the static pore-water volume and the volume released through elastic storage under confined conditions. The 47.02 km² reservoir area represents the total assessment boundary delineated from the integrated geological, WFEM, and CSAMT interpretations. It should not be interpreted as an area uniformly verified by drilling. Within this boundary, 27.97 km², accounting for 59.49% of the total area, is classified as the controlled area constrained by Well JCDR-1 and the surrounding geophysical profiles. The remaining 19.05 km², accounting for 40.51%, is classified as inferred because its reservoir extent and properties are based primarily on geophysical interpretation and lateral extrapolation from the single deep well.

The geothermal-fluid storage volume can be expressed as Equation 13:

V w = A ϕ M + A S H (13)

The heat stored in the reservoir rock and geothermal fluid can be calculated as Equations 14, 15:

Q ϵ = A M ρ ϵ c ϵ 1 − ϕ T ϵ − T 0 (14)

Q w = V w ρ w c w T r − T 0 (15)

Accordingly, the total in-place heat of the geothermal reservoir is (as shown Equation 16):where V _w is the geothermal-fluid storage volume m³; Q _r, Q _w, and Q denote the heat stored in the reservoir rock, the heat stored in the geothermal fluid, and the total in-place heat of the geothermal reservoir (J), respectively; A is the geothermal reservoir area (m²); M is the geothermal reservoir thickness (m); ϕ is the porosity; S is the elastic storage coefficient; H is the hydraulic head above the top of the geothermal reservoir (m); ρ _r and ρ _w are the densities of the reservoir rock and geothermal fluid (kg/m³), respectively; c _r and c _w are the specific heat capacities of the reservoir rock and water [J/(kg·K)], respectively; T _r is the mean geothermal reservoir temperature (°C); and T ₀ is the local mean annual air temperature (°C).

Q = Q t + Q w (16)

The stratified reservoir architecture and the corresponding variations in thickness, porosity, and temperature are illustrated in Figure 14. At Well JCDR-1, the lower Neogene geothermal reservoir is 79.50 m thick and has a representative porosity and temperature of 0.197 °C and 44.4 °C, respectively. The Cretaceous geothermal reservoir is 348.55 m thick and has corresponding values of 0.185 °C and 55.3 °C. The combined effective thickness is 428.05 m, and the thickness-weighted porosity and reservoir temperature are 0.186 °C and 53.3 °C, respectively. These parameters are directly constrained at Well JCDR-1 but remain spatially uncertain when extrapolated across the full assessment area. The parameters used in the reference heat-in-place calculation are listed in Table 5.

TABLE 5. Parameters used to calculate the in-place heat of the geothermal reservoir.

The calculated geothermal-fluid storage volume is 3.87 × 10⁹ m³. The heat stored in the reservoir rock and geothermal fluid is 1.50 × 10¹⁸ J and 7.74 × 10¹⁷ J, respectively, yielding a total in-place heat of 2.27 × 10¹⁸ J. Rock-stored heat accounts for 66.1% of the total, indicating that most of the geothermal energy in the study area is stored within the solid reservoir framework, whereas the geothermal fluid primarily serves as the medium for heat transfer and production. The reference value of 2.27 × 10¹⁸ J applies to the full 47.02 km² geophysically delineated assessment area. When the resource is separated according to the level of spatial control, the 27.97 km² controlled area contains approximately 1.35 × 10¹⁸ J, whereas the 19.05 km² inferred area contains approximately 0.92 × 10¹⁸J. The inferred contribution therefore accounts for approximately 40.51% of the total reference estimate and has a lower level of confidence than the controlled contribution.

On the basis of the estimated in-place heat, the recoverable heat was further evaluated using the recovery-factor method (Equation 17):where Q _rec is the recoverable heat (J), and R _E is the geothermal energy recovery factor. The recovery factor was selected with reference to the lithology, effective porosity, and current exploration level of the geothermal reservoir. The Methods and Regulations for Geothermal Resources Assessment (DZ/T 0331–2020) recommends that recovery factors be determined according to reservoir lithology, effective porosity, reservoir temperature, and production-reinjection conditions. When site-specific data are limited, a recovery factor of 25% is recommended for Cenozoic sandstone porous reservoirs in large sedimentary basins with porosity greater than 20%, whereas values of 5%–10% are recommended for Mesozoic sandstone and igneous-rock fractured reservoirs (Wang et al., 2017). Recovery factors of 0.10–0.25 have also been reported for porous geothermal systems in the Pannonian Basin, where 0.10 was adopted as a conservative value (Ministry of Natural Resources of the People’s Republic of China, 2020). The reservoir evaluated in this study comprises a 79.50 m thick lower Neogene unit and a 348.55 m thick Cretaceous unit, with a thickness-weighted porosity of 18.6%. Because this mixed lower Neogene–Cretaceous clastic reservoir is constrained by only one deep well and has not been evaluated through long-term production-reinjection testing, RE = 10% was retained as the reference value. To evaluate the influence of the recovery-factor assumption, study-specific deterministic sensitivity values of 5% and 15% were additionally considered. The 5% value represents a lower sensitivity bound consistent with the lower empirical value applicable to less well-characterized Mesozoic sandstone reservoirs. The 15% value lies within the range reported for porous geothermal systems but was selected as a cautious upper sensitivity bound below the 25% value applicable to better-characterized Cenozoic sandstone reservoirs with porosity greater than 20%. Therefore, the 5%–15% interval should not be interpreted as a universal literature-derived range or a statistical confidence interval. At the reference heat-in-place estimate of 2.27 × 10¹⁸ J, recovery factors of 5%, 10%, and 15% correspond to recoverable heat values of 1.14 × 10¹⁷, 2.27 × 10¹⁷, and 3.41 × 10¹⁷ J, respectively. The 10% case is retained as the reference estimate, whereas the 5% and 15% cases are used only to demonstrate the sensitivity of recoverable heat to the assumed recovery factor.

Q ec = R E Q (17)

To propagate the principal uncertainties into the resource estimate, deterministic conservative, reference, and upper scenarios were evaluated using alternative values of reservoir area, effective thickness, porosity, reservoir temperature, and recovery factor. The conservative scenario includes only the 27.97 km² controlled area, whereas the reference and upper scenarios use the full 47.02 km² mapped assessment boundary without extending the geothermal reservoir beyond the geophysically delineated area. Because spatial thickness variations are not constrained by additional deep wells, effective thicknesses 20% below and above the reference value were adopted as deterministic screening bounds. The porosity and temperature ranges correspond to the measured values of the lower Neogene and Cretaceous reservoir units. The scenario assumptions and propagated results are presented in Table 6. The resulting heat-in-place range is 8.84 × 10¹⁷–2.86 × 10¹⁸ J, and the corresponding recoverable-heat range is 4.42 × 10¹⁶–4.29 × 10¹⁷ J. These values represent deterministic scenario bounds rather than statistical confidence intervals. A Monte Carlo calculation was not applied because the single-well dataset does not provide sufficient information to define defensible probability distributions or spatial correlations among reservoir area, thickness, porosity, temperature, and recovery factor.

TABLE 6. Deterministic uncertainty scenarios for the geothermal heat-in-place assessment.

5.2.2 Long-term single-well production scenario and preliminary field-scale development scenario scenarios

The long-term single-well calculation was implemented as an analytical screening assessment by coupling a radial confined-aquifer equation with a volumetric thermal-energy-balance equation. It was not a coupled transient groundwater-flow and heat-transport simulation. The daily analytical single-well production rate was calculated as Equations 19–22: where Q s w is the analytical single-well production rate (m³/d); T is transmissivity (m²/d); S a is the allowable additional drawdown assigned to an individual well (m); R p is the thermal-protection radius (m); r w is the pumping-well radius (m); M is the cumulative effective aquifer thickness (m); f is the ratio between the volumetric heat capacity of the geothermal water and that of the saturated reservoir; ρ w and ρ r are the densities of geothermal water and reservoir rock, respectively; c w and c r are their specific heat capacities; ϕ is porosity; T r is the representative reservoir temperature; T 0 is the reference temperature; Δ T is the allowable reservoir-temperature decrease; and E T represents the heat stored in the reservoir at temperature T. The factor 36,500 represents 100 years of continuous production expressed in days.

Q s w = 2 π T S a ln 0.473 R p r w (18)

R p = 36500 Q s w f δ M π (19)

f = ρ w c w ρ e c e (20)

ρ e c e = ϕ ρ w c w + 1 − ϕ ρ r c r (21)

δ = E T r − E T r − Δ T E T r = Δ T T r − T 0 (22)

The two equations for Q s w and R p were solved iteratively. An initial value of R p was used to calculate Q s w, after which R p was updated using the thermal-energy-balance equation. The procedure was repeated until the changes in both variables between successive iterations were less than 0.1%.

The calculation assumes an equivalent homogeneous and isotropic confined reservoir, radial flow toward an equivalent fully penetrating well, uniform initial reservoir temperature and thermophysical properties, and average volumetric heat depletion within a cylindrical reservoir volume. External hydraulic boundaries, leakage, recharge, conductive heat replenishment, advective heat transport, thermal dispersion, reinjection, thermal breakthrough, wellbore storage, and changes in produced-fluid temperature were not explicitly simulated. Therefore, the specified temperature-decline criterion represents an average volumetric heat-depletion threshold rather than a prediction of the time-dependent produced-fluid temperature.

Quantitative sensitivity was evaluated using three deterministic scenarios. Transmissivity was varied from 19.69 to 23.40 m²/d according to the range obtained from the pumping-test interpretations, allowable additional drawdown was varied from 15 to 25 m, and the allowable average temperature decline was varied from 1 °C to 3 °C. The evaluation period and reservoir thermophysical properties were held constant.

For preliminary field-scale screening, a rectangular areal-well model was used to represent an idealized uniformly distributed well group as an areal production system. The relationship between production intensity and drawdown at the center of the field is expressed as follows Equations 23, 24:

ε = Q mf 4 l x l y (23)

s c = ε t op μ e S * α , β (24)

After rearrangement, the preliminary analytical field-scale production rate can be expressed as Equation 25:where ε is the production intensity per unit area [m³/(d·m²)]; Q _wf is the preliminary analytical field-scale production rate (m³/d); l _*x* and l _*y* are the half-length and half-width, respectively, from the center of the rectangular production area to its boundaries (m); S _c is the allowable drawdown at the center of the production area (m); t op is the evaluated production period (d); μ _e is the elastic storage coefficient; S ^*(a,β) is the rectangular areal-well function; and (a,β) is dimensionless geometric parameters.

Q w i = 4 l x l y μ c s c t o p S * α , β (25)

The thermal power was calculated from the production rate and geothermal-water temperature as follows Equation 26:where P _t is the thermal power (kW), Q is the geothermal-fluid production rate (L/s), T _p is the produced-fluid temperature (°C), and T ₀ is the local mean annual air temperature (°C). The parameters used in the single-well and field-scale calculations are listed in Table 7.

P 1 = 4.1868 Q T p − T 0 (26)

TABLE 7. Parameters used in the analytical single-well and field-scale production assessments.

Under the reference assumptions, the analytical calculation yielded a single-well production rate of 335.82 m³/day and a thermal-influence radius of 629.39 m. The corresponding equivalent square protection area was approximately 1.58 km². Based on a produced-fluid temperature of 47.0 °C and a reference temperature of 4.8 °C, the corresponding thermal power was approximately 0.69 MW.

The deterministic sensitivity scenarios are presented in Table 8. The calculated single-well production rates ranged from 224.62 to 455.69 m³/d, the analytical thermal-influence radii ranged from 598.62 to 727.95 m, and the corresponding thermal powers ranged from 0.46 to 0.93 MW. The reference value of 335.82 m³/d should therefore be interpreted as a preliminary analytical production scenario rather than a verified sustainable yield.

TABLE 8. Deterministic sensitivity scenarios for the analytical single-well production assessment.

The rectangular areal-well model yielded a preliminary field-scale analytical production scenario of 4,951.30 m³/d, corresponding to a thermal power of 10.12 MW and an annual utilizable heat output of approximately 3.19 × 10⁸ MJ. This result represents an idealized multiple-well scenario based on homogeneous and isotropic reservoir properties, uniformly distributed production, fixed hydraulic boundaries, and spatial extrapolation of hydraulic parameters obtained from a single well. It should therefore not be interpreted as a verified field-scale production capacity or an engineering installed-capacity target.

The maximum short-term stabilized discharge of 2,016.58 m³/d, the analytical long-term single-well production scenarios of 224.62–455.69 m³/d, and the preliminary field-scale analytical production scenario of 4,951.30 m³/d have different physical meanings. The first is a directly observed short-term well response under a drawdown of 103.80 m. The second consists of management-constrained analytical single-well scenarios for a 100-year evaluation period. The third is a regional extrapolation based on an idealized multiple-well model. These values are therefore not reported or compared as equivalent measures of sustainable production capacity.

5.2.3 Uncertainty analysis

Uncertainty in this study is treated as a hierarchy of evidence rather than as a single statistical error term. The principal inputs and assessment outputs are classified as directly measured, derived from geophysical inversion, estimated analytically, adopted from technical standards or literature, or extrapolated spatially. Measurements from JCDR-1 provide direct constraints at the well scale, whereas regional resource and field-scale production estimates additionally depend on inversion interpretation, spatial extrapolation, assumed parameters, and analytical model structure.

The reliability of the geothermal resource assessment depends on the spatial extent of the reservoir and the representativeness of the thickness, porosity, temperature, and thermophysical parameters applied within that area. The reference heat-in-place estimate of 2.27 × 10¹⁸ J applies to the full 47.02 km² geophysically delineated assessment area. However, only 27.97 km² is classified as controlled, with an estimated heat in place of 1.35 × 10¹⁸ J. The remaining 19.05 km² is classified as inferred and contributes approximately 0.92 × 10¹⁸ J to the reference total. Because the inferred area accounts for 40.51% of the total assessment area, the full-area estimate has a lower level of confidence than the controlled-area estimate and should be interpreted accordingly.

To identify the dominant parameters controlling the in-place heat estimate, Equations 1–4 can be rearranged as follows Equation 27:

Q = A T r − T 0 M ρ r c r 1 − ϕ + ϕ ρ w c w + S H ρ w c w (27)

A normalized sensitivity coefficient was used to evaluate the relative influence of parameter (X) on the estimated in-place heat (Equation 28):where S x is the normalized sensitivity coefficient of parameter X. When all other parameters remain constant, S x represents the percentage change in in-place heat caused by a 1% relative change in parameter (X).

S x = X Q ∂ Q ∂ X (28)

As shown in Figure 15, the local normalized sensitivity coefficients indicate that the calculated heat in place is most responsive to the geothermal reservoir area, temperature difference, and effective reservoir thickness. These coefficients describe the response of the calculation to a relative change in one parameter while holding the other parameters constant; they do not represent parameter uncertainty ranges or statistical confidence intervals. The deterministic scenarios in Table 7 additionally propagate the specified ranges of reservoir area, thickness, porosity, temperature, and recovery factor. The resulting heat-in-place estimates range from 8.84 × 10¹⁷ to 2.86 × 10¹⁸ J, compared with the reference value of 2.27 × 10¹⁸ J.

Uncertainty also arises from the single-well pumping-test interpretation. The Dupuit, water-level recovery, and Theis methods yielded transmissivity estimates of 19.69, 21.97, and 23.40 m²/d, respectively, with a representative mean of 21.69 m²/d. The corresponding hydraulic conductivity estimates ranged from 0.046 to 0.055 m/d, with a representative mean of approximately 0.051 m/d. The individual transmissivity estimates differed from the mean by approximately −9.2% to +7.9%. This range represents method-related variability rather than a formal statistical confidence interval. Moreover, the analytical models did not explicitly represent leakage, external boundaries, vertical heterogeneity, partial penetration, wellbore storage, skin effects, or nonlinear well losses. Additional observation-well pumping tests and multiwell interference tests are required to constrain reservoir-scale hydraulic properties and boundary effects.

These hydraulic-parameter uncertainties propagate directly into the analytical single-well and field-scale production scenarios. The 100-year single-well calculation is controlled by transmissivity, allowable additional drawdown, and the allowable average temperature-depletion criterion. The deterministic sensitivity assessment yielded analytical production rates of 224.62–455.69 m³/d, compared with the reference value of 335.82 m³/d.

The volumetric thermal-energy balance represents average heat depletion within an idealized cylindrical reservoir volume. It does not simulate conductive heat replenishment, advective heat transport, thermal dispersion, reinjection, thermal breakthrough, or the time-dependent temperature of the produced fluid. Consequently, the 2 °C criterion should not be interpreted as a prediction that the produced-fluid temperature will decline by exactly 2 °C after 100 years.

The hydraulic and thermophysical parameters used in the field-scale calculation were derived predominantly from Well JCDR-1 and spatially extrapolated across the assessment area. The field-scale scenario further assumes uniformly distributed production wells, homogeneous and isotropic reservoir properties, and fixed hydraulic boundaries. Consequently, the calculated production rate of 4,951.30 m³/d and thermal power of 10.12 MW represent conditional outputs of a preliminary analytical development scenario rather than verified sustainable production limits or engineering design values. Additional drilling, observation-well and multiwell interference tests, long-term production–reinjection tests, and coupled groundwater-flow and heat-transport simulations are required before sustainable field-scale production and engineering development limits can be established.

To integrate the uncertainty sources arising from the geophysical, drilling, temperature-logging, pumping-test, isotope, recharge, resource-assessment, and production analyses, Table 9 summarizes the evidence basis, principal uncertainty, treatment, and remaining limitation of each major parameter or assessment output.

TABLE 9. Uncertainty hierarchy and treatment of the principal data, parameters, and assessment outputs.

5.3 Development suitability and constraints on sustainable utilization

The temperature and hydrochemical characteristics of the Sanjiaochenggou geothermal field indicate that the resource is more suitable for direct-use applications, including district heating, building heating, greenhouse climate control, and health and wellness services, than for geothermal power generation. During Stage S1, the wellhead temperature was approximately 47 °C. The maximum short-term stabilized discharge of 2,016.58 m³/d represents the pumping-test response of Well JCDR-1 under a drawdown of 103.80 m and should not be used as a long-term production target. The reference analytical long-term single-well production scenario of 335.82 m³/d, corresponding to approximately 0.69 MW, provides a preliminary constraint for pilot-scale planning because it incorporates a 100-year evaluation period, a 20 m allowable additional drawdown, and a 2 °C average volumetric heat-depletion criterion. However, it remains a conditional analytical scenario rather than a field-validated sustainable yield. Similarly, the preliminary field-scale analytical production and thermal-power scenario of 4,951.30 m³/d and 10.12 MW should not be converted into an engineering production or installed-capacity target before additional drilling and multiwell production-reinjection data become available.

The Cl–Na-type geothermal brine represents an important constraint on development. Its TDS concentration ranges from 38.573 to 44.800 g/L, exceeding the commonly used 35 g/L threshold for brine, while the Cl⁻ concentration reaches 18.493–25.900 g/L. This composition creates a substantial chloride-corrosion risk for well casing, pumps, pipelines, heat exchangers, and reinjection facilities. The elevated Ca²⁺, Mg²⁺, SO₄ ^2-, hardness, and dissolved-silica concentrations also indicate potential carbonate-, sulfate-, and silica-bearing deposition as temperature, pressure, pH, and gas conditions change during production and cooling. Corrosion-resistant materials should therefore be selected for components in contact with the brine, atmospheric-oxygen ingress should be minimized, and corrosion- and scale-inhibitor performance should be verified under site-specific operating conditions.

Direct surface discharge of the spent geothermal brine is unsuitable because it could cause soil salinization and deterioration of shallow-groundwater and surface-water quality. Reinjection should therefore be incorporated into the development scheme (Osborn et al., 2013). Before field-scale operation, production and reinjection brines should be tested for chemical compatibility and precipitation potential, and a pilot reinjection test should be used to determine the acceptable reinjection temperature, pretreatment requirements, filtration specifications, inhibitor dosage, and injectivity response. Operational monitoring should include temperature, pH, TDS, Cl⁻, Ca²⁺, Mg²⁺, SO₄ ^2-, dissolved silica, suspended solids, injection pressure, corrosion products, and changes in injectivity.

The geothermal reservoir in the study area is a confined system characterized by lateral recharge, slow groundwater flow, and relatively closed hydrodynamic conditions. Its long-term sustainability is controlled not only by the amount of stored heat but also by pressure maintenance and reinjection performance. Reinjection can reduce the environmental risks associated with the discharge of highly mineralized spent geothermal water while helping to maintain reservoir pressure and mitigate continuous water-level decline. Reinjection should therefore be regarded as a prerequisite for development rather than as an auxiliary measure implemented after production. Because long-term coupled production–reinjection tests have not yet been conducted, the reinjection horizon, interwell spacing, reinjection temperature, and production-to-reinjection ratio should be determined through field testing.

A coordinated development mode integrating “conservative production, cascade utilization, reinjection compatibility, and dynamic monitoring” is recommended. During the initial stage, paired production and reinjection tests should be conducted with reference to the analytical long-term single-well production scenarios, and baseline data for water level, temperature, flow rate, and hydrochemistry should be established. Subsequently, operation over a complete heating season should be used to characterize drawdown recovery, temperature decline, reinjection pressure, and hydrochemical evolution. Expansion through additional wells, stratified exploitation, and staged development should be implemented only after the model parameters have been calibrated against long-term dynamic monitoring data.

Overall, the development suitability of the Sanjiaochenggou geothermal field can be summarized as follows: the conditions are favorable for the direct utilization of medium- and low-temperature geothermal energy; long-term single-well production should be conservatively controlled; highly mineralized geothermal fluids must be disposed of through reinjection; and field-scale expansion requires dynamic verification. Accordingly, a development strategy consisting of “conservative initiation, production–reinjection verification, model calibration, and phased expansion” is recommended so that the resource assessment results can be progressively converted into sustainable engineering development capacity.

5.4 Recharge origin, groundwater residence time, and modern recharge constraints

The isotopic and hydrochemical evidence supports a meteoric origin of the geothermal fluid, with recharge conceptually linked to the mountain-front region along the basin margin and subsequent deep lateral groundwater flow toward the study area. However, the recharge area should not be regarded as a precisely delineated boundary. The available data constrain the recharge setting at the regional scale, but do not provide sufficient information to determine the exact recharge extent, recharge elevation, or spatial distribution of recharge flux. Therefore, the mountain-front recharge zone shown in the conceptual model is interpreted as a plausible source region rather than a uniquely verified recharge boundary. This interpretation is consistent with the present evidence from stable isotopes, tritium, radiocarbon, hydrochemistry, and regional hydrogeology, while recognizing that recharge rates and detailed flow paths remain insufficiently constrained.

The ¹⁴C ages of approximately 52,000–57,000 years provide evidence for a long residence time and substantial contribution from old groundwater in the geothermal system. These ages should not, however, be interpreted directly as the travel time from the basin-margin recharge zone to the study area. A quantitative comparison between groundwater travel time and the ¹⁴C ages would require independent constraints on recharge location, flow-path length, hydraulic gradient, effective hydraulic conductivity, groundwater velocity, and the geochemical correction applied to the initial ¹⁴C activity. Such information is currently unavailable from the single-well dataset. Consequently, the present study interprets the radiocarbon ages primarily as evidence for prolonged groundwater residence and deep circulation rather than as a direct measurement of basin-scale groundwater travel time.

The description of the geothermal system as relatively closed should likewise be understood in a hydrodynamic rather than an absolute sense. The deep reservoir is characterized by slow groundwater movement, limited hydraulic exchange with shallow groundwater, and long residence times, indicating that the contribution of modern recharge to the stored geothermal water is probably limited. This interpretation does not imply that recharge has been completely absent throughout the geological history of the basin. The present geothermal water may represent water that entered the subsurface during earlier recharge periods and subsequently underwent prolonged deep circulation and water–rock interaction. Thus, paleorecharge and modern effective recharge should be distinguished in the conceptual model: evidence for a meteoric origin demonstrates the ultimate source of the water, whereas it does not by itself demonstrate that the present reservoir is being rapidly or continuously replenished under current climatic and hydrological conditions.

The available isotopic evidence also places an important constraint on the interpretation of modern recharge. Stable isotopes support the meteoric origin of the geothermal fluid, whereas the old ¹⁴C ages and the limited indication of modern water contribution from tritium are more consistent with a long-residence, relatively weakly renewed deep groundwater system. Nevertheless, the present dataset does not allow the current recharge rate to be quantified. In particular, the absence of a regional network of monitoring wells, hydraulic-head observations, environmental tracers, and distributed recharge measurements prevents a defensible estimate of modern recharge flux. We therefore avoid treating the geothermal reservoir as a continuously replenished system in the resource assessment and instead regard modern recharge as an important but currently unresolved component of the long-term water balance.

This distinction is important for geothermal development. The large heat-in-place estimate reflects the amount of stored thermal energy within the delineated reservoir and should not be interpreted as evidence of an equivalent rate of natural replenishment. Given the inferred slow renewal of the deep groundwater system, long-term production cannot be assumed to be sustained by natural recharge alone. Reinjection therefore represents an important strategy for pressure maintenance, reduction of reservoir depletion, and environmentally responsible management of the highly mineralized geothermal water. However, the feasibility and effectiveness of reinjection cannot be demonstrated from the present single-well dataset. Dedicated reinjection tests and, preferably, multiwell production–reinjection monitoring are required to quantify injectivity, pressure propagation, hydraulic connectivity, potential thermal breakthrough, and the long-term balance between production and reinjection. Accordingly, the present development assessment should be regarded as a preliminary resource-screening framework rather than proof of a sustainable field-scale production capacity under current recharge conditions.

5.5 Methodological limitations

The experience from the Sanjiaochenggou area also provides several methodological implications for geothermal exploration in comparable sedimentary basins. Electromagnetic methods such as WFEM and CSAMT are particularly useful for delineating regional stratigraphic architecture, identifying major resistivity contrasts, and constraining the spatial extent of potential caprock–reservoir systems. However, their ability to identify the exact hydraulic top of a reservoir may be limited where the reservoir consists of interbedded clastic units, where saline formation water produces resistivity equivalence with clay-rich strata, or where the reservoir boundary is gradational rather than sharply defined. In such settings, the geophysical interpretation should preferably provide a depth interval or uncertainty envelope rather than a single deterministic reservoir-top depth. The uncertainty can be reduced by combining multiple electromagnetic methods, performing sensitivity tests on inversion parameters, increasing profile density near prospective targets, and calibrating the electrical model with drilling and well-logging information. Therefore, the principal value of geophysical exploration in such basins is not to replace drilling verification, but to reduce the search space and improve the geological targeting of subsequent boreholes.

6 Conclusion

This study integrated geophysical exploration, drilling, temperature logging, pumping tests, hydrochemical analyses, and isotopic investigations to characterize the geothermal reservoir, hydrodynamic conditions, genetic mechanism, and resource potential of the Sanjiaochenggou geothermal field. The main conclusions are as follows:

• The Sanjiaochenggou geothermal system is a sedimentary-basin-hosted medium- and low-temperature confined geothermal system. The principal geothermal reservoirs are hosted in sandstone and sandy conglomerate of the Lower Neogene and Cretaceous strata, with overlying argillaceous formations providing an effective caprock.
• The geothermal system is predominantly conduction-dominated. The observed temperature increase with depth and the absence of clear evidence for major fault-controlled convection indicate that conductive heat transfer through the surrounding rocks is the principal thermal mechanism within the investigated area.
• The geothermal reservoir exhibits confined hydrodynamic characteristics and has a measurable capacity for groundwater transmission and pressure recovery. The pumping-test results indicate moderate reservoir transmissivity and support the feasibility of geothermal-water production, although long-term production capacity requires further dynamic verification.
• The geothermal fluid is characterized by a Cl–Na hydrochemical facies and high mineralization. Hydrochemical and isotopic evidence indicates a predominantly meteoric origin, deep circulation, prolonged water–rock interaction, and a relatively old and slowly renewed groundwater component. Evaporite dissolution, reverse cation exchange, and evolution toward a saline formation-water component may have jointly contributed to the observed hydrochemical characteristics.
• The geophysically delineated assessment area indicates considerable geothermal resource potential, with the reference heat-in-place estimate providing a preliminary basis for resource assessment. However, the current resource and production estimates remain subject to uncertainties associated with geophysical reservoir-boundary interpretation, spatial extrapolation from a single deep well, recovery-factor assumptions, and simplified long-term production models. Therefore, sustainable development should proceed conservatively and be supported by long-term production–reinjection testing, monitoring, and further reservoir characterization.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Paolo Capuano, University of Salerno, Italy

Reviewed by: Zhengzheng Cao, Henan Polytechnic University, China

Ibrar Iqbal, Guilin University of Technology, China