The study area is located within the Muli Coalfield in Tianjun County, Qinghai Province, at an elevation of 4,000 to 4,300 m. The region experiences an average annual temperature of −5.1°C and is characterized by widespread permafrost with an island-like distribution and an average thickness of 95 m (Zhu et al., 2006; Pan et al., 2008). Geologically, the area is part of a rift-graben basin formed by the reactivation of the northern Qilian deep fault system during the Yanshanian orogeny. Tectonically, it is situated within a depression zone between the central and southern Qilian structural units (Wen et al., 2010; Fu and Zhou, 2000) (Figure 1A). Hu indicates that the Cretaceous orogeny in the Qilian Mountains led to the region’s early uplift. Significant tectonic and uplift events occurred during the early Eocene (53–40 Ma), late Eocene (∼33 Ma), the Miocene (13–8 Ma), and the Pliocene, raising the Qilian Mountains to near their current elevation (Hu et al., 2017). By the end of the Pliocene, the range had reached the critical height required for glacial activity, with the Middle Pleistocene showing pronounced effects from global cooling and coupling influences (Hu et al., 2017).

The Juhugeng mining area is one of the main components of the Muli coalfield. Its central region consists of Triassic strata forming an anticline, while coal-bearing Jurassic strata create two synclines to the north and south (Wen et al., 2010; Fu and Zhou, 2000). Large-scale thrust faults are developed on both sides of the anticline, while significant NE-trending shear faults are present within the northern and southern synclines, segmenting the depression into discontinuous fault blocks (Lu et al., 2010). The area exhibits distinct structural characteristics, characterized by a north-south zonal and east-west sectional division (Figure 1B). The exposed strata in the Juhugeng mining area mainly consist of Quaternary, Middle Jurassic, and Upper Triassic formations. The Upper Triassic strata are prominently exposed in the northern and southern parts of the mining area, as well as along the axis of the anticline. These formations primarily comprise black siltstone, mudstone, and thin coal seams, exhibiting a parallel unconformity with the overlying Jurassic strata (Figure 1B). The Muli Formation, from bottom to top, consists of a sequence of medium-to coarse-grained clastic rocks formed in a braided river alluvial plain environment, occasionally interbedded with thin layers of carbonaceous mudstone or thin coal seams. The basal conglomerates are well-developed, transitioning to dark gray siltstone, fine-grained sandstone, and medium-grained sandstone, as well as coarse-grained sandstone formed in a lacustrine-swamp environment, with two main coal seams interbedded. The Jiangcang Formation, from bottom to top, consists of gray fine-grained sandstone, medium-grained sandstone, and dark gray mudstone and siltstone, deposited in a delta-lacustrine environment, containing 2 to six coal seams (Fu and Zhou, 2000; Hu et al., 2017; Pang et al., 2013). It transitions to fine clastic mudstone and siltstone formed in a shallow to semi-deep lake environment, interbedded with gray siltstone and lenticular siderite layers. The Quaternary strata are widely distributed in the drilling and consist of alluvial and diluvial humus, sand, gravel, colluvial breccia, and glacial deposits including mud, sand, ice layers, and erratic boulder.

Since 2008, the China Geological Survey has drilled 16 exploratory wells and 3 pilot production wells in the permafrost region of the Qilian Mountains. NGH were discovered in 9 of the exploratory wells and 2 of the pilot production wells, while the remaining exploratory wells exhibited only hydrate-related anomalies. Shenhua Qinghai Energy Co., Ltd. Has drilled 14 exploratory wells for NGH, with discoveries in 4 of these wells; the remaining wells showed only hydrate-related anomalies (Wang et al., 2019; Wen et al., 2015). Recent drilling investigations and laboratory studies have significantly clarified the fundamental geological characteristics of NGH.

In the permafrost region of the Qilian Mountains, NGH deposits are classified into two types. The first type, known as “fracture-type” hydrate, is visibly present as thin layers, platelets, or nodules within the fractures of siltstone, mudstone, and oil shale. These typically appear as white or milky white crystals, though some may appear yellow due to contamination from drilling mud. The second type, “pore-type” hydrate, is found as impregnations within the pores of fine siltstone. The crystals are not visible to the naked eye and must be inferred from indirect indicators such as the continuous emergence of gas bubbles and water droplets from core samples, or dispersed low-temperature anomalies detected by thermal infrared imaging. NGH are distributed within a depth range of 140–330 m and have a thickness of nearly 200 m (Wang et al., 2019), which is relatively shallow compared to polar permafrost regions. The vitrinite reflectance (Ro) of the source rocks ranges from 0.78% to 1.1%, with the maximum pyrolysis peak temperature (Tmax) reaching 470°C, indicating that the source rocks are in a stage of thermal maturation where significant oil and gas generation occurs. The methane content in the hydrate layers ranges from 54% to 76%, ethane from 8% to 15%, and propane from 4% to 21%, with minor amounts of butane, pentane, and other hydrocarbons. The CO2 content generally ranges from 1% to 7%, with some samples reaching up to 15%–17%. The spectral curves of the hydrates are similar to those of deep-sea hydrates from the Gulf of Mexico, classifying them as Type II hydrates. Carbon isotope studies indicate that the gas sources for the NGH in the Qilian Mountains primarily originate from deeper formations. The gas generated in these source rocks migrates along faults to shallower levels, where it accumulates either directly or indirectly due to late-formed compressional faults blocking the gas. Since the late Pleistocene glaciations, this gas has either formed hydrates or remained in the strata as free or adsorbed gas (Zhang et al., 2022).

Permafrost is a necessary condition for the formation of terrestrial NGH. The long-term real-time temperature monitoring results from the DK-9 well indicate that the geothermal gradient within the permafrost layer is 1.38°C per 100 m, while the geothermal gradient below the permafrost layer is 4.85°C per 100 m. Based on the temperature-pressure conditions for hydrate stability, the depth of the stability zone’s lower boundary is between 510 and 617 m (Wang et al., 2017). The study area possesses favorable conditions for the formation of NGH. The drilling results show that the main gas hydrate bearing horizon is the Middle Jurassic Jiangcang Formation, located beneath the permafrost layer. The primary reservoir intervals are between depths of 133.0–396 m (Wang et al., 2015). The vertical distribution of NGH in the wells is discontinuous, and there is no clear lateral distribution pattern between wells. The rock fracture system plays a significant controlling role in the distribution of NGH (Wang et al., 2011). NGH are found beneath the permafrost layer, and no gas hydrates or their associated anomalies have been observed within the permafrost layer (Wang et al., 2015). There is a significant variation in the estimated gas hydrate saturation range. The average hydrate saturation in sandstone pores is 5.1% (Lu et al., 2011), while the saturation ranges calculated from logging data are 1.3%–8.6% and 11.47%–81.61% (Guo and Zhu, 2011; Lin et al., 2018), with relatively similar results. However, when selecting saturation calculation models, certain assumptions are made, and further work is required, including rock physics experiments, petrophysical testing, and formation water analysis (Lin et al., 2017).

Hg is highly volatile, and during the heating process in the TRM method, both inorganic and organic Hg in the soil is converted into Hg vapor, which may mix with Hg related to oil and gas. The presence of Hg unrelated to oil and gas can introduce significant biases in the application of TRM. In geochemical exploration for oil and gas, considering the total Hg (THg) content can help mitigate the interference affecting TRM indicators. In soil samples, locations where TRM levels are high while THg levels are relatively low can be identified as anomaly points associated with micro-leakage of oil and gas. The statistical results for TRM and THg are presented in Table 1. The minimum TRM value is 5.9 ngg⁻¹, the maximum is 127.37 ngg⁻¹, and the average is 32.59 ngg⁻¹, with a coefficient of variation of 0.66. For THg in soil, the minimum value is 10 ngg⁻¹, the maximum is 317 ngg⁻¹, and the average is 83.40 ngg⁻¹, with a coefficient of variation of 0.63.

Figure 2 presents the contour map of soil TRM, identifying an anomalous lower limit of 39.24 ngg⁻¹ using a dual-logarithmic approach with frequency and concentration. A low-value anomaly is observed above the NGH deposits, where the thickness of the permafrost exceeds 65 m, meeting the temperature and pressure conditions necessary for hydrate formation. Exploration wells DK-4, DK-10, DK10-16, DK10-17, and DK10-18 were found to be dry, marking the boundary of the hydrate, while high-value anomalies in soil TRM correspond to this hydrate boundary, consistent with the findings from the exploration wells. Based on geochemical exploration results, including the current soil TRM survey, four additional wells—DK-9, DK13-11, DK12-13, and DK11-14—were strategically placed in favorable hydrate zones, all of which confirmed the presence of NGH (Figure 2).

The collected samples were also tested for acid-soluble hydrocarbons, analyzing the concentrations of hydrocarbon components from C₁ to C₅. This study utilized data on acid-soluble methane and heavy hydrocarbons, conducting initial analyses using EXCEL software. Through frequency grouping and examination of data distribution characteristics, the data were divided into 15 logarithmic intervals. The kriging method was then employed for interpolation, resulting in the creation of geochemical maps.

Figures 3, 4 present the geochemical maps for soil acid-soluble methane and heavy hydrocarbons, respectively. The distribution of acid-soluble methane and heavy hydrocarbons exhibits both similarities and differences when compared to the anomalies in soil TRM. These variations are primarily reflected in the following aspects: (1) The correlation between acid-soluble methane and heavy hydrocarbons is notably high (R ² = 0.6614), indicating a common origin. In the northwest coalfield region and directly above the NGH deposits, both exhibit anomalous distributions. The higher concentrations in the coalfield area are attributed to coalbed methane (Zhang et al., 2022). Conversely, the acid-soluble hydrocarbon anomalies above the NGH deposits are a result of hydrocarbon migration associated with the hydrates, suggesting that these anomalies could serve as indicators of NGH reservoirs. (2) The acid-soluble hydrocarbon anomalies display an annular pattern in areas where NGH are anomalous. Similarly, the soil TRM also shows an annular anomaly, indicating a high degree of overlap between the two. Both anomalies are characterized by large areas and high intensities, reinforcing their correlation. (3) In the southeastern part of the study area, anomalies in soil acid-soluble hydrocarbons, including methane and heavier hydrocarbons, correspond to the annular anomalies in soil TRM. This correlation indicates that the area could be a priority target for NGH exploration. The presence of TRM can serve as an auxiliary indicator to enhance the success rate of hydrocarbon exploration. (4) The central part of the study area exhibits significant TRM anomalies, which are attributed to the active geological gas migration associated with fault development in the region. In this area, the intensity of acid-soluble hydrocarbon anomalies, particularly methane, is relatively low, suggesting a reduced likelihood of NGH formation.

Fractal theory, which is defined by characteristics such as fractal dimensions, self-similarity, statistical self-similarity, and power-law distributions, has been applied in the field of geology (Li and Ma, 1999; Cheng et al., 1994). It provides an effective approach for uncovering the spatiotemporal structural features of mineralization systems (Cheng, 1999; Cheng, 2001; Shen, 2002; Li and Cheng, 2004; Zuo and Wang, 2015). This approach enables geoscientists to explore the complex, often non-linear patterns in geological formations, enhancing our understanding of the processes that govern resource distribution and refining exploration technologies.

The fractal statistical model is Formula 1. N r = C r − D , r > 0 (1) where r denotes the content of TRM; C is the proportional constant, which is >0; D is the fractal dimension, which is >0, and N(r) represents the amount of He, Ne, and CH₄ in the drill core headspace gas with content equal to or greater than r. This fractal statistical model effectively captures the overall spatial structural characteristics of geochemical anomalies. By taking the logarithm of both sides of the fractal statistical model for TRM in soil, the equation can be transformed into a simple linear regression Formula 2: lg N r   =   − D ⁡ lg r + lg C , r > 0 (2)

Figure 5 presents the fractal plot of TRM in soil, constructed by plotting the Hg concentration against its frequency in a scatter diagram. The fractal dimension (D) of the regression model was determined using the least squares method. From the figure, it can be observed that: (1) The TRM in the soil exhibits a multifractal nature, with fractal dimensions D₁ = 0.292, D₂ = 1.537, D₃ = 2.260, and D₄ = 2.835 (Table 2). The lower fractal dimension in the first layer indicates a region with lower values near the detection limit, reflecting the geochemical distribution of the original TRM in the soil. (2) The fractal dimension of the second layer is also relatively low, indicating a narrow range of concentrations coupled with a higher frequency, which reflects the background levels of TRM in the soil. (3) The third layer exhibits a larger fractal dimension, with an expanded concentration range and a lower frequency. This indicates the presence of anomalies in TRM, reflecting the geochemical distribution associated with gas migration. (4) The fourth layer has the highest fractal dimension, the largest concentration range, and the lowest frequency. This suggests the presence of localized high anomaly values in TRM, reflecting the geochemical distribution characteristics associated with leakage along fractures in the soil. (5) The multifractal distribution of TRM in the Juhugeng area not only reveals the distribution patterns influenced by various geological processes at different levels, but also establishes the lower limit for elemental anomalies. The lower limit determined through this fractal analysis is 39.24 ngg⁻¹ (Table 2). (6) The fractal dimensions D1, D2, and D3 of methane are 0.169, 0.847, and 0.809, respectively. Methane is multifractal and the fractal dimension is comparatively small, which indicates that various geochemical processes such as microleakage and microbial oxidation occur during the vertical migration of methane.

In permafrost regions, NGH can be divided into four distinct zones: the source rock layer, the NGH layer, the permafrost layer, and the seasonal permafrost layer. The source rock layer is rich in organic matter and serves as a stable source of methane for the formation of NGH. The hydrocarbons in the source rocks migrate upward, providing the necessary methane for hydrate formation. The NGH layer refers to the stratum where NGH are stored. Hydrocarbons migrating upward from deeper oil and gas reservoirs undergo geochemical differentiation, with some forming NGH. Hg ions, with their small ionic radius and high geochemical activity, can easily penetrate overlying rock and permafrost layers. This results in the formation of geochemical anomalies above the hydrate deposits, as Hg accumulates in the overlying layers. The permafrost layer extends from the surface’s seasonally thawed layer to the base of the permafrost. It remains frozen for extended periods, which suppresses conventional geochemical reactions. However, soil gas migration remains active, allowing Hg and other trace elements to move upward with the gas flow and accumulate near the surface. The seasonal thaw layer refers to the near-surface zone that undergoes periodic freezing and thawing. Within this layer, soil organic carbon and minerals such as quartz, feldspar, illite, and kaolinite actively absorb Hg migrating from deeper layers.

The geogas migration mechanism of TRM is intricate. It involves factors such as the amount of Hg absorbed during the formation of the hydrate, the vertical migration of hydrocarbons through the permafrost, and the infiltration and enrichment of Hg. Compared to the surrounding environment, Hg concentrations in NGH deposits show noticeable differences. By integrating this data with other geophysical and geochemical exploration methods, it becomes possible to delineate and predict favorable areas for NGH exploration.