Helium (He) and neon (Ne) are inert noble gases whose isotopic compositions in various reservoirs make them effective geological tracers for mantle-derived fluids. Among the eight isotopic forms of helium, ³He and ⁴He are stable, while ⁵He through ¹⁰He are unstable. The ratios ⁴He/²⁰Ne and ³He/⁴He are commonly employed to differentiate crustal from mantle-derived fluids (Sano and Wakita, 1985; Shao et al., 2024). The R/Ra ratio, representing He isotopic characteristics, is defined as the ratio of ³He/⁴He in a sample relative to that in the atmosphere.

In Earth’s atmosphere, He is predominantly composed of ⁴He, which constitutes ∼99.99986% of atmospheric He. The concentration of He in the atmosphere is relatively low, at 5.239 ± 0.004 ppm (Walia et al., 2010). The atmospheric ³He/⁴He (Ra) value is 1.4×10⁻⁶, and the ⁴He/²⁰Ne value is ∼0.318 (Sano and Wakita, 1985). Most atmospheric ⁴He is radiogenic, originating from the α-decay of radioactive isotopes such as ²³⁸U, ²³⁵U, and ²³²Th (Figure 1). The He abundance in Earth’s crust is estimated at ∼ 5.5 × 10⁻⁷%. Crustal He typically exhibits an R/Ra value of ∼0.02 and a ⁴He/²⁰Ne value of 1,000 (Andrews, 1985). Conversely, the ³He isotope, thought to originate from the solar nebula or solar wind radiation present during Earth’s formation, has accumulated in the mantle throughout Earth’s history. Mantle-derived He generally displays an R/Ra value exceeding 5 (Lupton, 1983), with mid-ocean ridge basalt (MORB) inclusions showing an R/Ra value of 8.0 and a ⁴He/²⁰Ne value of 1,000 (Graham, 2002). R/Ra values between 5 and 50 are indicative of He from the lower mantle (White, 1957). The highest recorded R/Ra value of 67.2 ± 1.8 was found in olivine from 62 Ma-old lava flows on Baffin Island, suggesting a possible origin from Earth’s core (Horton et al., 2023). Due to He’s chemical inertness, stable physical properties, and low solubility in water, gases such as N₂ and CO₂, along with groundwater, often act as carriers for He migration (Hong et al., 2010; Walia et al., 2010; Lee et al., 2019). He typically accumulates in sedimentary basins and is released to the surface via faults or fractures (Gao et al., 2024).

Radon (Rn) is the only naturally occurring radioactive noble gas, existing in 34 unstable isotopic forms, ranging from ²¹⁵Rn to ²⁴²Rn. In nature, radon is found primarily in three isotopes: ²¹⁹Rn (with a half-life of 3.96 s), ²²⁰Rn (half-life of 55 s), and ²²²Rn (half-life of 3.82 days) (Audi et al., 2003). Of these, ²²²Rn is a decay product of ²²⁶Ra in the ²³⁸U decay chain (Figure 1), with the longest half-life, and its concentration in the atmosphere is typically ranging from 10 to 100 Bq·m^-3 (Porstendörfer, 1994).

Uranium (U) and radium (Ra), naturally occurring radioactive elements, are widely distributed across the lithosphere, hydrosphere, and atmosphere. Uranium, which has 28 unstable isotopes (²¹⁵U to ²⁴²U), is found in concentrations of ∼3 ×10⁻⁴% in the lithosphere and ∼1 × 10⁻⁴% in soil. Radium, with 33 unstable isotopes (²⁰²Ra to ²³⁴Ra), has lithospheric and soil concentrations of ∼1 × 10⁻⁴% and ∼8×10⁻¹¹%, respectively (Cheng et al., 2005). The levels of U and Ra in soil or rock directly influence Rn release in soil gas (Pereira et al., 2017). Experimental studies on rock gas emissions have demonstrated that granite, which is rich in U and Ra, releases significantly higher Rn concentrations than limestone or sandstone (King, 1978; El-Arabi et al., 2006). Consequently, regions with extensive granite outcrops typically exhibit elevated Rn levels (Pereira et al., 2017).

Within mineral particles, radium undergoes α-decay, releasing α-particles (⁴He) and enabling Rn to escape. The fraction of Rn atoms generated from the decay of ²²⁶Ra that escape into rock pores is defined as the Rn emanation coefficient (Martinelli et al., 1995; Miklyaev et al., 2020; Phong Thu et al., 2020). Rn recoil can take three paths: 1) remaining within the same particle, 2) passing through a pore and embedding in adjacent particles, or 3) escaping into an open pore (Sakoda et al., 2011). Only Rn escaping into pore space is considered emanated (A, B, E, and F in Figure 2); otherwise, it is non-emanated (C, D, and G). The recoil range of Rn is 77 nm in water and 53 mm in the atmosphere, with the latter being 688 times greater (Sakoda et al., 2011). This difference indicates that rainfall and moisture content can significantly impact Rn diffusion.

Gas transport through porous media often occurs via two primary processes: diffusion and convection. Diffusion, driven by concentration gradients, involves the movement of substances from areas of high concentration to low concentration due to random molecular motion (Flügge and Zimens, 1939). Convection, also known as advection, mass transport, or viscous flow, is driven by pressure gradients (Ciotoli et al., 2007). In natural environments, gas transport typically results from a combination of these two mechanisms.

Due to Rn’s relatively large atomic mass and chemical inertness, deep-source gases such as CO₂, N₂, and CH₄ often serve as carrier gases that facilitate its migration to the surface (Yuce et al., 2017). CO₂, the most prevalent component of Earth’s interior, frequently acts as the carrier gas for Rn as it migrates along fault zones. Consequently, increased soil gas Rn concentrations are often observed in conjunction with rising CO₂ levels in fault zones (Li et al., 2013). In rock fractures and pores, typically ranging from 10^–2 to 10¹ mm in size at depths of several hundred to several thousand meters (Etiope and Martinelli, 2002; Girault and Perrier, 2014), Rn convection velocities can reach up to 10⁰ to 10⁴ m·d^-1(Etiope and Martinelli, 2002; Muto et al., 2021). For example, convection velocities of Rn in the Osaka Basin, Baikal Rift, and North Caucasus are estimated at 340 m·d^-1, 5.2 m·d^-1, and 28 m·d^-1, respectively (Miklyaev et al., 2020; Muto et al., 2021). When CO₂ acts as the carrier gas, Rn may originate from depths of several hundred to several thousand meters in areas of high permeability (Girault and Perrier, 2014). Moreover, groundwater transport and deposition also contribute to the movement of Rn’s parent elements, uranium, and radium (Chen et al., 2018).

Data from the Mauna Loa Observatory in Hawaii proves that the CO₂ concentration in the atmosphere continues to increase, rising from 315.70 ppm in March 1958 to 422.80 ppm on 5 February 2024 (http://www.co2.earth). CO₂ primarily originates from three sources: the decomposition of organic material, the breakdown of carbonate rocks, and mantle degassing (Barnes et al., 1978). The origin of CO₂ can generally be determined using δ¹³C_CO2 values and CO₂ concentrations, which identify three distinct end-member sources: 1) Deep-source CO₂, derived from magmatic degassing and the decarbonation of carbonate rocks, typically exhibits concentrations near 100% with δ¹³C_CO2 values ∼0‰ (Parks et al., 2013); 2) Biogenic CO₂, usually characterized by concentrations of ∼4% and δ¹³C_CO2 values ∼ −23‰ (Di Martino et al., 2016); and 3) Atmospheric CO₂, currently at 422.80 ppm, with δ¹³C_CO2 values ∼ −8‰ (Keeling et al., 2005).

The range of δ¹³C_CO2 from different sources can overlap each other, such as those from Mid-Ocean Ridge Basalts (MORB) and carbonate rocks (Bergfeld et al., 2001). CO₂ also serves as the primary carrier gas for He migration in the crust (Hong et al., 2010; Walia et al., 2010; Lee et al., 2019). Therefore, the He-CO₂ system is often utilized to further deduce the source of CO₂ (Tian et al., 2021; Shao et al., 2024). Analysis of CO₂ origins can be conducted using R/Ra ratios and δ¹³C_CO2 values, which help distinguish between contributions from organic material, carbonate rock metamorphism, and mantle magma degassing (Barnes et al., 1978).

The decomposition of carbonate rocks involves processes such as water-rock interactions, mechanical grinding by faults, thermal metamorphism, and weathering (Rovira and Vallejo, 2008; Tamir et al., 2011). These processes can lead to the release of substantial amounts of CO₂, which then becomes a crustal fluid, potentially contaminating mantle-derived volatiles. Typically, thermal metamorphism of carbonate rocks occurs ∼400°C; however, CO₂ release can begin at temperatures above 70°C when water is involved (Pankina G et al., 1979). Extensive fracture networks and fluid interactions can enhance water-rock reactions within the rock, producing significant quantities of CO₂ (Randazzo et al., 2021). Additionally, in regions of significant tectonic uplift, carbon stored in carbonate rocks for millions of years can be released through weathering (Zondervan et al., 2023).

The exploration of soil gases, referred to as “geogas”, dates back to 1913 (Klusman, 1993). Globally, regions of strong gas release often overlap with tectonic suture zones, volcanic belts, geothermal areas, and seismic zones (Barnes et al., 1978; Tamburello et al., 2018). Regionally, the intensity of fluid release and the geochemical characteristics within fault zones are closely related to fault activity. Significant anomalies in soil gas concentrations (such as Rn, CO₂, He, H₂, and CH₄) have been observed in various fault zones, including the Stivos Fault in Greece (Papastefanou, 2010), the Khlong Marui Fault in Thailand (Bhongsuwan et al., 2011), the Kütahya Simav Fault in Turkey (Manisa et al., 2022), and the Mat Fault in India (Jaishi et al., 2014). Field observations suggest that stronger fault activity correlates with increased soil gas release, making soil gas concentrations a useful metric for assessing fault activity (Seminsky et al., 2013; Capaccioni et al., 2015). Additionally, different fault types (normal, reverse, and strike-slip) exhibit distinct concentrations and flux characteristics (Annunziatellis et al., 2008; Sun et al., 2018). Therefore, tectonic zones with significant gas release are valuable for reconstructing regional geodynamic processes and monitoring subsurface tectonic activity (Faulkner et al., 2010; Tian et al., 2021; Li et al., 2023).

At a global scale, crustal permeability exhibits significant stratification, influenced by both internal and external forces. In the deeper crust, internal processes such as metamorphism and magmatism are dominant, while in the shallow crust, external factors, particularly the hydrologic cycle, play a more crucial role in shaping permeability (Rojstaczer et al., 2008). The difference in permeability of the crust determines the different distribution patterns of fluids underground. Rock deformation experiments indicate that when differential stress exceeds rock shear strength, pre-existing fractures close, forming new microcracks and pores. Continued stress can link these microcracks into macroscopic fractures, providing new pathways for fluid migration (Tuccimei et al., 2010). Under tectonic stress, the number of microcracks in fault zones increases (Li et al., 2013; Hansberry et al., 2021), accelerating the migration and release of deep gases, which can cause anomalies in gas concentrations and fluxes in shallow soils (Martinelli, 2020; Miklyaev et al., 2020). Research has shown that high sliding rates increase the permeability of sandstone and granite by three orders of magnitude, indicating that high sliding rates can sustain high permeability in fault zones (Tanikawa et al., 2010). Consequently, variations in soil gas release are primarily influenced by changes in fault zone permeability.

Active faults and fractures generally exhibit higher permeability and porosity than surrounding hard rock, resulting in greater deep-sourced gas release in fault zones compared to non-active tectonic areas (Annunziatellis et al., 2008; Giammanco et al., 2009; Weinlich, 2014; Voltattorni et al., 2015; Singh et al., 2016; Bond et al., 2017). In regions outside fault zones with lower permeability, the correlation between Rn and CO₂ concentrations is weak. In contrast, well-connected faults show a stronger positive correlation between Rn and CO₂ (Padrón et al., 2013; Ciotoli et al., 2014). Extensional structures with high permeability are more conducive to deep fluid release than thrust or strike-slip faults, with the scale of extensional faults directly influencing CO₂ emissions (Tamburello et al., 2018). For example, CO₂ emissions from the East African Rift are ∼71 Mt·yr^-1 (Lee et al., 2016), from active rifts ∼40 Mt·yr^-1 (Brune et al., 2017), and from the eastern Ethiopian Rift ∼20 Mt·yr^-1 (Hunt et al., 2017) (Figure 3). Although active faults are key pathways for the release of mantle-derived and crust-derived gases (Caracausi et al., 2022), atmospheric gases can also enter the Earth’s interior through high-permeability fractures, with diffusion rates reaching 10 m·d^-1 and maximum depths of 300 m (Arai et al., 2001; Giammanco et al., 2009). Additionally, thick sedimentary layers can obstruct gas migration, influencing atmospheric mixing and the release of deep-sourced gases, while shallow organic gases may mix with rising fluids (Liu, 2006). Therefore, the connectivity of fault zones significantly affects underground gas release, with surface gases reflecting a mix of various sources.

Deep and large active fault zones act as links across different Earth layers. Stable isotopes of deep fluids may undergo equilibrium or kinetic fractionation during geological processes, and fluid isotope tracers can provide important information on fluid sources and migration in active fault zones (Zheng et al., 2013; Zhang et al., 2021). For instance, (Hernández Perez et al., 2003) identified mantle-derived CO₂ in soil gases of the Hakkoda Fault zone in northern Japan, with a contribution of up to 6.7%. Kulongoski et al. (2013) detected high ³He/⁴He ratios and CO₂ concentrations in hot spring gases from the San Andreas Fault zone, with mantle-derived He contributing up to 44%. Shao et al. (2024) analyzed hot spring gases in the southern segment of the eastern boundary of the Sichuan-Yunnan rhombic block, finding no intersection between the Red River Fault and Xiaojiang Fault. Zhang et al. (2021) studied the southeastern margin of the Tibetan Plateau using the He-CO₂-N₂ system in hydrothermal fluids, finding that He isotopes provided evidence for the lateral expansion and localized surface uplift of the Tibetan Plateau. These studies demonstrate that surface-emitted gases and isotopes in hot springs or soil gases are effective indicators of tectonic activity and fluid dynamics.

Stress changes induced by earthquakes can trigger variations in pore pressure and the number of micro-cracks within fault zones, affecting the interaction between fluids and rocks and altering the release of deep gases at the surface (Camarda et al., 2016; Randazzo et al., 2021; Zhao et al., 2021; Caracausi et al., 2022). These processes can enhance fluid migration along active faults and modify the contribution of different fluid sources to soil gases and hot spring emissions, leading to observable pre-seismic anomalies or post-seismic responses (Martinelli and Dadomo, 2017). Between 1967 and 2014, analysis of 134 global seismic cases revealed that 69% showed anomalies in soil and groundwater Rn, 20% in geochemical parameters of soil and groundwater gases, and 10% in physical groundwater parameters (Woith, 2015) (Figure 4).

Recent studies have increasingly applied geochemical methods for analyzing soil gases to understand seismic activity trends and to develop earthquake monitoring and prediction theories. In tectonically active regions, stress accumulation from seismic activity enhances the release of deep-sourced gases like Rn, CO₂, and He, which accumulate in rock fractures along fault zones (Ciotoli et al., 2014; Yuce et al., 2017; Chen et al., 2015). The correlation between Rn and CO₂ concentrations tends to increase before earthquakes (Fu et al., 2017). The vibroseis truck (Gresse et al., 2016) and active seismic source (Liu et al., 2023b) experiments have demonstrated that seismic waves can boost the release of gases trapped in rock and soil pores. Moreover, low-magnitude earthquakes (M < 4) can release crustal He into the atmosphere, with the He release amount being quantitatively related to the fault zone volume (Caracausi et al., 2022). Periodic monitoring of soil gases in Italy’s Emilia region revealed significant increases in CO₂, CH₄, and H₂ concentrations before and after the 2012 Emilia-Romagna earthquake swarm (Sciarra et al., 2017). In Gujarat, India, continuous Rn monitoring successfully detected significant increases in Rn concentrations days to weeks before four earthquakes with magnitudes ranging from 4.0 to 4.1 (Sahoo et al., 2020; Torkar et al., 2010) used soil gas Rn to predict 10 out of 13 earthquake events using an artificial neural network with a backpropagation algorithm. These findings highlight that seismic activity induces the release of deep-sourced gases along fault zones, leading to changes in soil gas concentrations that can serve as indicators for seismic activity and earthquake monitoring.

Hot spring gas geochemistry also shows potential as an indicator of seismic activity. Before the 2008 Tibet M 6.3 earthquake in China, significant anomalies in He and Rn concentrations were observed in hot springs at Bakreswar and Tatta Pani in India (Chaudhuri et al., 2011). Prior to the 1955 Kobe M _W 6.9 earthquake in Japan, Rn release rates in groundwater and atmospheric Rn concentrations significantly increased, correlating with crustal strain fluctuations (Yasuoka et al., 2009). During the 2016 Kumamoto M 7.3 earthquake in Japan, He concentration changes in deep groundwater correlated with volumetric strain changes (Sano et al., 2016). Thus, hot spring gas concentrations can be crucial for earthquake monitoring.

Throughout different stages of earthquake preparation and occurrence, the contribution of deep-sourced and shallow-sourced fluids dynamically evolves. For instance, before and after the 2011 Van M _W 7.2 earthquake in Turkey (Aydın et al., 2015) and the 2013 Lushan M _S 7.0 earthquake in China (Chen et al., 2015), significant increases in ³He/⁴He and δ¹³C_CO2 values were observed in hot spring gases in fault zones. As aftershock activity waned, the supply of mantle-derived gases decreased, leading to a decline in ³He/⁴He and δ¹³C_CO2 values. Following the 2008 Iwate-Miyagi M 7.2 earthquake in Japan, the ascent of mantle-derived fluids caused a maximum 85% increase in the ³He/⁴He value in hot spring gases near the epicenter within a week (Horiguchi and Matsuda, 2008). After 2 M 6.0 earthquakes in the Emilia, Italy in 2012, the δ¹³C_CO2 and δ¹³C_CH4 values of gases released from fault zones in the epicentral area significantly decreased, likely due to the seismic-induced release of shallow biogenic CH₄ and CO₂, overshadowing deep thermogenic gases (Sciarra et al., 2017). These changes in He and C isotopes in hot spring gases near fault zones before and after earthquakes underscore how seismic activity promotes the mixing of gases from various sources, particularly the ascent of mantle-derived fluids.

Atmospheric gas variations induced by seismic activity are integral to understanding the lithosphere-atmosphere coupling mechanism (Veefkind et al., 2012; Jing et al., 2019). Advances in hyperspectral sensors with atmospheric detection capabilities have enabled extensive studies on gas changes associated with seismic and volcanic events (Tramutoli et al., 2013) (Figure 5). Notable anomalies in gases such as CH₄, CO, CO₂ and O₃ have been documented before and after significant earthquakes, such as the 2004 Sumatra-Andaman M _W 9.1 earthquake and the 2005 Sumatra-Nias M _W 8.6 earthquake (Cui et al., 2023), the 2008 Wenchuan M _S 8.0 earthquake and 2013 Lushan M _S 7.0 earthquake in China (Cui et al., 2017), and the 2015 Gorkha M 7.8 earthquake and Dolakha M 7.3 earthquake in Nepal (Jing et al., 2019). Furthermore, a statistical analysis using the Adaboost machine learning algorithm examined infrared and hyperspectral gas parameters among 10 different variables before and after 1,371 global earthquakes of magnitude ≥6 from 2006 to 2013, identifying O₃ and CO₂ as significant contributors to earthquake prediction (Xiong et al., 2021).