Advances in Geochemical Monitoring Technologies for CO₂ Geological Storage

Abstract

CO₂ geological storage, as a large-scale, low-cost, carbon reduction technology, has garnered widespread attention due to its safety. Monitoring potential leaks is critical to ensuring the safety of the carbon storage system. Geochemical monitoring employs methods such as gas monitoring, groundwater monitoring, tracer monitoring, and isotope monitoring to analyze the reservoir’s storage state and secondary changes after a CO₂ injection. This paper summarizes the recent applications and limitations of geochemical monitoring technologies in CO₂ geological storage. In gas monitoring, the combined monitoring of multiple surface gasses can analyze potential gas sources in the storage area. In water monitoring, pH and conductivity measurements are the most direct, while ion composition monitoring methods are emerging. In tracer monitoring, although artificial tracers are effective, the environmental compatibility of natural tracers provides them with greater development potential. In isotope monitoring, C and O isotopes can effectively reveal gas sources. Future CO₂ geological storage project monitoring should integrate various monitoring methods to comprehensively assess the risk and sources of CO₂ leakage. The incorporation of artificial intelligence, machine learning technologies, and IoT monitoring will significantly enhance the accuracy and intelligence of numerical simulations and baseline monitoring, ensuring the long-term safety and sustainability of CO₂ geological storage projects.

1. Introduction

The rapid warming resulting from human CO₂ emissions has disrupted the Earth’s climate regulation system, causing significant changes to the global climate and ecosystems. Controlling CO₂ emissions has become an urgent necessity for all countries [1]. CO₂ geological sequestration is currently the most significant form of artificial carbon sink and is a critical technology and safeguard to control global warming within 2 °C [2]. According to predictions, by 2030, the global average annual emission reduction through carbon capture and storage (CCS) technology needs to reach 490 million tons; by 2050, it must reach 4.66 billion tons [2,3]. However, there is a risk of leakage after CO₂ is injected into underground reservoirs. The stored CO₂ can increase the pore pressure in reservoir rocks, altering the original temperature, fluid pressure, and stress state, potentially causing surface uplift, fault activation, and CO₂ leakage. Additionally, the stored CO₂ may react chemically with reservoir minerals and groundwater, causing groundwater acidification and mineral dissolution, which compromises geological stability [4]. The leakage of CO₂ from the reservoir can severely impact the ecological environment and human health [5]. Therefore, reducing the leakage risk is crucial for promoting CCS projects, and the continuous monitoring of CO₂ geological sequestration sites is essential for ensuring the effectiveness and safety of sequestration.

Considering the movement, migration state, and phase changes in underground CO₂, monitoring must be dynamic. This requires monitoring methods that not only provide continuous surveillance but also offer timely feedback and warnings [6]. Conventional geophysical monitoring methods have limitations in terms of their monitoring frequency and cost, which cannot meet the demands of long-term, low-cost monitoring [7,8]. Geochemical monitoring methods, by analyzing the chemical composition of gasses and water, tracers, and isotopes, can detect subtle geochemical changes or early warning signs of leakage. These methods help analyze the CO₂ sequestration state, migration paths, and leakage scenarios [9], providing valuable information on the sequestration state and dynamic changes in CO₂. Moreover, geochemical monitoring is relatively cost-effective, and the collected data can inform numerical simulations, assessing the model’s accuracy and reliability [10,11]. Recently, with the extensive implementation of CO₂ geological sequestration projects, geochemical monitoring has become a standard part of monitoring due to its effectiveness and low cost [12,13,14,15].

This paper systematically reviews the principles and application progress of various geochemical monitoring technologies and their effectiveness in CO₂ geological sequestration projects. However, with advancements in information technology, traditional methods require upgrades to meet new demands. Future research should focus on developing low-cost, routine geochemical monitoring, leveraging extensive existing data, and integrating the Internet of Things, artificial intelligence, and machine learning technologies to create predictive models for early warning and risk assessment. These findings will provide a reference for the future development of geochemical monitoring methods.

2. Overview of Geochemical Monitoring Techniques

Based on monitoring principles, geochemical monitoring techniques can be divided into direct and indirect types. Direct monitoring involves measuring changes in the chemical composition or concentration of water and gasses, directly reflecting changes in the underground environment, and the geochemical reactions and impacts during CO₂ sequestration. For example, changes in CO₂ concentration in groundwater can be monitored to assess CO₂ migration and the effectiveness of sequestration. Indirect monitoring involves analyzing parameters such as the composition and distribution patterns of tracers or isotopes to infer or simulate information about the underground environment and the CO₂ migration process. For instance, groundwater flow models and geochemical simulations can predict CO₂ migration paths and underground reactions [16].

2.1. Gas Monitoring

Gas monitoring determines changes in the underground environment by measuring changes in the components and concentrations of gasses [17]. Techniques such as Infrared Gas Analyzer [18], Long Open Path IR [19], Eddy Covariance [20], Accumulated Chamber [21], and Light Detection and Ranging [22] are commonly used to monitor CO₂ concentrations. Currently, monitoring systems integrating soil, surface, and low-altitude CO₂ are used to analyze changes in the composition and concentration of gasses, providing information on the CO₂ leakage pathways, diffusion range, and leakage scale [23]. However, due to the relatively high and fluctuating atmospheric CO₂ concentrations [24], trace amounts of CO₂ leakage in carbon sequestration projects may be obscured by background fluctuations. Additionally, geological sequestration areas are typically large and remote, limiting gas monitoring techniques’ stability, monitoring accuracy, and frequency, which may hinder the timely detection of leakages. To mitigate the impact of environmental changes, current gas monitoring research focuses on two aspects:

• Enhancing the accuracy of CO₂ concentration monitoring through sensor upgrades or data calibration. Sensors are critical components when collecting environmental parameters. In recent years, CO₂ concentration monitoring equipment such as Fourier Transform Infrared Spectrometer (FT-IR) and Photoacoustic Spectroscopy (PAS) sensors have developed rapidly. FT-IR, combined with a long-path gas absorption cell, can improve detection sensitivity to the ppb level, meeting monitoring needs in complex environments such as high-temperature and high-humidity environments [25,26,27]. PAS can monitor multi-component mixed gasses with high sensitivity, making it ideal for online CO₂ monitoring, although the stability of PAS still needs to be improved [28,29]. Since the precision components of sensors are susceptible to environmental influences (Figure 1a) [30,31,32], designs typically include temperature, humidity, and pressure compensation mechanisms or algorithms to correct the initial measurements and enhance the credibility of the data [30,31]. By integrating measurements from different sensors, reliable data under specific environmental conditions can be output based on algorithms, collectively representing the CO₂ concentration changes in the region.
• The coordinated monitoring of multiple gasses to indirectly reflect gas leakages. During biological photosynthesis and respiration, changes in O₂ and CO₂ concentrations have a good linear relationship, so the ratio of O₂ to CO₂ concentration changes can be used to determine whether CO₂ leakage has occurred. If CO₂ leakage occurs, there will be a significant abrupt leakage signal (Figure 1b) [33]. However, factors such as water–rock–CO₂ interactions, methane oxidation, rock weathering, and groundwater flow can generate or consume CO₂ or O₂ [34,35]. Therefore, monitoring changes in the composition and concentration of multiple gasses, such as CO₂, O₂, N₂, CH₄, Ar, and He, is needed to assist in analyses of the CO₂ source and corresponding geochemical processes (Figure 1c) [36]; this can reduce the impact of environmental background changes and achieve effective monitoring [37].

The development direction of gas monitoring is towards refinement, providing quantitative change trends. However, high-sensitivity sensors are expensive and have poor environmental tolerance [38]. Therefore, the future mainstream direction of gas monitoring should involve calibrating low-cost sensor monitoring data and the coordinated monitoring of multiple gasses.

Figure 1. (a) Temperature correction diagram of the infrared spectrum sensor (modified from [30]); (b) the variations in the relative concentration of CO₂ and O₂ (modified from [33]); (c) processes defined by relationships between O₂ and CO₂ [36].

Figure 1. (a) Temperature correction diagram of the infrared spectrum sensor (modified from [30]); (b) the variations in the relative concentration of CO₂ and O₂ (modified from [33]); (c) processes defined by relationships between O₂ and CO₂ [36].

2.2. Water Monitoring

Water monitoring assesses CO₂ leakage and its scale by monitoring changes in the chemical composition, isotope ratios, and hydrogeological characteristics of groundwater and surface water. Overall, research on water monitoring has primarily focused on the direct impacts of CO₂ leakage and its secondary changes. The direct impacts of CO₂ leakage include a decrease in pH and an increase in the partial pressure of CO₂ in water [39]. A significant decrease in water pH is one of the best diagnostic indicators of CO₂ leakage [16]. Even a small amount of CO₂ leakage into the water can cause the pH of the regional water to drop rapidly, by 1–2 [40,41]. Similar to gas monitoring, the composition and concentration of dissolved gasses in water may change when CO₂ leakage occurs. For example, the concentration of dissolved CO₂ increases, which may affect the concentrations of other dissolved gasses, such as N₂, O₂, and CH₄. When CO₂ dissolves, other dissolved gasses in the water may be released, causing changes in the concentration and composition of gasses in the water, indirectly indicating whether CO₂ leakage has occurred [42]. Dissolved gas analysis can also provide information about the redox state of the underground environment. An increase in CH₄ concentration might indicate microbial activity or other geochemical processes related to CO₂ leakage [43,44]. The composition of dissolved gasses in water is also affected by environmental factors such as temperature and pressure, and requires a comprehensive comparison. Overall, monitoring the partial pressure of pH and CO₂ is simple and highly accurate, and has good application prospects for determining the location and range of underground CO₂ plumes and analyzing leakage pathways [39].

Monitoring the changes in ion composition due to pH changes is another mainstream direction in water monitoring. Since the water in carbon sequestration areas has been in long-term chemical equilibrium with the surrounding rocks, it is particularly sensitive to local equilibrium changes caused by the addition of large amounts of CO₂. A decrease in water pH can dissolve surrounding rock or reservoir minerals, leading to changes in the ion composition and concentration, electrical conductivity, and redox properties of the water [45,46]. Numerous studies have examined the changes in ion composition caused by the dissolution of surrounding rock minerals following the intrusion of CO₂ into groundwater. The research indicates that the dissolution of carbonate rocks can enrich elements such as Ca and Mg [40,47,48], while the dissolution of silicate rocks can release additional Si and Fe [49]. Other elements, such as Al, Si, Zn, Ni, Ba, As, U, Mn, and Sr, may also increase in concentration [16,40,41,50,51]. Recent studies suggest that rare earth elements can reflect the interaction between groundwater and surrounding rock caused by CO₂ injection at the nanoscale. In CO₂ leakage areas, heavy rare earth elements are more enriched compared to light rare earth elements, and the La/Yb and Y/Ho ratios change significantly after CO₂ disturbance [40]. Do et al. [39] summarized the trends in the changes in certain elements after CO₂ injection and classified them into three categories (

Table 1. Coefficient of the correlation (r) between log pCO₂ and hydrogeochemical parameters (adapted from [39]).

Period	Pre-Injection	Syn-Injection	Post-Injection
pH
Type 1
Type 2
Type 3

): (1) pulse type, where ion concentrations rise rapidly when the CO₂ plume arrives and decrease with continuous CO₂ injection, such as HCO₃, Ca, Mg, Na, K, Sr, and Ba; (2) delayed type, where ion concentrations in the water rise and fall relatively slowly, reaching a maximum at the end of the CO₂ injection, such as SiO₂ and Mn; (3) ion concentrations in the water rise quickly but fall slowly, such as Li. Changes in ion concentrations in the water are influenced by the scale of leakage, the properties of the surrounding rocks, and the buffering and clearing capacities of the underground aquifer [50,52]. Mechanisms for ion changes may include rock dissolution, ion exchange, the precipitation of secondary minerals, adsorption/desorption, and redox processes [40,53].

Overall, water monitoring is cost-effective and efficient. Combining water monitoring with geophysical methods, such as Electrical Resistance Tomography, holds significant potential [54].

2.3. Tracer Monitoring

Tracer monitoring involves the use of artificially injected compounds or markers to trace the migration and distribution of CO₂ in underground reservoirs. Tracers are typically compounds with specific chemical properties or isotopic compositions that distinguish them from natural substances in the underground environment. This technology, widely employed in the oil, gas, and geothermal industries, provides insights into reservoir connectivity and flow pathways, and estimates of residual oil or native water saturation [13]. Tracers are valuable tools in the field of CO₂ geological storage, offering low-cost solutions with high informational returns [4].

Tracers are categorized into chemical and natural types. Chemical tracers, such as sulfur hexafluoride (SF₆), perfluorocarbons (PFCs), and poly-fluoroalkyl substances, are synthetically produced compounds known for their low natural background concentrations and compatibility with CO₂ plumes, making them viable for numerous storage projects. SF₆, which has been widely used in meteorology and electrical industries for nearly 60 years [55], maintains a natural atmospheric concentration of approximately 1.0 × 10⁻¹¹ (v/v) with a detection limit as low as 1.0 × 10⁻¹⁴ (v/v) [56]. Under various environmental conditions, about 300 kg of SF₆ was added to an injected CO₂ mixture of 100,000 tons the Otway site, Australia [57], while approximately 2 tons of SF₆ were added to 300,000 tons of injected CO₂ at the Shenhua storage site, China [58]. PFCs, which have a relatively low global warming potential and pose a minimal environmental hazard, have seen extensive use in regions like InSalah and the North Sea [59,60]. However, the environmental risks, high equipment costs, and challenges in terms of sustained field monitoring limit the widespread use of chemical tracers [13].

In contrast, natural tracers offer superior environmental compatibility, and naturally occurring elements are better suited as in situ tracers [39,46,61]. The injection of CO₂ disrupts the geochemical equilibrium of the subsurface, causing changes in the chemical and/or isotopic signatures of specific fluid endmembers [61]. Metal elements, which migrate and transform in response to CO₂ plumes or acidic waters, can serve as effective natural tracers for studying underground CO₂ migration [46,62]. Joun et al. [63] improved pCO₂ measurement systems and achieved recovery rates of 26%, 85%, and 95% for SF₆, Kr, and U, respectively, demonstrating that uranium (U) is a particularly effective and conservative tracer. Lithium (Li), which shows a strong correlation with water CO₂ at partial pressure, is useful for monitoring CO₂ migration, pH, and conductivity changes in aquifers [39].

Additionally, rare gasses are commonly employed as natural tracers. These gasses are chemically stable, have low background concentrations, and migrate quickly, meaning that they are present in trace amounts in both natural and anthropogenic CO₂ sources [14]. By analyzing the types and isotopic compositions of rare gasses in reservoirs, one can gain insights into CO₂ migration paths, leakage, and groundwater flow. Rare gasses, such as helium (He), argon (Ar), and krypton (Kr), have been used effectively as tracers in various storage sites, including Cranfield and Otway, to reveal the extent of CO₂ dissolution and assess changes in CO₂ displacement efficiency [14,15]. Kr has historically been an early indicator of CO₂ migration, while He and Ar are released in the initial stages of CO₂ migration [64]. In experiments involving the injection of Kr and SF₆ into CO₂-containing groundwater, no gas loss to external systems was initially, although, later, the tracers were progressively diluted by groundwater (Figure 2) [63].

Overall, chemical tracers are a cost-effective and efficient monitoring method, but natural tracers with high environmental affinity are more likely to receive application approval. Tracer monitoring typically requires large-scale mass spectrometers or chromatographs, which are expensive and limit the feasibility of continuous monitoring. Natural tracers have multiple sources in nature, and there may be a natural accumulation of methane and rare gasses in the sequestration area [65], which can impact the efficiency of tracer monitoring due to leakage and site-specific conditions [66]. In the short term, chemical tracers remain indispensable for control purposes. For instance, in experiments conducted in the North Sea, Roberts et al. [13] found that tracers like SF₆ and ¹⁴C are not viable monitoring methods due to their environmental impacts and cost constraints. To achieve a more detailed characterization of the sequestration conditions and develop a comprehensive geochemical profile of the sequestration site, further research is needed on the dissolution and distribution mechanisms of natural tracers in CO₂ plumes. This includes understanding processes such as dissolution, residual gas capture, convective mixing, and vertical CO₂ migration.

2.4. Isotope Monitoring

Isotope monitoring involves the use of the isotopic composition of compounds to passively track and assess the status of CO₂ sequestration. Since CO₂ consists of carbon and oxygen, monitoring carbon and oxygen isotopes is the most common method. Different sources of CO₂ have varying abundances of carbon isotopes such as ¹²C, ¹³C, and ¹⁴C. For instance, CO₂ from fossil fuel combustion lacks ¹⁴C, whereas biogenic CO₂ has similar ¹⁴C levels to modern atmospheric levels [67]. Additionally, in biogenic CO₂, the ¹³C composition varies depending on the source, making carbon isotopic composition one of the most precise tools for identifying CO₂ sources [67,68]. Changes in the carbon isotopic composition of CO₂, if the initial isotopic composition is known, can be used to infer the migration path and potential leakage scale of CO₂ [69,70]. For example, in the Weyburn sequestration area, carbon isotopic analysis has been repeatedly used before and after CO₂ injection to evaluate CO₂ sources, migration paths, and the sequestration status [68,71]. This requires the carbon isotopic composition of the injected CO₂ to be significantly different from the background CO₂ sources, which typically include dissolved inorganic carbon, bedrock-derived carbon, and soil CO₂, placing high demands on the background values of the sequestration environment [69].

Oxygen isotopes, including ¹⁶O, ¹⁷O, and ¹⁸O, generally maintain a constant ratio because the oxygen content in atmospheric CO₂ is negligible compared to that in water. However, the injected supercritical CO₂ becomes a major source of oxygen in the reservoir and can reach isotopic equilibrium with groundwater oxygen within a period ranging from hours to days. This results in rapid changes in isotopic values, which can reveal the CO₂ migration processes [69]. Nonetheless, oxygen isotope monitoring is not a conservative tracer and may not accurately identify CO₂ sources due to potential groundwater influences [59,72]. Therefore, it is essential to first evaluate the original C and O isotopic composition and abundance of the target sequestration site to determine its applicability [61]. The use of only carbon or oxygen isotopes for monitoring can be limited by the conditions at the sequestration site, so a combined approach using multiple isotopes, including carbon and oxygen, is more suitable.

In addition, the isotopic composition of noble gasses can provide insights into the storage conditions. Noble gasses show a strong correlation with CO₂ [14], and the CO₂/³He ratio in natural reservoirs typically ranges from 1 × 10⁹ to 1 × 10¹⁰ [73]. A decrease in this ratio can indicate underground CO₂ leakage [17]. The extent of CO₂ dissolution can be assessed by examining changes in the isotopic composition of He. At the Cranfield storage site in the United States, the CO₂/³He ratio suggests that approximately 0.2% (around 7 kt) of the 1 million tons of injected CO₂ has dissolved into the formation water, with a pH of 5.8 [14]. Furthermore, noble gasses such as ²⁰Ne, ³⁶Ar, ⁸⁴Kr, and ¹³²Xe can be monitored alongside carbon isotopes to track CO₂ dissolution in water and its interaction with residual oil [14]. The combination of noble gas and carbon isotope geochemistry can effectively identify short-term and long-term physicochemical processes in geological reservoirs. In the short term, the mass fractionation ratio of Ne and Ar isotopes serves as a reliable leakage indicator, with higher values indicating a faster gas flow. Over the long term, noble gas and carbon isotope tracers can identify dissolution and/or precipitation processes [74]. However, the water formation also contains noble gasses, and their concentration can be influenced by CO₂–oil interactions. Thus, interpretations based on noble gas isotope ratios may affect the monitoring and tracking results of stored CO₂ recovery.

Essentially, isotope monitoring functions as a natural tracer, utilizing the inherent isotopic composition of the subsurface environment for passive tracing. Although isotope monitoring is relatively costly, it can provide substantial information. Future research should integrate various isotope types to further refine the understanding of isotopic composition and changes during the storage process, elucidate the mechanisms of these, and account for environmental variables (

Table 2. A summary of the potential tracers for CCS, and their properties, cost, and environmental impact (data from [13]).

Tracers		Cost		Environment Impact				Type
		Cost per Mt (CO2)	Logistics Cost	GWP (100 Years)	Biological Impact	Bio-Degradable	Use Restricted	In CO2 Stream?	In Storage Reservoir?
Artificial	SF6	£1~100	Acceptable	22,850	-	Uncertain	Yes	No	No
	PFCs	£1~100	Acceptable	9540	Possible	Yes	No	No	No
	CD4	£1000~10,000	Acceptable	>36	Possible	Yes	No	No	No
Natural	14C (in CO2)	£10,000~100,000	Acceptable	1	Possible	-	Yes	Yes	No
	14C/12C	£1~100	Acceptable	1	Possible	-	Yes	Yes	No
	13C/12C	£100,000~1,000,000	Restrictive	1	No	-	No	Yes	Yes
	18O	-	Restrictive	1	No	-	No	Yes	Yes
	CH4	£1000~10,000	Acceptable	36	Possible	Yes	No	No	Yes
	3He/4He	£100~1000	Acceptable	None	No	No	No	Yes	Yes
	124,129Xe/130Xe	£1000~10,000	Acceptable	None	No	No	No	Yes	Yes
	80,83,86Kr/84Kr	£100,000~1,000,000	Restrictive	None	No	No	No	Yes	Yes

).

3. Research Developments

3.1. Development of Monitoring Strategies

As previously mentioned, with the advancement of CO₂ geological storage projects, various geochemical monitoring methods have been widely applied in these projects (

Table 3. Comparison of different geochemical monitoring methods.

Method	Principle	Application	Main Mechanism Types	Leakage Scale	CO2 Sources	Leakage Path	Cycle	Accuracy	Case
gas	gas flux and composition	surface seepage or leakage spread	shallow/surface effects	direct	No	Yes	continuous/regular	normal	almost all
water	pH, ion concentration, and composition	cap integrity, plume migration	geochemical effects	indirect	No	Yes	continuous/regular	normal	InSalah, Outway, CO2SINK, Weyburn, Cranfield
noble gas	species and composition of noble gasses	plume migration, sequestration state	physical effects	indirect	Yes	Yes	regular	high	Outway, Weyburn, Cranfield
isotope	isotopic value	underground characteristics, storage state	geochemical effects	indirect	Yes	Yes	regular	high	Otway, Weyburn
tracer	the amount of tracer	plume migration, sequestration state	geochemical effects	indirect	Yes	Yes	regular	high	SECARB (SF6, (PFCs), InSalah (PFCs), Outway (CD4), Weyburn (PFCs), Shenhua (SF6)

). However, different geochemical monitoring methods vary in their mechanisms, scope, and duration. For example, geochemical monitoring mechanisms mainly include [75]: (1) physical effects (pressure effects or fluid displacement); (2) geochemical effects (the dissolution of reservoirs and cap rocks; the activation of heavy metals); and (3) shallow/surface effects (toxic compounds affecting the soil, microbial communities, and groundwater quality). Different mechanisms can lead to variations in monitoring results [15,64]. Additionally, the original geochemical properties of different storage sites can vary significantly, which may also impact the monitoring outcomes [13]. Clearly, given such complex conditions, a single monitoring method cannot meet the monitoring needs of all CO₂ geological storage sites. Currently, a combination of multiple monitoring methods is generally employed to comprehensively characterize and determine the geochemical characteristics of the region. Therefore, developing effective monitoring strategies is a key focus of current research.

The development of monitoring strategies needs to meet various requirements. First, it is necessary to consider the constraints of the geological conditions, the hydrological conditions, the injection processes, environmental pollution, and budgets, among other factors. Different geological characteristics influence the migration and geochemical processes of CO₂. For example, saline aquifers and oil and gas reservoirs, as different types of storage environments, show significant differences in their groundwater chemical composition and organic and inorganic geochemical properties. In marine carbon storage sites, monitoring methods must also account for seawater pressure, adsorption or dispersion, and environmental impacts. Terrestrial carbon storage sites are more affected by human activities, and background value changes can be more complex [13]. Additionally, specific monitoring objectives need to be achieved, such as determining the CO₂ storage volume and distribution, tracking CO₂ migration paths, identifying sources of CO₂ leakage, and determining the scale of the leakage. Finally, considering the movement, migration state, and phase changes in underground CO₂, the monitoring strategy must be dynamic. It should enable continuous monitoring and timely feedback and warnings [6]. Regular evaluations of monitoring effectiveness and storage status are also necessary, with timely adjustments to the monitoring strategy based on actual conditions to ensure the effectiveness and sustainability of the monitoring efforts [76]. For example, the adaptive monitoring strategy in the FutureGen 2.0 project continuously assesses the monitoring results and modifies the monitoring network as needed to identify the geochemical characteristics of CO₂ or brine leakage. This continuous assessment throughout the project’s design and operation phases determines the need for additional near-surface monitoring methods, such as surface aquifers, surface water, soil gas, and atmospheric monitoring [8].

To meet the requirements of a monitoring strategy, almost all current CO₂ geological storage projects employ multiple geochemical monitoring methods, including gas concentration, pH value, ion composition, chemical tracers, noble gasses, and isotopes (Table 3). The selection of specific monitoring types requires further comparison. For example, in the Cranfield storage area in the United States, noble gasses are minor natural components in the injected CO₂. Geochemical monitoring in this area extensively uses noble gasses and their isotopic composition to monitor CO₂-driven oil recovery efficiency and CO₂ loss [77]. In controlled-release experiments in the North Sea, artificial tracers are more effective due to their very low concentrations in the marine environment [13]. These studies indicate that the choice of monitoring methods should be based on specific site factors.

Additionally, specific monitoring requirements must be met, encompassing two aspects: meeting the monitoring needs at different stages of the CCS project and fulfilling the overall project monitoring requirements, such as minimizing costs and maximizing safety. As shown in Figure 3, the risk of leakage varies significantly at different stages of the CCS project [78]. Therefore, during the low-risk background phase, a baseline monitoring of normal gas and water composition can be conducted, with periodic or continuous monitoring being carried out to meet the project’s needs. As the injection progresses, continuous monitoring with high-frequency sampling and analysis is required. Thus, continuous monitoring or high-frequency periodic monitoring is preferable. In the later stages of injection, when the risk decreases, low-frequency monitoring is sufficient.

The overall monitoring requirements of the project may include reducing costs, enhancing safety, and analyzing storage status. Different combinations should be used based on the objectives of the project. The quantitative evaluation of monitoring indicators is a promising direction, scoring various parameters such as stability, cost, and accuracy to assess the advantages and disadvantages of different indicators [79]. Li et al. [80] selected ten monitoring targets, including injection scale and storage efficiency, and assigned empirical values to different monitoring technologies to indicate their recommendation levels. While this method meets the basic needs of CO₂ geological storage and allows for different monitoring combinations, it is overly subjective and lacks quantitative standards. The Weighted Sum Method (WSM) can reduce subjectivity by assigning weights to each target, converting the multi-objective problem into a single-objective problem. The weight ratio can be determined using the Analytic Hierarchy Process (AHP) to establish a judgment matrix, with specific evaluation factors based on site characteristics. This method is simple and has been applied at the Shenhua CCS storage site [58]. However, WSM and the empirical assignment method share similar issues, being suitable for finding the optimal solution under a single target condition and potentially not finding the true optimal solution. In contrast, multi-objective optimization methods have broad prospects in geochemical monitoring. They can consider multiple evaluation indicators simultaneously, such as accuracy, sensitivity, cost, and time efficiency, providing a comprehensive evaluation framework for selecting and optimizing geochemical monitoring technologies. Depending on the specific monitoring objectives, such as minimizing false alarm rates, maximizing detection sensitivity, and minimizing monitoring costs, suitable multi-objective optimization algorithms can be chosen. Among them, Multi-Objective Simulated Annealing (MOSA) is popular for its ability to adapt to various constraints and achieve global optimization [81]. MOSA first randomly selects a set of initial solutions ƒ₀ as the optimal solution, randomly alters the optimal solution to form a new solution ƒ₁, and compares the new solution with the original optimal solution to determine a new optimal solution (Figure 4). The final solution of this algorithm does not depend on the initial solution’s selection. The objective function should reasonably reflect the overall optimization requirements of the problem and be easy to calculate, improving the efficiency of the algorithm. Applying the simulated annealing algorithm to multi-objective optimization problems, by accepting certain deteriorated solutions to escape local optima, is a feasible path for optimizing geochemical monitoring schemes in the future.

Additionally, there is a certain correlation between different indicators, and the possibility of substituting one indicator for another can be considered. For example, there is a high correlation between CO₂ leakage concentration and radon isotope changes (r = 0.858) [82], and a good correlation between N₂ and O₂ concentrations and CO₂ concentration [36]. Moreover, Risk et al. [37] introduced the potential signal–noise ratio (PSNR) to represent the ideal potential signal-to-noise ratio for comparison in the absence of leakage, with higher PSNR values indicating a better indicator performance (Figure 5). While methods like ⁴He/²⁰Ne and ¹⁴C are effective, their high costs limit their routine use. Conversely, more affordable methods such as Ar and N₂ monitoring also show high PSNR values. Therefore, to reduce costs, inexpensive indicators can provide a substitute for expensive monitoring methods. For instance, pH, alkalinity, Ca²⁺, Fe²⁺, and CO₂ δ¹³C values are cost-effective indicators and proved to be the best monitoring parameters at the Pembina Cardium storage site in Canada [83].

Future research should focus on developing new monitoring methods tailored to the needs of geological storage sites, such as gas chromatograph–mass spectrometers for tracer monitoring and small CO₂ detectors for gas concentration measurements. Additionally, it is important to combine existing monitoring methods effectively, such as integrating gas and water monitoring to directly reflect changes in the subsurface environment and the geochemical reactions and impacts during CO₂ storage, providing real-time feedback on underground changes. Indirect methods like tracers and isotopes can be used to quantify and map CO₂ leakage, offering precise quantitative data, predicting and simulating subsurface environmental changes, and assessing the long-term effectiveness of CO₂ storage. He isotopes, Xe isotopes, PFCs, and CD₄ are viable options due to their stable geochemical behavior, low environmental impact, and cost-effectiveness [13]. A suitable monitoring plan can be both effective and cost-efficient, but in cases of suspected leakage, clear and definitive indicators should be employed. This approach allows for streamlined monitoring, ensuring that effective monitoring does not need to be excessively expensive [37].

3.2. Storage State and Leakage Assessment

With continuous technological advancements, geological storage monitoring techniques have become more refined and are moving toward quantitative characterization. The quantitative characterization of storage state and leakage assessments is crucial for ensuring the safety and effectiveness of storage projects and is a key focus in the current geochemical monitoring research.

Research on storage state can analyze the effectiveness of the CO₂ storage system. Geochemical monitoring evaluates the storage capacity and distribution of CO₂ by monitoring the concentration of CO₂ in underground rocks. Differences in oxygen isotope compositions between formation water and CO₂, based on oxygen isotope similarity calculations, can quickly determine the residual saturation within the storage area [84]. The combination of carbon and oxygen isotopes can indicate the presence of CO₂ and determine the storage mechanisms. In the Pembina Cardium CO₂ monitoring project in Canada, the equilibrium isotope exchange relationship and CO₂ solubility were used to calculate the saturation of fluids and gasses in the pore space (0.05~0.60), qualitatively and quantitatively determining the presence of CO₂ around observation wells [85]. Fluid and gas monitoring can also be used to trace the injected CO₂. By quantifying the mass of CO₂ injected as HCO₃⁻, the scale of CO₂ storage in reservoir fluids can be estimated. Studies show that the ionic capture of injected CO₂ is the most important source of HCO₃⁻ in the Weyburn storage area, with the dissolution of carbonate minerals being a secondary source. A reconstruction of the measured data based on reservoir conditions indicates that approximately 185,000 tons of injected CO₂ is stored in reservoir fluids [71].

Geochemical monitoring can provide data on changes in CO₂ concentrations in underground rocks and groundwater, assessing CO₂ storage rates and stability. At the Cranfield storage site, rare gasses are minor components of the injected CO₂. The He isotope ratio and ⁴⁰Ar*/⁴He ratio are significantly correlated with CO₂ concentrations and can serve as tracers for CO₂ [77,86]. Using rare gasses like Ar and He to trace the dissolution of CO₂ in formation water and its interaction with residual oil, the gas composition and rare gas isotope composition in different regions were measured, showing that the CO₂ stripped inert gasses from the formation water [14,86]. The He isotope composition indicates that about 0.2% (about 7 kt) of the injected CO₂ dissolved in formation water. Ne isotope composition was used to determine the isotope composition of natural gas. Using rare gas composition and isotope analysis to trace and quantify CO₂ concentration at the site during continuous injection, it was found that a significant portion of the CO₂ in gas phase samples collected from production wells lost inert gasses, with 22~96% of the CO₂ being lost in individual wells (Figure 6,

Table 4. The measured and reconstructed CO₂ concentrations from different wells (data derived from [59,86]).

Well	Measured CO2	Loss of CO2 from CO2 Mix (%)
		3He/4He	40Ar/4He	C3F8	SF6	Kr	CH4
28F-2 2009	0.9%	93	96	/	/	/	/
29F-1 2009	3.3%	85	91	/	/	/	/
44-2 2009	8.4%	77	86	/	/	/	/
29-5 2009	40.0%	30	22	/	/	/	/
27-5 2009	46.9%	28	42	/	/	/	/
370-1	86 kg	/	/	27.9	34.6	40.0	42.1
370-2	86 kg	/	/	34.7	41.0	47.9	43.4
372	86 kg	/	/	64.3	70.6	70.8	51.6
373	86 kg	/	/	46.1	52.1	55.2	44.3
376	143 kg	/	/	36.6	52.5	53.0	40.8

). This suggests that inert gasses effectively quantified the CO₂ storage potential [14,86]. Analyzing the migration paths and sources of CO₂ can help to determine the causes of leakage events, the scale of the leakage, and the means of the underground propagation of leakage substances, which is significant for reducing the risk related to storage systems and ensuring their sustainability and safety. Many scholars have used synthetic chemical tracers and naturally occurring elements, such as stable isotopes of light elements (¹⁸O, D, ¹³C, ³⁴S, and ¹⁵N), rare gasses (He, Ne, Ar, Kr, and Xe), and radioactive isotopes (such as tritium, ¹⁴C, ³⁶Cl, ¹²⁵I, and ¹³¹I), to analyze the migration paths and sources of CO₂ [15,16,57,70].

The analysis of CO₂ migration pathways involves tracking CO₂ concentrations in groundwater and rocks by injecting tracers or isotopic markers. CO₂ leakage typically affects the chemical characteristics of groundwater, such as the changes in pH, dissolved oxygen levels, and ion concentrations. Changes in the chemical components related to CO₂ leakage in subsurface rocks can be used to identify the CO₂ source. The migration and transformation of pre-existing trace metals caused by CO₂ plumes or acidified waters can reveal the movement of underground CO₂ [46,62]. In addition to CO₂, other gasses’ compositions can also provide valuable information. For instance, some hydrocarbons have significantly higher solubility in supercritical CO₂ than in water [87], so monitoring organic gasses or volatile organic compounds (VOCs) can also indicate CO₂ leakage. During migration, the composition of various rare gasses changes. The clear relationship between rare gasses and CO₂ plumes shows that degassing and mixing mainly control CO₂ retention in shallow groundwater. Kr is an early indicator of CO₂ migration, while He and Ar are diluted and vary with the CO₂ migration pathways [64]. Combining rare gas isotope ratios (⁴He/²⁰Ne) can indirectly indicate the source of leaked CO₂ [88]. Integrating CO₂ plume isotopes with trace gas compositions to track the fate of injected gasses in storage projects can identify unintended CO₂ migration to a shallow subsurface or surface [89]. For example, Flohr et al. [59] simulated the leakage of CO₂ from a seabed reservoir in the North Sea, releasing CO₂ along with various natural tracers (¹³C, ¹⁸O) and added non-toxic tracer gasses (PFC, SF₆, Kr, methane). The results revealed that the solubility of CO₂ in sediment pore water ranged from 35% at the lowest injection rate to 41% at the highest injection rate. Between 22% and 48% of the injected CO₂ left the seabed at the lowest and highest injection rates, respectively, while the remaining CO₂ accumulated in sediment gas pockets. These methods quickly confirmed the CO₂ leakage source analysis and other aspects of the analysis (Table 4).

The use of isotopic tracer technology is an effective method for determining CO₂ sources. CO₂ from different sources usually has a distinct isotopic composition. After determining the δ¹³C value of the storage environment, changes in regional δ¹³C values often accompany increases in CO₂ or DIC concentrations, aiding in evaluating the movement of injected CO₂ in the target reservoir and leakage at the CO₂ storage site [69]. The sole reliance on ¹³C isotopes may not be sufficient for characterization. For example, ¹⁴CO₂ can be used to assess CO₂ leakage [37]. Although leakages are mainly composed of CO₂, the high variability of natural-source CO₂ makes it challenging to pinpoint leakage locations based solely on CO₂ isotopes. Therefore, monitoring δ¹³C values, methane concentrations, and soil CO₂ concentrations over time can help determine whether CO₂ is leaking and the extent of the leakage [68]. Alternatively, coupling carbon isotopes with oxygen isotopes [90], hydrogen isotopes [72], and sulfur isotopes [71] can provide a comprehensive reflection of the leakage. The characteristics of the injected CO₂ vary, and carbon isotope ratios are effective for tracking the movement and reactions of the CO₂ in mature oil fields [91]. The monitoring of oxygen isotopes has been widely applied in storage projects such as the Otway project in Australia, Frio in the U.S., and Pembina in Canada, revealing the groundwater saturation levels at different injection stages [84]. The isotopic “fingerprint” formed by carbon and oxygen isotope ratios can track the fate of the injected CO₂ in respective reservoirs [92]. By monitoring changes in CO₂ isotope ratios in groundwater and rocks in the storage area, whether the CO₂ originates from the storage site can be determined. For example, if CO₂ isotope ratios are inconsistent with known CO₂ sources in the storage area, this may indicate an external CO₂ leakage. In the Weyburn and Pembina Cardium projects, the C and O isotope values of the injected CO₂ differed significantly from the C and O isotope values of background reservoir CO₂. This not only shows the fate of the injected CO₂ in the reservoir but also allows for the future quantification of CO₂ dissolution in the water based on the quantity of oxygen in the CO₂ sources [93]. However, the use of oxygen isotopes alone is limited by their non-conservative nature. Karolyte et al. [90] found that water–rock reactions are unlikely to significantly affect observed δ¹⁸O values. Therefore, the displacement of δ¹⁸O values in water can be used to monitor CO₂’s impact on shallow groundwater aquifers, provided there is sufficient CO₂ and a difference between water δ¹⁸O values and CO₂ δ¹⁸O values. Combining hydrogen isotopes with oxygen isotopes can also reflect the migration pathways of underground fluids [72], but current research on this is still relatively limited.

4. Outlook

4.1. Application of Artificial Intelligence and Machine Learning

Recent advancements in artificial intelligence (AI) and machine learning (ML) have introduced new opportunities and challenges for geochemical monitoring. AI and ML can potentially enhance numerical simulations and data analyses in geochemical monitoring. By integrating geochemical simulation technologies with data on groundwater flow and rock pore structures, we can model the migration pathways of underground CO₂. These simulations help assess the migration and distribution of CO₂, as well as the potential sources of CO₂ leakage, by simulating different leakage scenarios. This approach typically combines groundwater flow models and rock pore structure models to predict the diffusion and transport of CO₂.

Simulation experiments of CO₂ injection and leakage are essential for evaluating the feasibility, safety, and effectiveness of various geological storage strategies. Repetitive simulation experiments provide extended time series data, improving the reliability of the monitoring methods [94]. Currently, field-based simulations are commonly used (Figure 7), including (1) CO₂ controlled-release experiments, which model the underground environment’s response and surface changes during CO₂ leakage from the storage site [59,95,96]; (2) CO₂ injection experiments in underground aquifers, which simulate CO₂ plume migration and storage mechanisms [40,64]; (3) the monitoring of natural CO₂ leakage points, such as volcanic rock areas, with a focus on applying different monitoring techniques [97,98,99]. While these field simulation methods provide continuous data, such as the chemical and isotopic composition of groundwater and rocks, offering direct support to the research and verifying the storage effectiveness, they are conducted in open environments. Consequently, factors such as temperature, pressure, and solution composition cannot be controlled, meaning that the technique may not meet specific environmental requirements.

Numerical simulations can be performed in controlled environments to model specific underground geochemical processes and validate real-world scenarios, offering an effective complement to field simulations. They can also predict changes over several years to several decades based on geological conditions [100]. During numerical simulations, extensive geochemical data (such as major and trace elements and isotope values) can be used as input parameters, resulting in high consistency between the simulation results and actual measurements [12]. For example, Yang et al. [101] used Fluent to simulate CO₂ leakage patterns in the Yanchang Oilfield under various scenarios. By comparing the monitoring data with simulation results, they validated the assumptions and parameters of their CO₂ numerical simulation model and optimized the placement of monitoring points. The structured data generated from these simulations are valuable for refining and optimizing CO₂ models and can be applied to similar scenarios [102]. Wang et al. [103] used a BP neural network to predict CO₂ injection capacity in saline aquifers. Testing the model on projects like Sleipner, Quest, and Illinois showed that integrating numerical simulations with machine learning can significantly enhance calculation speed and accuracy. Uncertainty quantification methods enable accurate and rapid predictions of CO₂ concentration distributions without traditional inversion modeling [104].

Geochemical monitoring generates vast amounts of data, which traditional processing methods often struggle to handle. AI and ML technologies, particularly deep learning, enable models with multiple processing layers to learn complex data representations, improving their processing efficiency and accuracy [105]. These technologies can build sophisticated geochemical models and prediction systems. By training neural network models, it is possible to simulate CO₂ migration paths and changes in reservoirs and predict potential leakage locations and scales. This includes calculating parameters such as storage solubility, capture efficiency, and various capture indices [106,107,108]. Such predictive capabilities can help decision-makers take proactive measures to mitigate the environmental risks. AI and ML will increasingly play a role in integrating diverse data sources, including geological, geophysical, and chemical data. By merging these sources, AI algorithms can provide more comprehensive and accurate monitoring results and site selection criteria [109]. AI and ML technologies hold significant promise for geochemical monitoring at CO₂ storage sites, enhancing efficiency and effectiveness in areas such as real-time data processing, model construction, automated monitoring, and anomaly detection.

In geochemical monitoring simulations, machine learning algorithms can analyze historical data to create anomaly detection models, quickly identifying anomalies in monitoring data. As illustrated in Figure 8, algorithms like artificial neural networks (ANN) and convolutional neural networks (CNN) are commonly used to predict CO₂ storage properties, assess mechanical stability, and monitor CO₂ plume migration and leakage [110,111,112]. Deep convolutional network algorithms, in particular, have shown a strong performance in analyzing complex geochemical data from groundwater and gas samples in offshore Brazil experiments, effectively detecting abnormal signals and potential CO₂ leakage points [113]. Support vector machines (SVM) and support vector regression (SVR) methods are used for predicting rock properties and analyzing sensitivity to various factors [114,115]. Generative adversarial networks (GAN) and long short-term memory networks (LSTM) are employed for the real-time monitoring of CO₂ migration and leakage [116,117,118]. Decision trees (DT) and random forests (RF) are used to develop risk assessment frameworks and decision analysis tools to estimate CCS success probabilities [110,119,120]. Overall, deep learning methods like CNN and GAN show promise in various scenarios in CCS storage sites and merit further exploration in the future.

Machine learning (ML) methods, with their distinct advantages and capabilities, are becoming increasingly important in various research areas of carbon capture and storage (CCS). To fully utilize the potential of machine learning and achieve the best results, it is essential to select the most suitable ML methods for specific research needs and to develop integrated models that combine different techniques. However, due to the significant geological variability and heterogeneity across regions, as well as the dynamic nature of underground CO₂ migration, the application of machine learning and artificial intelligence technologies requires extensive data for effective training. Consequently, sites with limited monitoring data may struggle to make accurate predictions. Recently, the rapid growth in large-scale models has led to a proliferation of machine learning algorithms, some of which have limited intersections with geological disciplines. Future advancements must overcome challenges related to data volume, algorithm transparency, and standardization. By continuously refining the application processes of these technologies, the safety and effectiveness of CO₂ geological storage can be further enhanced.

4.2. Baseline Survey and Internet of Things Monitoring

To ensure the long-term safe storage of CO₂ underground, conducting a baseline survey is essential. This survey aims to evaluate the initial geochemical state of the storage site. A baseline survey involves a comprehensive and systematic assessment of the natural conditions of the target area before starting experiments or monitoring. This includes evaluating fundamental geological features, hydrological characteristics, and the environmental background [121]. Comparing the subsequent data with the baseline data allows for the timely detection of changes in water quality, atmospheric composition, and other factors, providing early warnings of environmental changes. Baseline monitoring serves as the foundation for both the initial monitoring efforts and ongoing environmental protection measures.

For instance, the Otway storage site initiated long-term environmental baseline monitoring before CO₂ injection. Due to the influence of the Campbell Port Limestone, the natural baseline is complex. The long-term monitoring of groundwater composition, tracer compounds, CO₂ levels, water levels, and seasonal variations in water flow rates and directions helped establish the connectivity, fluid composition, and migration timeline of the freshwater aquifer [122]. The Quest project established hydrological and geochemical baselines for four target aquifers. Based on existing water head data, well pumping tests, and aquifer characteristics, the inferred regional groundwater flow direction and velocity were calculated to determine if fluid and/or gas migration occurred in the reservoir [123]. In the Weyburn storage area in Canada, environmental benchmarks, including N₂, Ar, He, and CO₂/O₂ ratios, were used to reduce the impact of background variability on monitoring indicators. Typically, after establishing baselines for water chemistry and carbon isotope composition, injection simulation experiments were conducted. The arrival of CO₂ plumes in monitoring wells is determined by decreases in pH, increases in CO₂ partial pressure, and changes in conductivity [39]. Column experiments have shown that calcium (Ca) is highly responsive to changes in CO₂ flux, and once a baseline concentration is established, calcium (Ca) can be used as a parameter for monitoring CO₂ leakage [48]. Baseline surveys are therefore a necessary part of the pre- and post-storage processes. However, these methods heavily rely on data-intensive baseline studies to identify anomalies. After establishing the baseline, comparing results from different monitoring dates can help detect leaks and identify potential leakage locations and scales (Figure 9). Nevertheless, predicting future leakages based on pre-injection baseline surveys is challenging due to natural variations in CO₂ concentrations in the soil, the groundwater, and the atmosphere [124]. Thus, adjusting baseline values to account for the continuously changing environmental conditions will be a critical aspect of future baseline surveys.

In recent years, advancements in sensing, computing, and communication technologies have enabled the integration of computing, communication, and processing functions into low-cost, low-power sensors. Internet of Things (IoT) online monitoring technology uses wired or wireless communication to connect various types of sensors, forming wireless sensor networks (WSN) that transmit data over the internet for remote access. This allows for the simultaneous monitoring of parameters such as temperature, humidity, pressure, and spatial location (Figure 10) [126]. IoT online monitoring can continuously collect geochemical data, including groundwater chemical composition, gas concentrations, pressure, and temperature, providing a real-time, precise monitoring of environmental parameters. This continuous data collection captures dynamic changes in environmental parameters at storage sites, offering comprehensive baseline information, which is a key advantage in baseline surveys. Moreover, IoT online monitoring also has promising applications in routine monitoring post-storage. Given the large and remote nature of CO₂ geological storage sites, deploying numerous sensors across these areas is necessary. Sensors with precise positioning capabilities allow for high-precision monitoring over extensive areas and enable adjustments in sensor density as needed [127].

Recent studies have explored the use of IoT online monitoring for baseline and routine monitoring in the geosciences. Wang et al. [128] developed a real-time monitoring system for urban soil pollution data using IoT. They integrated the Spring Cloud Alibaba microservice framework with the EMQX platform for soil data collection and storage, and created a WebGIS module for map rendering and soil concentration visualization, aiding in pollution prediction and warnings. Yang Hui et al. [129] used IoT to collect CO₂ concentration data and other related information in the Junggar Basin and Turpan Basin of the Xinjiang Uygur Autonomous Region. They built a neural network model for a regression analysis of key factors like NDVI and topographic undulation, revealing high CO₂ concentration accumulation in energy resource areas influenced by vegetation, meteorological conditions, and topography, and identifying varying correlations between NDVI, topographic undulation, and CO₂ concentration across different regions [23]. Ma et al. [127] suggest that IoT online monitoring technology effectively meets the needs of geological storage sites, proposing sensor design ideas, and deployment strategies, and establishing a monitoring system that can be extended to geological engineering construction monitoring [130].

Although IoT online monitoring involves high initial costs for sensor deployment, it is a promising technology with significant potential when combined with long-term baseline and routine monitoring. Additionally, using machine learning to establish environmental baselines and analyze interactions between CO₂ leakage and environmental factors such as temperature, humidity, and wind speed can help mitigate the impact of environmental changes [82]. Therefore, the use of IoT online monitoring technology with CO₂ sensors, which is adaptable to monitoring range and frequency requirements, represents a promising direction for future development [23,131].

5. Conclusions

CO₂ geological storage projects are critical technologies for addressing climate change. Geochemical monitoring is essential for the successful implementation of storage projects and for ensuring the safety and effectiveness of storage systems. This paper reviews methods such as gas monitoring, groundwater monitoring, and tracer monitoring, detailing the effectiveness of geochemical methods in various contexts. This highlights the role of geochemical monitoring in identifying leakage sources and scales, analyzing migration pathways, and characterizing storage conditions. Based on existing research and the strengths and limitations of these methods, the following recommendations to enhance monitoring strategies are offered:

• Implement a comprehensive multi-method monitoring approach to improve accuracy and coverage.
• Strengthen baseline surveys to establish reliable environmental reference standards.
• Utilize Internet of Things (IoT) technology for real-time data collection.
• Integrate artificial intelligence (AI) and machine learning (ML) to enhance data processing and achieve more accurate anomaly detection.
• Adjust monitoring plans dynamically based on the results obtained.

Combining IoT technology with AI and ML could lead to the development of advanced automated monitoring systems. At CO₂ geological storage sites, sensor networks can provide real-time geochemical data, while AI algorithms facilitate automatic analysis and reporting, thereby greatly improving monitoring efficiency and responsiveness.