Remote Sensing for Quantifying Greenhouse Gas Emissions at Carbon Capture, Utilisation and Storage Facilities: A Review

Highlights

What are the main findings?

• We synthesise sensors, retrieval methods and inversion frameworks to quantify greenhouse gas emissions at carbon capture, utilisation and storage sites.
• We demonstrate how to combine in situ, drone and satellite observations and provide indicative detection limits and cost tiers to guide method selection.

What is the implication of the main finding?

• The proposed framework enables more reliable verification of emissions and leaks, supporting regulatory compliance and carbon-market reporting.
• Harmonised standards and data assimilation across platforms can reduce uncertainty and costs while improving transparency for operators and authorities.

Abstract

Carbon capture, utilisation and storage technologies are increasingly recognised as critical components of global climate mitigation strategies. However, the effective monitoring and verification of greenhouse gas emission reductions from carbon capture, utilisation and storage facilities remain significant challenges. This review synthesises current monitoring methods, including in situ sensing, drone-based observations and satellite remote sensing, and critically evaluates their strengths, limitations and applicability to various carbon capture, utilisation and storage contexts. We analyse the regulatory frameworks that govern monitoring practices across jurisdictions, identify methodological gaps and assess the performance of existing technologies with respect to detection thresholds, the integration of multiple data sources and the requirements for long-term verification. Particular emphasis is placed on the role of data assimilation and inversion modelling in interpreting measurements and quantifying emissions. Based on this synthesis, we recommend a more harmonised, concentration-based approach to monitoring that combines diverse observation platforms to enhance the accuracy, transparency and cost-effectiveness of verification efforts. This review aims to support the development of best practices for environmental monitoring and assessment in the context of carbon capture, utilisation and storage deployment.

1. Introduction

Carbon capture, utilisation and storage (CCUS) technologies have garnered considerable attention as critical tools for achieving the Paris Agreement targets [1]. While multiple countries and organisations have begun large-scale CCUS deployment, an unresolved challenge is ensuring that these facilities truly deliver net reductions in greenhouse gas (GHG) emissions [2].

The multifaceted nature of CCUS technologies allows for diverse applications, ranging from direct air capture (DAC [3]) to sequestration of emissions at large-scale industrial facilities for utilising the CO₂ for enhanced oil recovery (EOR [4]) and synthesis of valuable compounds [5]. With the potential to significantly reduce emissions while contributing to economic growth, CCUS has become increasingly integrated into national and international climate mitigation strategies.

CCUS systems include CO₂ capture from large point sources, transport via pipelines or shipping, utilisation in industrial processes and long-term geological storage in deep reservoirs. Each phase has specific monitoring needs, particularly leak detection and verification of permanent storage, which motivates the development of remote-sensing-based monitoring frameworks.

Despite significant progress, multiple recent studies have documented persistent underestimation in official GHG inventories. For instance, sector-wide evaluations have observed gaps of up to 25% [6], global satellite analyses of oil and gas “super-emitters” indicate a 30–70% discrepancy [7] and in a Danish case study, emissions were found to be 56–75% higher than reported estimates [8]. Such discrepancies can arise from data uncertainty, reporting biases and the inherent complexity of tracking fugitive emissions in large-scale industrial settings. In particular, CCUS monitoring faces methodological gaps related to accurately quantifying fugitive emissions and distinguishing anthropogenic signals from natural or biogenic sources. Traditional “bookkeeping” methods tend to underestimate small leaks or ignore intermittent emissions, while more advanced techniques (including direct measurements via remote sensing [9] and in situ sensors [10]) are rarely integrated into a unified framework. Consequently, operators, regulators and carbon trading platforms all face the same core questions: how can we ensure consistent, verifiable evidence that CCUS facilities provide meaningful net reductions in GHG emissions and which emerging technologies can best address the existing uncertainties?

While several recent reviews have addressed aspects of greenhouse gas monitoring, few have specifically focused on the integrated use of remote sensing to quantify emissions from CCUS facilities. Existing studies often treat satellite, airborne and in situ techniques separately or emphasise natural and fossil-fuel sources rather than engineered storage systems. This review therefore synthesises the state of knowledge on remote sensing sensors, retrieval algorithms and inverse modelling frameworks as they apply to CCUS monitoring, identifying current limitations in spatial resolution, detection thresholds and data integration.

The literature considered covers the period from 2001 to mid-2025 and includes peer-reviewed journal articles and technical guidelines relevant to greenhouse gas verification. Studies were selected based on methodological relevance, novelty and their demonstrated applicability to CCUS-scale monitoring.

In this review, we critically examine the current methods used to monitor and verify greenhouse gas emissions from CCUS operations. We synthesise recent technological developments, compare international regulatory frameworks and evaluate the performance and limitations of in situ, drone- and satellite-based monitoring approaches. Based on this synthesis, we propose recommendations for a harmonised, concentration-based monitoring framework designed to enhance transparency, cost-effectiveness and environmental accountability in future CCUS deployments.

This review addresses remote-sensing sensors, retrieval algorithms and inversion/data-assimilation methods that transform observations into robust, policy-relevant emission estimates from facility to regional scales.

2. Monitoring Requirements and Regulatory Frameworks for CCUS

2.1. Global Deployment and the Need for Monitoring

As of early 2025, the International Energy Agency (IEA) reports that just over 50 Mt CO₂ yr⁻¹ of capture and storage capacity is operational globally, with approximately 79 large-scale CCUS facilities in operation across nine industrial sectors. More than 600 additional CCUS projects spanning capture, transport, utilisation and storage are in various stages of planning and development [11]. While many operational facilities remain associated with natural gas processing and enhanced oil recovery (EOR), the portfolio of projects is rapidly expanding to include bioenergy with carbon capture and storage (BECCS), direct air capture (DAC) and dedicated geological storage sites, both onshore and offshore. This accelerating deployment underscores the growing importance of robust, transparent and scalable monitoring frameworks to verify storage performance, detect potential leaks and ensure climate integrity.

EOR has traditionally been one of the primary applications for CCUS, as the injected CO₂ aids in recovering residual hydrocarbons. However, the net climate benefit remains contentious. Cooney et al. [12] reported EOR scenarios with net emissions of approximately 100 g CO₂e/MJ, depending on system boundaries and the ultimate combustion of the extracted oil. While lower emissions in this range can indicate near carbon-neutral operations (where most of the injected CO₂ remains underground), higher values suggest that a significant share of the injected CO₂ eventually returns to the atmosphere, offsetting much of the potential climate benefit. This wide variability underscores the importance of implementing robust monitoring and verification frameworks to accurately gauge whether EOR genuinely contributes to net emission reductions.

BECCS is widely regarded as a negative emission technology, since the CO₂ assimilated by biomass can be permanently sequestered rather than re-released to the atmosphere. In a recent analysis, Pratama et al. [13] documented a net removal potential ranging from approximately 0.3 to 1.0 tons of CO₂ per ton of dry biomass, subject to variations in feedstock type, land-use practices and the efficiency of the capture process. These findings reinforce the importance of robust measurement and verification schemes, as even minor fugitive emissions or inefficiencies along the supply chain may reduce the overall climate benefit that BECCS is intended to deliver.

In order to evaluate the net result of a given facility, it is thus important to have a measurement, monitoring and verification (MMV) programme that is capable of taking into account both the sink and the emissions associated with given facilities.

Among the reasons why a CCUS facility needs a MMV programme, two are of the utmost importance, namely

• Documentation for legislation
• Documentation carbon emission trading

Below we will go through the requirements needed for the two kinds of documentation.

2.2. Legislative and Regulatory Frameworks

The Global CCS Institute maintains a CCS legal and regulatory indicator (the CCS-LRI), which is described in Havercroft [14]. The indicator describes the state of CCS-specific laws or existing laws that are applicable across most parts of the CCS project cycle. Countries like Australia, Canada, Denmark, the United Kingdom and the USA score high on this indicator. In general, most European countries score high on the indicator, which is likely related to the EU CCS directive that has been in place since 2009 and had to be transposed into national law by June 2011. In the following, we go through the main points in the EU CCS directive.

The aim of the directive is to establish a legal framework for environmentally safe geological storage of CO₂. This is achieved by outlining the procedures for obtaining permits for the operation of CCS facilities. These procedures include submitting an application to the EU member state hosting the facility. Such an application should include the following points:

• The name and address of the potential operator;
• Evidence of the operator’s technical competence;
• A comprehensive characterisation of the storage site and an assessment of its expected security;
• The total volume of CO₂ to be injected and stored, including details on prospective sources, transport methods, CO₂ stream composition, injection rates, pressures and the location of injection facilities;
• A description of measures implemented to prevent significant irregularities;
• A detailed monitoring plan;
• A corrective measures plan for addressing any detected issues;
• A provisional post-closure plan.

The directive further requires that the monitoring of the CCS facility (covering both the injection points and the storage complex up to the potential extent of the CO₂ plume) includes

• Comparison between the actual behaviour and the modelled behaviour of CO₂ in the storage site;
• Detection of significant irregularities;
• Detection of CO₂ migration;
• Detection of CO₂ leakage;
• Detection of significant adverse effects on the surrounding environment, particularly on drinking water, human populations or ecosystem users;
• Assessment of the effectiveness of any corrective measures implemented in response to leakages or irregularities;
• Periodic updating of the safety and integrity assessment of the storage complex to confirm that the stored CO₂ remains completely and permanently contained.

The main problem with this description and with the EU CCS directive, in general, is that is does not provide any indication of how small leakages or irregularities must be to be detected as part of the monitoring programme. In Section 5, we provide a review of how small leakages are realistic to detect.

2.3. Carbon Trading and Verification Schemes

The relationships among key stakeholders in carbon emission trading, regulatory authorities, validators and operators are summarised here. Regulators oversee the trading framework, validators ensure the accuracy and compliance of reported emissions and operators buy and sell allowances within this verified system, maintaining transparency and accountability throughout the process.

At the time of writing, there are more than 35 operating carbon trading systems worldwide, but only a few have reached large-scale implementation and long-term stability. The six most established systems are

• The European Union Emissions Trading System (EU ETS)
• The California Cap-and-Trade Programme
• The Regional Greenhouse Gas Initiative (RGGI)
• The China’s National Emissions Trading System (ETS)
• The New Zealand Emissions Trading Scheme
• The South Korea Emissions Trading Scheme

These systems vary in design, coverage and maturity. The EU ETS, established in 2005, remains the world’s largest and most developed carbon market, covering over 10,000 installations across 30 countries. China’s national ETS, launched in 2021, is the second largest, initially focusing on the power sector but expanding to heavy industry. The California Cap-and-Trade Programme (launched in 2013) and the Regional Greenhouse Gas Initiative (launched in 2009) are well-established North American markets, while South Korea (launched in 2015) and New Zealand (launched in 2008) operate national-level schemes that integrate offset mechanisms. All rely on MVV procedures to ensure transparency and compliance, which are directly relevant to CCUS implementation. Here, we will focus on the EU ETS.

Operating on a cap-and-trade principle, the EU sets a limit or “cap” on the total greenhouse gas emissions permitted from energy-intensive sectors like power generation, manufacturing and aviation. Companies within the system receive emissions allowances, each representing the right to emit one ton of CO₂ or its equivalent. These allowances can be distributed either for free to regulated sectors or auctioned by governments. Participants must surrender enough allowances at the end of each compliance period to cover their actual emissions. Failure to do so results in significant penalties.

Trading of allowances occurs among companies, enabling those needing more allowances to purchase them from others who have a surplus, establishing a market where prices are determined by the balance of supply and demand. Presently, EU ETS does not permit the acquisition of carbon credits from sinks, such as tree planting initiatives aimed at reducing CO₂ levels. The reason for this is linked to the difficulty of monitoring, measuring and verifying the effective removal capacity of such sinks.

2.4. Regulatory and Policy Frameworks for Monitoring

The EU monitoring and reporting regulation (MRR) outlines the requirements for monitoring and reporting GHG emissions. Under this framework, facilities can choose between a “calculation approach for CO₂” or a “measurement approach for CO₂.” The measurement approach allows for the use of continuous emission monitoring systems (CEMSs), which are particularly applicable to CCUS facilities. The required precision levels for CEMSs are specified in Annex VIII of the EU Monitoring and Reporting Regulation No. 2018/2066.

In the EU ETS context, the term “tier” refers to predefined levels of methodological rigour used to estimate greenhouse gas emissions (see

Table 1. Tiers for CEMSs (maximum permissible uncertainty for each tier).

Tier 1	Tier 2	Tier 3	Tier 4
±10%	±7.5%	± 5%	± 2.5%

). Higher tiers correspond to more detailed and accurate measurement or calculation approaches. Category A installations, typically high-emission, energy-intensive industrial facilities, are required to apply at least Tier 2 methodologies, which involve direct measurement of fuel consumption or process parameters rather than default emission factors. This ensures greater accuracy in reported emission data.

It is thus clear that the requirements for carbon emission trading are much more concrete than the requirements from legislation for CCUS. It must, however, be expected that the uncertainty requirements for CEMSs will be adopted by the legislation requirements for CCUS in the future.

While the EU ETS remains among the most comprehensive carbon markets globally [15], other major systems also influence CCUS verification frameworks. For instance, the California Cap-and-Trade Programme employs mandatory reporting and third-party verification under its protocols [16], which focus on monitoring, leakage detection and long-term liability. China’s National ETS, the largest by covered emissions and operational since 2021, continues to refine its verification regime as it anticipates inclusion of CCUS as an offset mechanism [17]. Each of these frameworks therefore establishes tiers of methodological rigour and auditing procedures to assure transparent and consistent emission accounting.

As shown in

Table 2. Comparison of International Regulatory Standards for CCUS Monitoring.

Jurisdiction	Primary Programme & Scope	Monitoring Requirements	Key Gaps/Challenges
European Union	EU ETS, CCS Directive; storage across multiple EU member states	Broad guidelines for site characterisation and MMV; corrective measures required in case of leak	No explicit detection threshold for CO2 leakage; enforcement varies
California	California Cap-and-Trade; large stationary sources with CCUS offset protocols	Mandatory reporting for large emitters; evolving offset rules for CO2 sequestration	Inconsistent coverage for smaller facilities; complex third-party validation
China	National ETS (gradual expansion); initially power sector with potential CCUS inclusion	Likely requirement for continuous monitoring and periodic audits; offset protocols in development	No finalised leak-detection thresholds; unclear long-term liability
Australia	Safeguard Mechanism, Emissions Reduction Fund; major industrial sites	Project-based crediting if permanence and additionality are demonstrated; operator-driven MMV	Complex approval process; uncertain long-term stewardship guidelines
Quest	Province-level + Federal oversight; large-scale onshore CCUS	Extensive baseline data; continuous wellhead monitoring; 3D seismic surveys	High monitoring costs; specialised expertise required

, the international policy landscape for CCUS monitoring is highly diverse. For example, the EU ETS and CCS Directive provide broad guidelines but lack explicit leak detection thresholds, while California’s Cap-and-Trade system imposes strict reporting requirements yet faces challenges in covering smaller facilities. In contrast, China’s National ETS is still evolving its CCUS standards and Australia’s framework emphasises project-based crediting with operational complexities. The Canadian example, exemplified by the Quest project, further illustrates the potentials and challenges of comprehensive monitoring in a multi-level regulatory environment.

These policy differences directly influence the selection and implementation of monitoring technologies. Regions with clear, stringent standards are more likely to adopt advanced, high-precision systems, while less prescriptive frameworks may continue to rely on traditional methods. Ultimately, harmonised international standards could drive innovation by creating a competitive market for high-precision GHG monitoring solutions.

2.5. Identified Gaps and Challenges

Despite the ongoing expansion of CCUS facilities worldwide and the evolving legislative landscape, systematic shortcomings persist in the methodologies used to monitor and verify net emission reductions. These gaps impede consistent reporting, effective leak detection and long-term confidence in CCUS as a viable climate solution. The primary challenges can be summarised as follows:

• Inadequate detection thresholds

Many regulatory frameworks provide only broad guidelines (e.g., periodic leak checks) without specifying minimum detection limits. Consequently, small-scale or intermittent leaks can remain undetected, potentially undermining the net-negative impact of certain CCUS operations.

• Limited integration of data sources

CCUS facilities often rely on a single platform, such as bookkeeping estimates or periodic in situ sampling, while ignoring more advanced or complementary approaches like satellite remote sensing and drone surveys. This siloed use of monitoring methods heightens uncertainties in attributing measured emissions to specific process streams or leaks.

• High operational costs

Methodologies capable of achieving high spatial and temporal resolutions (e.g., drone-based hyperspectral imaging or extensive well-log monitoring) can be prohibitively expensive, especially for smaller operators. No standardised cost–benefit framework exists to guide these actors in selecting the most appropriate monitoring suite based on site-specific risks.

• Inconsistent long-term verification

Post-injection monitoring can span decades, but current guidelines offer limited clarity on optimal frequencies and methods for long-term surveillance. As a result, operators may underfund or deprioritize extended monitoring, leading to missed detections of slow-onset leaks.

• Uncertain attribution of emission sources

Geological complexities, combined with natural variability in background concentrations, complicate efforts to distinguish fugitive emissions from ambient fluxes. Accurate attribution requires more robust data-assimilation frameworks and higher data density than many facilities currently maintain.

3. Monitoring Technologies and Platforms

Monitoring technologies used in CCUS systems vary significantly depending on the phase of the CO₂ handling process and the intended monitoring objective. A review by the IEA Greenhouse Gas Programme (IEAGHG) in 2020 identified 43 technologies used in CO₂ storage monitoring and grouped them into three categories: atmospheric, surface/near-surface and reservoir zone techniques [18]. More than half of these (55%) were related to the reservoir zone, which is consistent with the interests of oil and gas operators seeking to maximise hydrocarbon recovery. Consequently, this area has received considerable investment and technological development.

An example of applied monitoring practice is found in the QUEST project [19], where a risk-based MMV plan was developed for containment and conformance monitoring. Over time, the project refined its strategy using a cost–benefit assessment of monitoring technologies based on lifecycle estimates. Technologies were ranked by the number of tasks they could fulfil, weighted by likelihood of success. High-ranking technologies included wellhead CO₂ detectors, satellite or airborne hyperspectral imaging and interferometric synthetic aperture radar (InSAR). In contrast, technologies associated with well logging, although common in reservoir monitoring, scored lower on cost-efficiency.

Given that the focus of this review is on the detection and verification of fugitive emissions rather than subsurface containment, this section concentrates on atmospheric and near-surface methods. These include in situ measurements, drone-based techniques and satellite remote sensing.

Table 3. Overview of GHG observation methods, including indicative detection limits (DLs), measurement precision (Prec.), response time or temporal coverage (Cov.) and approximate device cost ranges in EUR. Cost estimates refer to hardware acquisition only and may exclude installation, calibration, and operational costs.

Method	DL CO2	DL CH4	Prec. CO2	Prec. CH4	Cov. CO2	Cov. CH4	Approx. Cost
In situ
NDIR sensors	1–5 ppm	10–50 ppm	∼2%	∼2%	1–2 s		EUR 500–EUR 5k
Photoacoustic sensors	1 ppm	1–10 ppb	∼1%	∼1%	<1 s		EUR 2k–EUR 15k
TDLAS sensors	0.1–1 ppm	1–10 ppb	∼0.1%	∼0.1%	<1 ms		EUR 10k–EUR 100k
Drones
Sniffer sensors	1–3 ppm	1–80 ppb	∼2%	∼2%	1–2 s		EUR 2k–EUR 20k
Thermal infrared imaging	∼5 ppm	∼10 ppm	∼2%	∼2%	<100 ms		EUR 3k–EUR 50k
Hyperspectral imaging	1–10 ppm	1–10 ppm	∼1%	∼1%	<1 s		EUR 50k–EUR 300k
LiDAR	1–10 ppm	1–10 ppm	∼1%	∼1%	<1 s		EUR 20k–EUR 200k
Satellites
Spectro-scopic sensors	1–4 ppm	10–50 ppb	1–2 ppm	10–20 ppb	1–3 days		Multi-million
Interferometry	–	∼100 ppb	–	10–20 ppb	1–2 weeks		Multi-million
Satellite LiDAR	1–2 ppm	10–50 ppb	1 ppm	10 ppb	1–7 days		Multi-million

provides an overview of the most common approaches, along with indicative detection thresholds, response times and cost estimates.

3.1. In Situ Monitoring Techniques

In situ measurements directly quantify GHG concentrations close to the emission source. The most widely used sensor types include non-dispersive infrared (NDIR), photoacoustic and tunable diode laser absorption spectroscopy (TDLAS) systems.

NDIR sensors determine gas concentrations by detecting the absorption of infrared light at wavelengths specific to the target molecule. Light is passed through a sample chamber and the reduction in signal is measured by a detector. Photoacoustic sensors operate on the principle that gases absorbing modulated light emit acoustic waves; the amplitude of these waves is proportional to the gas concentration. TDLAS systems use a finely tuned laser to interact with the gas sample at precise absorption bands, allowing for rapid, high-sensitivity measurements.

Beyond point sensors, flux towers equipped with multiple GHG sensors and meteorological instrumentation are used to estimate gas fluxes over time [20,21]. Such towers can help link local emissions with broader meteorological dynamics.

Ground-based GHG measurements at CCUS sites commonly employ NDIR and TDLAS techniques. For example, NDIR and cavity-ring-down systems have achieved high-accuracy CO₂/CH₄ mole-fraction measurements under challenging humid conditions [22]. TDLAS-based detectors have been applied in field settings to quantify CH₄ emissions using mobile or stationary platforms [23]. More recently, open-path laser absorption systems have enabled high-temporal-resolution CO₂ and CH₄ flux quantification at industrial-site scale [24]. Real-world applications also include controlled release experiments in Australia and the Netherlands using NDIR-based CO₂ sensors [25,26]. Similar work was carried out under the MONICO project, where the authors tested a CO₂ sensor network under varying wind conditions. The results showed that relatively small leaks, in the order of a few kilogrammes per hour, could be detected up to 15 metres from the source under favourable conditions.

However, the study also revealed key challenges: accurate and hyperlocal wind monitoring was essential for interpreting dispersion patterns; sensors required frequent calibration; and precise background CO₂ levels were critical to avoid false positives from ambient or biogenic fluctuations. Figure 1 presents results from a controlled CO₂ release experiment performed at the Trige test site. The figure shows the temporal evolution of measured concentrations at multiple sensors positioned around the leak source. During periods of steady wind, the CO₂ plume was clearly detected at several sensors downwind of the source. These data confirm that even modest emission rates, on the order of a few kilogrammes per hour, can be detected with relatively simple instrumentation when the network geometry and meteorological conditions are favourable. Such tests are critical for evaluating the sensitivity, response time, and practical deployment strategies of ground-based CO₂ monitoring systems for CCUS applications.

3.2. Drone-Based Observations

Drones provide a mobile and adaptable platform for deploying sensors over complex terrain or industrial sites. Capable of carrying gas detectors, imaging systems and meteorological instruments, drones offer enhanced spatial resolution and flexibility compared with fixed or vehicle-mounted systems.

Drone-based systems increasingly integrate real-time calibration, GPS and inertial data to ensure accurate geolocation of measurements. Advanced processing methods, including spectral unmixing (for hyperspectral imaging) and thermal anomaly detection (for infrared cameras), are used to extract meaningful emissions data. These data-processing steps, including filtering, background correction and dispersion-model inversion, have been employed in field studies of GHG emissions. For example, Deutscher et al. [27] validated open-path near-infrared measurements for CO₂/CH₄ over kilometre paths, Schmitt and et al. [28] demonstrated urban open-path flux quantification using laser absorption spectroscopy and Wietzel et al. [29] applied mobile transects with Gaussian plume modelling at a biogas facility to quantify CH₄ emissions. Machine learning is also being employed to automate leak detection and quantify emission plumes.

Different sensor types are suited to specific applications: sniffer sensors, electrochemical, NDIR, photoacoustic or TDLAS detection methods can quantify GHG concentrations while performing transects or hovering downwind of potential leaks. Thermal infrared cameras detect emissions by identifying heat differentials from infrastructure or ground surfaces. Hyperspectral imagers capture detailed wavelength-specific data, enabling identification of gas-specific absorption features [30]. LiDAR systems, when drone-mounted, generate 3D maps and support concentration measurements using reflected light characteristics.

A demonstration of drone-based CO₂ monitoring was carried out at the Stenlille gas storage facility, where the Drone Flux Measurement (DFM) method was tested as part of the MONICO project. Drones flew multiple horizontal transects at different altitudes to build a “flux wall” that characterised plume structure in three dimensions. Figure 2 and Figure 3 summarise the results of drone-based CO₂ measurements at the Stenlille facility, including horizontal and vertical concentration variations observed during flight transects, as well as the corresponding overhead map visualisation of CO₂ dispersion across the site.

3.3. Satellite Remote Sensing

Satellite instruments enable broad-scale GHG monitoring over large spatial areas. Though less often applied to individual CCUS facilities, satellites play a vital role in regional or national inventory validation.

Representative satellite missions relevant to CO₂ and CH₄ monitoring are summarised in

Table 4. Representative satellite sensors relevant to CO₂/CH₄ emission quantification. Values indicate typical use-cases rather than strict limits.

Mission/Sensor	Sampling/Footprint	Typical Role	Notes
GOSAT/GOSAT-2	sparse nadir, ∼10 km	regional trends	Long record; XCH4/XCO2.
OCO-2/OCO-3	tracks/snapshot maps, ∼1–3 km	urban plumes	Targeted XCO2; snapshot area mapping (OCO-3).
Sentinel-5P	global daily, ∼3–7 km	regional mapping	High-coverage CH4 enhancements.
GHGSat	tasking, ∼25–50 m	facility detection	High-resolution CH4; taskable scenes.
PRISMA/EnMAP	hyperspectral, 30 m	facility detection	Imaging spectroscopy for CH4 under favourable conditions.
EMIT	60 m, targeted	facility detection	Imaging spectroscopy; mineral payload with CH4 capability.
MethaneSAT	wide swath, ∼100 m	regional + facility	Bridging scale for CH4.
Tanager	30 m class, tasking	facility detection	Dedicated imaging spectroscopy for CH4.
CO2M (future)	wide swath, ∼2 km	national-scale CO2	Policy-facing XCO2 for inventory verification.

, illustrating their typical spatial resolution, coverage and roles in emission quantification from regional to facility scales.

Spectrometers onboard satellites like GOSAT, OCO-2 and Sentinel-5P measure absorption of sunlight reflected off Earth’s surface to determine column-averaged gas concentrations [31,32,33]. Interferometry, as used by GHGSat or the CLERREO mission, provides detailed spectral data across images using interferometers. LiDAR missions like MERLIN offer vertical profiles of gas concentrations by analysing the delay and backscatter of laser pulses [34].

Satellite-data processing includes radiative transfer modelling, albedo correction and inversion algorithms to translate radiances into concentrations. When combined with ground-based data and dispersion models, satellite data offer powerful top-down emission estimates.

A notable example is the work by Lauvaux et al. [7], which used satellite data to detect methane super-emitters. By integrating high-resolution satellite observations with inversion modelling, they demonstrated detection sensitivities as low as 10–50 ppb above background, identifying emissions that would have likely been missed by local monitoring. While such applications are rare in CCUS contexts today, the methodology has strong potential for future integration.

3.4. Comparative Strengths and Limitations

Each monitoring approach has distinct advantages and limitations: in situ systems offer high precision and rapid response but limited spatial coverage. They are best suited for near-source detection, though they require regular calibration and careful background correction. Drones bridge the gap between point sensors and satellite imagery. Their mobility and multi-sensor integration make them valuable for site-wide plume mapping, but their use is constrained by flight time, weather conditions and operational logistics. Satellites provide global or regional coverage and trend analysis, but spatial resolution is coarser and real-time application to CCUS remains challenging. An integrated monitoring framework that combines multiple platforms is increasingly seen as best practice.

In practice, satellite observations provide wide-area screening to identify potential leakage anomalies on regional to basin scales. When such anomalies are detected, airborne remote sensing systems enable targeted plume mapping and quantification at facility scale. Finally, in situ sensors provide continuous, high-frequency monitoring near wells and infrastructure for real-time verification and rapid leak detection. Together, these systems form a multiscale monitoring framework that supports both routine verification and rapid diagnostic response at CCUS sites.

4. Data Processing and Emission Estimation Methods

All the measurement methods described above require preprocessing before they can provide meaningful GHG concentration or flux data. This preprocessing differs slightly between in situ sensing and imaging techniques but, generally, involves two main steps. First, retrieval algorithms are used to convert raw sensor data into measured concentrations. Then, these concentrations are used as input for dispersion or transport models, which, together with meteorological and topographic information, allow for the estimation of GHG emissions. These models can also incorporate different types of observations through data assimilation.

4.1. Retrieval Algorithms

Retrieval algorithms are responsible for processing the raw outputs from different measurement platforms to estimate GHG concentrations. The specific steps involved depend on whether the observations originate from in situ sensors, drones or satellites. Below, we outline the typical processing steps for each platform.

• In Situ Sensors

For point-based measurements such as NDIR, photoacoustic and TDLAS sensors, retrieval, generally, includes the following steps:

1.

Calibration: Align the sensor readings with known reference standards, correcting for baseline drift and signal noise.

2.

Concentration calculation: Use physical models (e.g., the Beer–Lambert law) to derive concentrations from signal intensity or absorption spectra.

3.

Environmental corrections: Apply adjustments for temperature and pressure variations to improve accuracy.

• Drone-based techniques

Drones enable spatially resolved GHG measurements using sniffer sensors or imaging instruments such as thermal infrared and hyperspectral cameras. The general processing steps are

1.

Preprocessing: Remove sensor noise, apply atmospheric corrections and calibrate raw data.

2.

Feature extraction:

−

For sniffer sensors: identify concentration peaks to locate emission sources.

−

For imaging systems: detect spectral or radiometric features indicative of GHGs.

3.

Concentration mapping: Estimate concentrations across the area surveyed, with corrections for surface emissivity (thermal) or spectral unmixing (hyperspectral).

4.

Validation: Compare drone-derived results with ground-based observations or model simulations for quality control.

• Satellite-based observations

Satellites observe atmospheric columns and retrieve GHG concentrations using more complex algorithms, due to the need to account for atmospheric scattering and surface reflectance:

1.

Spectral calibration: Compare measured spectra with synthetic ones using radiative transfer models, accounting for molecular absorption, pressure and temperature [35,36,37].

2.

Albedo corrections: Adjust for surface reflectivity to reduce bias in retrieved gas concentrations [36].

3.

Concentration retrieval: Apply inversion methods to derive gas concentrations from spectral radiances.

4.

Validation: Assess accuracy using ground-based or airborne measurements.

While a detailed review of retrieval physics is beyond the scope of this work, we note that most GHG retrievals rely on optimal estimation [38], with forward radiative transfer models such as the linearised discrete ordinate radiative transfer (LIDORT) model and the optimal spectral sampling (OSS) model used to simulate spectra, followed by inversion schemes to retrieve gas concentrations from top-of-atmosphere radiances. Benchmark retrieval descriptions for methane and CO₂ include those for TROPOMI [36,39], OCO-2/OCO-3 [40,41] and GOSAT [42]. Retrievals designed for point-source detection in industrial settings, including CCUS facilities, typically enhance prior-error characterisation and plume-scale regularisation [43,44].

4.2. Emission Estimation

Although retrieval algorithms provide concentration measurements at various spatial scales depending on the platform, in most practical applications the quantity of interest is the GHG emission flux. These fluxes are estimated using inversion techniques that compare observed concentrations with those predicted by a model.

The general relationship is expressed by

$$\mathbf{y}=\mathbf{F}\left(\mathbf{x}\right)+\epsilon$$

(1)

where $\mathbf{y}$ is the observed data vector, $\mathbf{x}$ is the state vector (e.g., GHG fluxes) and $\mathbf{F}$ is the forward model linking fluxes to concentrations. The term $\epsilon$ captures measurement and model uncertainties. The forward model $\mathbf{F}$ is typically a chemical transport model (CTM), several of which are available.

4.2.1. Chemical Transport Models

CTMs simulate the atmospheric transport of gases by solving the governing equations for advection, diffusion and, where relevant, chemical transformations. In the context of CCUS monitoring, CTMs provide the forward operator linking emissions from a facility to downwind concentration measurements. By representing meteorology, turbulence and boundary-layer processes, CTMs enable the prediction of concentration fields from assumed emissions, forming the basis for both forward simulations and inverse estimation in top-down verification frameworks.

CTMs can be grouped into three general categories:

• Eulerian models: These compute gas concentrations on a fixed grid, considering the transport, transformation and removal processes. Examples include WRF-Chem [45] and GEOS-Chem [46].
• Lagrangian models: These follow air parcels or particles through the atmosphere. FLEXPART is a Lagrangian particle dispersion model that can be run in both forward (emission-to-concentration) and backward (receptor-to-source sensitivity, i.e., “footprint”) modes, which makes it well suited for source–receptor analyses in CCUS settings [47,48,49]. In backward mode, FLEXPART produces sensitivity fields ($\mathbf{H}$) that can be used directly in linear Bayesian inversions.
• Plume models: These assume a Gaussian distribution of emissions from point sources and are often used in local-scale applications [50].

4.2.2. Data-Assimilation Frameworks for Inverse Emission Estimation

Inverse estimation of emissions is most naturally framed in a Bayesian context, where prior knowledge about the emission field is updated using new observations while accounting for the uncertainties of both. This approach provides a consistent statistical foundation for combining models and measurements. Within this framework, two main families are commonly applied to atmospheric trace-gas studies: (i) variational methods and (ii) ensemble-based Kalman filters. A third, related computational strategy uses source–receptor or influence–function relationships to approximate how emissions affect observed concentrations. The following paragraphs summarise the key principles and relationships among these approaches [38,51,52,53,54,55,56].

• Bayesian formalism.

In a Bayesian framework, emission estimates are updated by optimally combining prior information with new atmospheric observations through a model that links emissions to measured concentrations. The result is a best estimate—often termed the maximum a posteriori solution—that represents the most probable emission field given the available data and their uncertainties. This formulation provides the statistical basis for both variational and ensemble data-assimilation methods.

• Variational methods.

Variational approaches determine the emission field that minimises the overall mismatch between model predictions and observations, subject to prior constraints. The four-dimensional variational (4D-Var) method extends this by explicitly accounting for the temporal evolution of the atmosphere, allowing information from observations at different times to influence the estimate simultaneously. Variational methods are well suited to problems involving large-state vectors and continuous physical constraints, but they depend on tangent-linear and adjoint models of atmospheric transport and require careful specification of background error statistics [52,53].

• Ensemble Kalman filters.

Ensemble-based methods use a collection of model realisations or ensemble members, to represent the uncertainty in both the prior and the observations. Each member evolves according to the atmospheric transport model, and their collective behaviour provides an estimate of how uncertainties propagate in time. When new observations become available, the ensemble is updated to produce a revised mean and covariance that better match the data. Ensemble Kalman filter techniques are particularly attractive for large, nonlinear systems because they avoid the explicit computation of adjoint models and provide dynamic estimates of uncertainty [54,55].

• Influence–function formulations.

For many CCUS applications, the relationship between emissions and observed concentrations can be approximated as linear. In this case, so-called source–receptor or influence–function methods describe how emissions from each source location contribute to concentrations at each observation point. These sensitivities can be derived from adjoint or backward Lagrangian transport models. Such formulations are mathematically equivalent to Bayesian optimal-estimation solutions but are often computationally more efficient for large problems. They are therefore best viewed as a practical implementation within the broader Bayesian data-assimilation framework.

• Practical guidance for CCUS.

The choice among variational, ensemble and influence–function approaches depends on several factors: the availability of adjoint models or ensembles, the number and type of emission sources to be resolved, the required response time (e.g., near-real-time leak detection versus periodic verification) and the degree of nonlinearity in the atmospheric model. For most CCUS monitoring scenarios, where atmospheric transport is approximately linear over the analysis period, influence–function methods combined with Bayesian regularisation provide a robust and computationally efficient solution. For dynamic operations or dense observation networks, variational or ensemble techniques offer additional flexibility and robustness.

5. Validation and Integration with National Inventories

Carbon budgets play a central role in the validation and calibration of GHG inventories, which are fundamental tools for countries to report progress under international agreements such as the Paris Agreement and the United Nations Framework Convention on Climate Change (UNFCCC). These inventories aim to quantify anthropogenic emissions and removals of CO₂, CH₄, nitrous oxide (N₂O) and fluorinated gases, based on sectoral activity data and emission factors. However, ensuring their accuracy and consistency remains a significant challenge.

The UNFCCC encourages countries to adopt both bottom-up and top-down approaches to validate their inventories [57,58]. The bottom-up method involves cross-checking national data against independent inventories such as the Emission Database for Global Atmospheric Research (EDGAR) [59], which compiles globally consistent emission estimates from energy, agriculture, waste and industrial sectors. These comparisons allow countries to identify potential gaps or misclassifications in activity data and emission factors.

In contrast, the top-down approach estimates emissions by inverting atmospheric concentration observations to infer fluxes that best reproduce measured mixing ratios [6,60,61,62]. These methods combine satellite, airborne and in situ data with atmospheric transport models to provide observation-based constraints on regional and facility-scale greenhouse gas fluxes.

In contrast, the top-down approach compares inventory estimates to atmospheric concentration measurements using inverse modelling. This method incorporates direct observational data from in situ stations or satellite instruments (e.g., GOSAT, OCO-2) and links them to surface emissions through atmospheric transport models. The UNFCCC provides a framework for this approach, recommending a sequence that includes (1) acquisition of observations, (2) preparation of gridded prior data, (3) inverse modelling, (4) quality assurance and (5) comparison with reported inventories.

A particularly important frontier in this integration effort is the treatment of biogenic emissions, which are driven by natural processes such as wetland emissions, forest carbon uptake and microbial soil activity. These fluxes can be substantial and variable, and failure to distinguish them from anthropogenic sources can lead to significant over- or underestimation in national inventories. Models such as dynamic global vegetation models or land surface schemes are essential for resolving these contributions and improving inventory robustness.

Despite the importance of top-down validation, only a handful of countries, notably Switzerland, the United Kingdom and Australia, have implemented such systems at a national scale. The reasons for this limited adoption include the technical complexity of inversion methods, limited access to high-resolution atmospheric data and uncertainties in transport modelling. There is a clear need for capacity building and standardisation of methods to support broader uptake.

At the global level, initiatives such as the Global Carbon Project provide valuable benchmarks by synthesising data from bottom-up inventories and atmospheric inversions. Their annual global carbon [63] and methane budgets [6] incorporate results from more than a dozen inversion models and compare them against established datasets like EDGAR, Community Emissions Data System (CEDS [64]), (USEPA [65]) and Air pollutant INteractions and Synergies (GAINS). In the case of CO₂, there is general agreement between net land and ocean sinks and top-down atmospheric inversions for the period 2013–2022 [63]. However, for CH₄, the discrepancies remain substantial. The most recent methane budget indicates a +30% difference between top-down and bottom-up estimates for the period 2008–2017 [6], likely driven by double-counting of natural sources and challenges in scaling up regional observations.

The persistent divergence between independent model estimates and official inventory data underscores the urgent need for harmonised approaches. For CCUS applications, this implies that national inventories must be capable of incorporating concentration-based data from sensors, drones and satellites, into their reporting frameworks. Establishing procedures for aligning observed emissions with inventory categories (e.g., IPCC sectors), including the uncertainty quantification and attribution of sources, will be critical. As monitoring technologies mature, especially those capable of identifying fugitive emissions from industrial processes or geological storage, these validation approaches must become integral components of national reporting systems.

Ultimately, the credibility of national inventories and by extension the transparency of international climate commitments, depends on their alignment with independent observational data. Developing robust frameworks for integrating bottom-up and top-down information, and for incorporating CCUS-specific emissions and removals, will be key to meeting the reporting and verification challenges of the coming decades.

6. Challenges and Gaps in Current Monitoring Practices

Despite significant advances in GHG monitoring technologies and modelling frameworks, key challenges persist in implementing effective, scalable and verifiable measurement systems for CCUS. These span technical, operational and regulatory domains and limit the ability of current systems to deliver robust verification and accountability.

A central issue is the lack of harmonised MMV standards across jurisdictions and CCUS facility types. Frameworks such as the EU MRR provide tiered requirements for CEMSs yet many national laws offer only general guidance, particularly for geological storage, where terms like “zero leakage” are rarely operationalised with defined detection thresholds. This ambiguity undermines the development of consistent monitoring plans.

Regulatory incoherence between climate legislation and carbon markets further complicates monitoring. Precision thresholds mandated for emissions trading (e.g., under the EU ETS) are often more detailed than those in environmental law, leaving CCUS operators uncertain about how to satisfy both legal compliance and offset verification requirements.

Technically, many monitoring systems still lack defined detection thresholds, whether in concentration units, resolution or frequency. This makes it difficult to determine whether emissions are within acceptable limits, particularly when dealing with diffuse, low-rate or intermittent leaks.

Even when elevated GHG levels are detected, source attribution remains a significant challenge. Differentiating anthropogenic signals from biogenic or background variability requires high-resolution data, detailed site characterisation and reliable inverse modelling. For instance, BECCS facilities near agricultural zones often experience attribution uncertainty, potentially leading to both false positives and undetected leaks.

Another barrier is the fragmentation of observation platforms. In situ sensors, drones and satellites are often operated in isolation, without coordinated data fusion or cross-calibration. This “siloed” monitoring approach prevents full exploitation of spatial and temporal synergies, undermining the effectiveness of GHG assessments across different scales.

While inversion techniques and data assimilation have strong potentials in improving source attribution and integrating multiple data streams, their use in operational settings is limited. High computational costs, a lack of regulatory validation and the absence of standardised models all pose obstacles to broader adoption.

Cost and scalability constraints also limit the uptake of advanced technologies like hyperspectral imaging, LiDAR and 3D seismic surveys. Many smaller CCUS projects cannot afford such systems, particularly during long-term or post-closure phases. Moreover, drone-based platforms remain constrained by regulatory restrictions requiring human pilots, which impedes continuous or high-frequency operation.

Finally, long-term monitoring protocols are often vague or absent. Few standards specify how frequently post-closure sites should be monitored, for how long or how to manage liabilities after project handover. The absence of mechanisms for ensuring monitoring continuity across ownership changes further erodes confidence in the permanence of stored CO₂, a critical requirement for carbon offsetting.

7. Recommendations for Advancing CCUS Monitoring Practices

To address these gaps, we propose the following actionable recommendations for enhancing the effectiveness and reliability of CCUS monitoring systems.

7.1. Adopt a Concentration-Based Monitoring Paradigm

We recommend shifting from purely emission-based accounting toward concentration-based monitoring. This approach prioritises direct measurements of CO₂ and CH₄ concentrations in and around CCUS facilities, rather than relying exclusively on inferred emissions or bookkeeping methods. Detection thresholds should be clearly defined in terms of measurable concentration increases above background levels.

As shown in Table 3, several techniques exist that allow the detection of CO₂ concentrations as low as 1–10 ppm and methane concentrations as low as 1–10 ppb. However, converting these concentrations into emission levels is, generally, not feasible, as the impact varies significantly depending on whether the emission occurs locally (e.g., at the injection point) or is fugitive over a large area (e.g., from a large geological storage site). One exception is the tracer gas technique, where detection limits of 5–25 kg/h for carbon dioxide and 0.2–0.5 kg/h for methane can be obtained, depending on wind speed and distance to the source. We suggest overcoming these limitations by defining measurable leakage in terms of the increase in concentration it causes, rather than by the emission rate itself.

7.2. Harmonise Regulatory and Carbon Market Requirements

Establish consistent requirements across environmental regulation and emissions trading schemes. Clearly define terms like “zero measurable leakage” and ensure that monitoring obligations under climate law and offset markets are mutually compatible.

By adopting background levels from the National Oceanic and Atmospheric Administration for 2024 (420 ppm for CO₂ and 1900 ppb for CH₄) and applying a 10% precision requirement aligned with current standards such as the EU MRR CEMS Tier 1 standard together with a $4\sigma$ detection limit, we derive the following detection thresholds. A 10% precision implies that measurements are accurate within 10% of the background value. Multiplying this margin by 4 yields an allowable increase of 168 ppm for carbon dioxide (i.e., $4\times 0.10\times 420=168$ ppm) and 760 ppb for methane (i.e., $4\times 0.10\times 1900=760$ ppb). Adding these increments to the background concentrations gives thresholds of 588 ppm for CO₂ and 2660 ppb for CH₄.

These thresholds offer a balance between technical feasibility, cost-effectiveness and environmental protection. They also provide a concrete, standardised definition of measurable leakage that is adaptable to various monitoring platforms.

7.3. Integrate Multi-Platform Observations into Unified Monitoring Systems

Break down platform silos by requiring cross-calibration and coordinated operation of in situ, drone and satellite systems. These different technologies offer complementary strengths: in situ sensors provide high temporal resolution and accuracy, drones offer spatial flexibility and localised detection and satellites ensure wide-area and regular coverage. Yet, in many cases these systems operate independently, missing opportunities for data synergy.

We recommend that monitoring plans mandate co-location campaigns to compare instrument performance under controlled conditions and establish data fusion protocols for combining observations with different temporal and spatial characteristics. Metadata standardisation, including uncertainty characterisation and sensor performance metrics, is essential for building interoperable monitoring systems that can deliver reliable results across regulatory and research domains.

7.4. Operationalise Data-Assimilation Frameworks

Support the routine use of data-assimilation frameworks to integrate heterogeneous observations and infer spatio-temporal GHG concentration fields around CCUS facilities. Unlike traditional inversion models that estimate emissions directly, data-assimilation frameworks prioritise the reconstruction of GHG concentration distributions by updating prior state vectors using real-time observations.

This approach is particularly well suited to a concentration-based monitoring paradigm, where compliance thresholds are defined in terms of measured values rather than inferred emissions. By continuously integrating data from in situ, drone and satellite sources, data-assimilation systems can identify and track anomalies in near real time, improving responsiveness to potential leak events.

7.5. Develop Monitoring Standards Across Facility Lifecycles

Ensure that monitoring obligations extend beyond the operational phase of CCUS projects. Currently, many regulations focus on active injection periods, with vague or insufficient guidance on post-closure monitoring. This poses long-term risks, especially given the geological timescales involved in CO₂ storage.

Standards should define post-closure monitoring frequency, acceptable risk thresholds for leak detection and the duration over which monitoring should be maintained. Importantly, they must also clarify who holds the monitoring and remediation responsibilities as facilities transition between ownership or regulatory frameworks.

7.6. Embed CCUS Monitoring into National Inventory Validation

Incorporate satellite and ground-based data into top-down national GHG inventory validation frameworks, in line with UNFCCC recommendations [57,58]. While bottom-up inventories are vital for understanding sectoral emissions, they are often based on static emission factors and incomplete activity data. Top-down approaches—using atmospheric observations and assimilation methods—offer a way to cross, check and improve the accuracy of national reporting.

We recommend that national authorities establish formal processes for integrating top-down estimates into inventory quality assurance, especially for sectors involving large point sources or diffuse emissions, such as CCUS and agriculture. This integration should be accompanied by methodological transparency, including uncertainty estimates and documentation of modelling assumptions.

8. Conclusions

CCUS technologies represent a critical pathway toward achieving climate targets, but their environmental legitimacy depends on credible and transparent monitoring. This review has highlighted persistent gaps in current MMV systems, including undefined detection thresholds, inconsistent regulatory standards and underutilised observational technologies.

By adopting a concentration-based monitoring framework, harmonising regulations and embracing data integration and inversion methods, the CCUS community can build a more robust and trustworthy verification ecosystem. Our proposed recommendations provide a clear strategy to align monitoring practices with climate goals.

Emerging innovations, including autonomous systems, AI-driven analytics and low-cost sensors, offer exciting opportunities to reduce costs and increase responsiveness. Realising their full potential, however, requires regulatory openness, investment in technical capacity, and a shared commitment to long-term stewardship.

In summary, current retrieval and inversion frameworks for CCUS monitoring exhibit complementary strengths. Optimal-estimation algorithms such as LIDORT and OSS remain the most robust for satellite-based column retrievals, while ensemble Kalman filters and variational approaches (4D-Var) have proven effective for regional- to facility-scale inverse modelling. Source–receptor (influence–function) techniques offer computational efficiency and remain well suited for near-real-time verification. Nevertheless, important gaps persist: shortwave infrared (SWIR) data for CO₂ and CH₄ remain underutilised in industrial contexts; synergistic use of multi-angle and hyperspectral sensors is still limited and quantitative error propagation between retrievals and emission estimates is often overlooked. Future research should focus on benchmarking inversion frameworks across standardised CCUS test cases and integrating multi-platform data into unified, traceable monitoring systems.

Ultimately, the success of CCUS in delivering durable climate benefits hinges not only on capturing carbon but on proving it, with precision, consistency and transparency. Strengthening MMV frameworks is both a technical and ethical imperative in the global mitigation effort.