I reviewed the manuscript entitled “Methane Quantification of an LNG Gas-Fired Power Plant in Seoul, South Korea” with great interest. The study addresses an important knowledge gap by providing observational constraints on methane emissions in an understudied region. The authors conducted six months of mobile measurements and applied top-down (TD)approaches using a Gaussian Plume Dispersion Model (GPDM) and the OTM-33A method to quantify emissions. In addition,bottom-up (BU) estimates were used and compared with TD results. This integrated approach is valuable and timely.

However, while the study is promising, the manuscript requires substantial technical strengthening before it can be considered for publication. Several key methodological assumptions require clearer justification, the uncertainty analysis mustbe significantly expanded, and the comparison between TD and BU estimates needs stronger conceptual and quantitative support. Major and detailed comments are provided below.

Major Comments

The major concerns relate to the following aspects:

Application of the GPDM and OTM-33A methods

Propagation and reporting of uncertainties

Comparison between TD and BU quantification; an inconsistency requires justification

Upscaling of short-term measurements to annual emissions

Attribution of detected emissions to specific industrial activities

Use of terminology, particularly “fugitive emissions”

Consideration of financial losses associated with methane emissions

Points 1 & 2: Applicability and Uncertainty of GPDM and OTM-33A

The Gaussian plume dispersion method relies on several strict assumptions, including flat terrain, homogeneous wind fields, steady-state atmospheric conditions, and a single well-defined point source. For OTM-33A, a consistent wind direction within ±30°, ideally for more than 80% of the sampling period, is additionally required. Measurement distance from the source is alsoa critical parameter.

The manuscript does not adequately specify:

The extent to which these assumptions were met,

How deviations from these conditions may have influenced the emission estimates,

How associated uncertainties were propagated into the final

The authors must explicitly report all measured parameters that influenced the final quantification (e.g., wind speed, wind direction variability, source-receptor distance, stability class) and quantitatively assess their contributions to total uncertainty.

We have significantly revised Section 2.2 and Section 2.3 in method part to explicitly describe our instrument calibration, data screening criteria, parameter determination, and uncertainty quantification methods.

We added the 2.2.1 Instrument calibration subsection, and revised as follows: “To ensure the accuracy of the measurements, a two-point calibration for LI-7810 and one-point calibration for GLA131-MEA were performed prior to each daily measurement. We conducted a two-point (zero and span) calibration for the CH₄ channel using CH₄-free zero air (0 µmol/mol; 0 ppm) and a certified CH₄ span standard of 2.03 µmol/mol (2.03 ppm). For the GLA131-MEA, we performed a zero and span calibration using certified reference gases through the instrument’s built-in calibration routine (reference-gas calibration). All reference gases used for calibration were Korea Research Institute of Standard and Science (KRISS)-certified reference gas mixtures, providing metrological traceability and international recognition under the CIPM MRA and produced within KRISS’s ISO/IEC 17025 and ISO 17034 quality system”.

We added the specific screening criteria to the “Quantification of methane emissions from ground-based top-down approaches” section of 2.3: “Strict data quality indicators (DQIs) were applied to ensure the reliability of the OTM-33A estimates. While standard guidelines recommend a sampling duration of 15–20 minutes from the U.S. EPA, we extended the stationary measurement period to 30 minutes for sites characterized by complex terrain or dense vegetation. This extension was implemented to ensure sufficient plume event accumulation and to mitigate the impact of near-field obstructions on plume transport. Data were strictly screened based on the following criteria: (1) the plume center was captured within ±30° of the mean wind direction, (2) the in-plume methane concentration exceeded 0.1 ppm, and (3) the Gaussian curve fit exhibited an R² > 0.8”.

We clarified data source for the Pasquill-Gifford stability classes to justify the atmospheric conditions used and revised in the section 2.3 as follows: “The stability classes were determined based on hourly average wind speed and solar altitude angle data obtained from the Korea Astronomy and Space Science Institute (KASSI)”.

We added more specific descriptions for uncertainty analysis to section 2.3: We quantified the uncertainty of methane emission rates using a Monte Carlo simulation with 10,000 iterations. Given that the target LNG power plant is situated in a complex environment that deviates from the ideal conditions of flat terrain and homogeneous wind fields, we assigned robust uncertainty ranges to key input parameters. Specifically, uncertainties were set at ± 10% for the source-receptor distance; ± 20% for wind speed to account for surface roughness and turbulence; and ± 30% for the source height, reflecting the uncertainty in estimating vents from underground facilities”.

Points 3 & 4: BU–TD Comparison and Annual Upscaling

The BU approach is inherently designed for annual emission estimation, whereas the TD approach in this study reflects snapshots during limited sampling periods. It appears that the BU estimates were applied to the temporal window of themeasurements, rather than being strictly annualized.

The manuscript must clearly explain:

How the TD annualized estimates were consistently compared with BU

From Table 3, inventory values are, in several cases, greater than the measured TD values for S1, S2, and in three out of five cases for S3. This contradicts the authors’ conclusion that inventories underestimate emissions. The data currently presented suggest the opposite. This inconsistency must be addressed. Additionally, annual variability in emissionsmust be discussed, as it directly affects the validity of BU–TD comparisons.

How short-term TD emission estimates were upscaled to annual values,

Whether temporal variability (seasonal, operational, maintenance-related) was accounted for,

We have revised Section 3 Results and Discussions to clarify our comparison methodology and to address the apparent inconsistencies in Table 3.

While the detailed methodology for downscaling the inventory is provided in the Methods section, we recognized that the comparison in the Results section could lead to confusion regarding the time scale. To explicitly clarify this and avoid any misunderstanding, we added the following sentence to the Section 3 (Results and Discussion): “To enable a direct comparison, the bottom-up annual emission estimates were converted into hourly emission rates (kg hr^-1) based on the monthly fuel consumption and operating hours corresponding to the measurement period”.

To address inconsistency between the TD and BU at S1 and S2, we have added sentences to Section 3: “The substantial discrepancy between the measured emission rate and the bottom-up inventory stems directly from the limitations of the input data described in the methodology. While we utilized 5-minute operational proxies (MW), the underlying activity data relied on 'monthly aggregated LNG consumption.' This temporal resolution inevitably smooths out instantaneous high-emission events. Furthermore, the applied emission factor (87.5 tCH₄ PJ-6) accounts only for routine 'post-meter leakage' under normal operating conditions and does not capture stochastic super-emitter events—such as incomplete combustion during start-ups or pressure relief venting—which likely caused the elevated plumes detected at the S3 hotspot”.

We explicitly discussed the limitations of using "monthly aggregated data," noting that it inevitably smooths out instantaneous variability found in our measurements. Furthermore, regarding the annual upscaling appearing in the newly added Section 3 (Results and Discussions), we would like to clarify that this was introduced specifically to address the reviewer’s valuable suggestion on estimating "financial impacts" (Point 7). To provide a meaningful estimate of the potential economic loss as requested, we extrapolated the maximum observed emission rate to an annual scale. However, to ensure scientific rigor, we clearly specified that this figure represents a "potential magnitude" or an upper-bound scenario assuming the super-emitter event persists, rather than a definitive claim of continuous annual emissions. The revised sentences were added to the end of section 3 (Results and discussions): “Based on the maximum emission rate at S3 (2,053 ± 283 kg hr-1), we estimated the potential annual financial loss. Using the lower heating value (LHV) of methane and the power generation charge rates from the Korea Gas Corporation (KOGAS), the estimated annual loss amounts to approximately 15.77 million USD. This calculation assumes that the maximum methane emission event occurs continuously, serving as an upper-bound estimate of the potential revenue loss”.

Point 5: Source Attribution

The manuscript provides insufficient methodological detail on how emissions were attributed to specific industrial activities. This attribution is central to the interpretation of results and to the applicability of BU estimates

More detailed description of for the source attribution were added to the section 2.5 (Measurement Strategy): “Due to strict security protocols restricting access to most LNG power plants in South Korea, indirect measurement strategies are often required. The target facility in this study is an underground power plant. Accordingly, we conducted measurements primarily at the surface level (ground level) where access was permitted. Our strategy involved repeated mobile transects combined with targeted walking surveys to ensure the reliability of source identification. We defined a location as a potential emission source only when elevated methane concentrations were consistently detected at the same spot on at least two or more separate occasions”.

Point 7: Financial Impacts of Methane Loss

Beyond environmental and health impacts, methane loss also represents a direct economic loss, particularly for LNG-importingcountries. The manuscript would be significantly strengthened by:

Estimating annual financial losses associated with both intentional and unintentional methane emissions,

Clearly separating fuel loss mechanisms,

Discussing implications for national energy security and

Financial Impacts of Methane Loss were added to the end of section 3 (Results and Discussions): “The fugitive methane emissions identified in this study also represent a significant economic loss. Based on the maximum emission rate at S3 (2,053 ± 283 kg hr-1), we estimated the potential annual financial loss. Using the lower heating value (LHV) of methane and the power generation charge rates from the Korea Gas Corporation (KOGAS), the estimated annual loss amounts to approximately 15.77 million USD. This calculation assumes that the maximum methane emission event occurs continuously, serving as an upper-bound estimate of the potential revenue loss. This substantial economic cost underscores the necessity and cost-effectiveness of implementing stricter leak detection and repair (LDAR) protocols to minimize fugitive emissions. The annual financial loss was calculated using Eq. (5)”.

Detailed Comments Abstract

The abstract should be more specific and quantitative, highlighting key numerical findings and methodological outcomes rather than relying on general statements.

The abstract was revised based on the reviewers' comments and the changes marked in red on the revised manuscript.

Introduction

Additional literature review is required to better contextualize methane emissions from industrial sites.

In the result and discussion you can compare quantifications of methane emissions from the LNG site with other sources in urban area of Seoul.

Additional literature reviews were added to Introduction section: “In particular, recent studies on industrial methane emissions have consistently revealed a substantial discrepancy between bottom-up inventories and ground-based top-down atmospheric measurements. Rutherford et al. (2021) showed that bottom-up inventories frequently underestimate emissions from global oil and gas infrastructure because they depend on fixed factors that do not account for stochastic "super-emitter" events. Such discrepancies are well-documented in upstream production fields (Alvarez et al., 2018), but downstream industrial facilities, such as LNG power plants and distribution networks, are also prone to substantial fugitive emissions due to equipment leaks and operational venting (Weller et al., 2020; Plant et al., 2022). Despite their critical role in urban environments, these downstream sites lack facility-level monitoring”.

In the results and discussions section, we added a detailed comparison between our results and other sources in Seoul from the 2023 Seoul Greenhouse Gas Inventory. These were revised as follows: “To further contextualize the magnitude of these emissions, we compared our findings with the official 2023 GHG inventory of Seoul (SCNSC). The reported annual methane emissions from the entire energy fugitive sector were approximately 4,464 tons, and the wastewater treatment sector accounted for approximately 1,607 tons. Although emissions at S3 hotspot are likely intermittent, extrapolating the average measured rate of 550 kg h^-1 to an annual scale yields a potential magnitude of approximately 4,818 tons. Notably, this potential figure from a single hotspot exceeds the total reported fugitive emissions for the entire Seoul and is nearly three times higher than the emissions from the wastewater treatment sector alone. This comparison illustrates how current bottom-up inventories may significantly underestimate urban methane budgets by failing to account for the intensity of industrial super-emitters”.

Materials and Methods

Key methodological steps require substantial clarification and

To improve the logical flow and clarity, we have restructured Section 2.2 into distinct subsections: 2.2.1 Instrument calibration and 2.2.2 Data preprocessing.

The background subtraction method is not described and must be explicitly

We explicitly added the calibration protocols and the use of certified reference gases to the section 2.2.1: “To ensure the accuracy of the measurements, a two-point calibration for LI-7810 and one-point calibration for GLA131-MEA were performed prior to each daily measurement. We conducted a two-point (zero and span) calibration for the CH₄ channel using CH₄-free zero air (0 µmol/mol; 0 ppm) and a certified CH₄ span standard of 2.03 µmol/mol (2.03 ppm). For the GLA131-MEA, we performed a zero and span calibration using certified reference gases through the instrument’s built-in calibration routine (reference-gas calibration). All reference gases used for calibration were Korea Research Institute of Standard and Science (KRISS)-certified reference gas mixtures, providing metrological traceability and international recognition under the CIPM MRA and produced within KRISS’s ISO/IEC 17025 and ISO 17034 quality system”.

L126: Frequency and protocol of instrument calibration must be

In the new Section 2.2.2, we provided a detailed description of the data quality control and the moving window method used for background determination. “To quantify the methane emissions from the LNG gas-fired power plant, the raw measurement data were first quality-controlled and time-synchronized. We corrected for the temporal lag between the GPS coordinates and the gas analyzer measurements, accounting for the residence time of air through the sampling inlet. The time-synchronized CH₄ concentration time series was merged with collocated GPS and meteorological data (wind speed and wind direction) to produce a point-by-point dataset. The CH₄ background concentration for each data point was defined as the median of measurements within a moving window of ±2.5 minutes centered on that timestamp (Weller et al., 2019; Maazallahi et al., 2020). Methane enhancement was then calculated by subtracting the estimated background from the measured concentration”

L158–169: Explicitly describe operational requirements and data screening criteria for GPDM and OTM-33A.

We have specified the strict data quality indicators (DQIs) used to select valid plumes (GPDM and OTM-33A) for analysis in Section 2.3. The detailed changes was described above (Major comments’s response 2) of point (1)).

L172: Clarify how atmospheric stability classes were

We clarified the data source for stability classification in Section 2.3. The detailed changes was described above (Major comments’s response 3) of point (a)).

L185: Allocating monthly consumption to 5-min intervals does not capture startup and shutdown behavior; discusslimitations and potential bias.

The detailed changes was described above (Major comments’s response 2) of point (b)).

L188: Justify the use of post-meter leakage emission factors for exhaust-based

The detailed changes was described above (Major comments’s response 2) of point (b)).

L194: Replace with “site access restrictions” for

Revised as” Due to strict security protocols restricting access to most LNG power plants in South Korea, indirect measurement strategies are often required”.

L205: Clarify whether this refers to the surrounding area of the power

We clarified the description of the surrounding environment to justify the background conditions and potential sources. The manuscript was revised as “The target LNG gas-fired power plant is situated in a complex urban environment surrounded by residential areas, the Han River waterfront park, and sewer networks. These factors act as potential sources of methane emission”.

L225–228: This section should be moved to Methods as a subsection on source attribution

We moved it to the section 2.5 and revised as “Our strategy involved repeated mobile transects combined with targeted walking surveys to ensure the reliability of source identification. We defined a location as a potential emission source only when elevated methane concentrations were consistently detected at the same spot on at least two or more separate occasions”.

Study Area Description

L113–117: If available, provide detailed information on:

LNG storage capacity,

Reported emissions from storage systems,

Number and type of exhausts,

Surrounding infrastructure affecting plume

Revised in the section 2.1 as “The plant operates as a combined cycle power plant (400 MW × 2 units), utilizing gas turbines and heat recovery steam generators (HRSGs). The plant is equipped with two main combustion exhaust stacks, each approximately 88 meters in height, which release emissions from the underground generation units (KOMIPO). LNG is supplied to the plant via pipelines from the Incheon LNG terminal, which is operated by the Korea Gas Corporation (KOGAS). Consequently, there are no on-site LNG storage tanks within the facility. The target LNG gas-fired power plant is situated in a complex urban environment surrounded by residential areas, the Han River waterfront park, and sewer networks. These factors act as potential sources of methane emission. To ensure accurate attribution, we conducted extensive preliminary surveys to characterize these sources before proceeding with the targeted quantification of the power plant's emissions”.

L129: GPS precision of ±20 m is coarse for GPDM applications; clarify whether this refers to horizontal coordinates or elevation.

The precision of GPS has corrected to position accuracy 3m (50% CEP, Circular Error Probability) in the manuscript and table.

Results and Discussion

Further clarification is required regarding:

Uncertainty ranges and their derivation,

This has been addressed in detail. Please refer to our response to Major Comment 1 (Uncertainty Analysis).

Annual upscaling and consistency with BU estimates,

This has been addressed in detail. Please refer to our response to Major Comment 3 & 4 (Comparison Methodology).

Interpretation of TD–BU discrepancies,

This has been addressed in detail. Please refer to our response to Major Comment 3 & 4.

Robust identification of emission sources,

This has been addressed in detail. Please refer to our response to Major Comment 5 (Source Attribution).

Estimation of financial

This has been addressed in detail. Please refer to our response to Major Comment 7 (Financial Loss).

L18–20: See major comment on source

Revised as “S1 and S2 were attributed to continuous fugitive methane leaks from underground pipeline, whereas S3 was identified as a stochastic super-emitter likely caused by incomplete combustion or pressure relief venting from the facility's exhaust”.

L20–21: Revise significant digits, clearly distinguish whether values are instantaneous or

We revised the emission rates to use appropriate significant digits to kg h^-1 for consistency.

L24–25: Place the root causes of TD–BU discrepancies

Revised as “This significant discrepancy arises because bottom-up inventories rely on aggregated monthly fuel consumption data and static emission factors. These inventories are unable to capture high-emission events like S3”.

L26–27: Justify why mobile measurements were particularly advantageous in the Seoul

Revised as “Our study shows that mobile measurement is an effective approach for identifying specific sources of methane emissions in Seoul, especially for facilities with strict access restrictions”.

L21–23: Provide deeper explanation of emission processes for S1–S3, especially

Please refer to our response to “L18–20: See major comment on source attribution”.

L25: Explicitly discuss limitations of BU

Please refer to our response to “L24–25: Place the root causes of TD–BU discrepancies here”.

L35–49: Overly long; tighten to focus on manuscript objectives.

We have deleted the sentence regarding China and Japan's LNG import statistics to streamline the Introduction and remove extraneous background information, as suggested. “China recently overtook Japan as the world's largest LNG importer, with imports reaching approximately 72 million tons, whereas Japan imported approximately 66 million tons in the same year (Union, 2024)”.

L48: Check citation

revised

L58–61: Clarify that uncertainties remain large for LNG-specific methane

Revised as “However, the accurate quantification of methane emissions from LNG power plants has proven challenging, and significant uncertainties remain regarding the magnitude of fugitive emissions (Lyon et al., 2015)”.

L80–81: Later use these values for cross-site

We have utilized these values in Section 3 (Results and Discussions) to contextualize our findings, specifically comparing our results with recent measurements from other LNG facilities (e.g., Jia et al., 2025).

L211–213: Avoid excessive decimal

Revised

L214: Methane enhancement alone does not prove source

Revised as “Methane enhancements from the mobile measurements indicate a maximum of 3,795.7 ppb (S1) with an average of 1,698.62 ppb, a maximum of 1,188.94 ppb (S2) with an average of 466.22 ppb, and a significant maximum of 56,039.06 ppb (S3) with an average of 19,963.97 ppb, identifying them as specific emission sources associated with the LNG gas-fired power plant”.

L233 onward: Reduce decimals; explain how such small uncertainty ranges were

Revised all emission rates and detailed descriptions for the uncertainty ranges described in section 2.3.

L247–250: Root-cause attribution appears speculative; consider intentional

This has been addressed in detail in our response to Major Comment 5 (Source Attribution & Super-emitters).

L251–272: Separate BU–TD comparison from variability

Please refer to our response to Major Comment 3 & 4 (Comparison Methodology) regarding the quantitative comparison, and Major Comment 6 (Super-emitters) regarding the interpretation of emission variability.

L268: Super-emitters must be defined consistently; include dispersed source contributions (Williams et , 2025).

Revised as “Williams et al. (2025) recently demonstrated that dispersed sources emitting less than 100 kg h-1 can collectively account for a majority of total methane emissions in oil and gas basins. In our study, while the super-emitter S3 dominates the total methane emissions, the routine leaks detected at S1 and S2 fall into this category of smaller, dispersed sources, highlighting the need for a comprehensive monitoring strategy that captures both heavy-tailed events and widely distributed smaller leaks”.

L289: Compare S3 with other documented super-emitting sites using flux-based

Revised as “Jia et al. (2025) quantified fugitive methane emissions from typical natural gas infrastructure components (e.g., valves, flanges) using on-site measurement techniques. They reported that the total annual emissions from an entire LNG terminal were approximately 5,202 kg year-1 (averaging 0.6 kg h-1), with individual components emitting significantly less. In comparison, the methane emissions identified in our study at sources S1 (90 ± 9 kg h-1) and S2 (18 ± 2 kg h-1) are orders of magnitude higher than these typical component-level methane emissions. These results suggest that the emissions at S1 and S2 are caused by high-pressure fugitive leaks or compromised infrastructure, which is different from the minor component-level leaks that are typical of standard distribution stations. Furthermore, Zimmerle et al. (2015) reported comprehensive transmission and storage (T&S) sector measurements. Their study determined the mean methane emission rates to be approximately 670 Mg year⁻¹ (equivalent to ~76.5 kg h⁻¹) for transmission stations and 847 Mg year⁻¹ (equivalent to ~96.7 kg h⁻¹) for underground storage facilities. When placed in this context, the emission rate at S1 (90 ± 9 kg h⁻¹) is comparable to the average emission load of an entire US transmission station, indicating a substantial facility-level emissions. In stark contrast, the maximum emission rate at S3 (2,053 ± 283 kg h⁻¹) is exceptionally high, exceeding the average emissions of these large-scale midstream facilities by a factor of over 20. While S1 and S2 represent significant localized leaks, S3 falls into the category of extreme super-emitters, exhibiting discharge rates far surpassing typical operational emissions found in standard natural gas infrastructure”.

L332–334: Focus conclusions on study results, not on method validation

Revised in conclusion section.

L339: Emphasize transparency

Revised in conclusion section.

L343: Expand on temporal variability and its

Revised in conclusion section.

L347–351: Discuss whether GPDM and OTM-33A assumptions were truly met in this complex

Revised in conclusion section: This study identified and quantified fugitive methane emissions from an urban LNG gas-fired power plant in Seoul using a mobile measurement platform. By integrating mobile measurements with GPDM and OTM-33A, we demonstrated that ground-based top-down approaches can effectively quantify emission sources even in complex urban environments where facility access is restricted. Our measurements revealed three distinct emission hotspots with varying characteristics. Sources S1 and S2 were identified as continuous fugitive leaks associated with underground pipeline infrastructure, exhibiting consistent emission rates of 90 ± 9 kg h-1 and 18 ± 2 kg h-1, respectively. In contrast, source S3 was characterized as a stochastic super-emitter, with a peak emission rate reaching 2,053 ± 283 kg h-1. The magnitude and intermittency of S3 suggest it is driven by operational venting or incomplete combustion, events that are typically invisible to standard leak detection protocols. A critical finding of this research is the substantial discrepancy between our ground-based top-down measurements and bottom-up inventory estimates. The high temporal variability observed at S3 demonstrates that emission factors used in current inventories fail to capture transient, high-intensity events. This disparity underscores the inherent limitations of current greenhouse gas inventory calculation methods, which often fail to account for fugitive emissions arising from specific, variable operational conditions. Furthermore, the estimated potential financial loss of approximately 15.77 million USD year-1 underscores the economic urgency of addressing these leaks.

In conclusion, relying solely on current inventories can lead to a significant underestimation of the methane emissions from urban energy infrastructure. We recommend implementing continuous methane monitoring systems and conducting regular mobile measurements to capture the full distribution of emissions, including those from heavy-tailed super-emitters. These transparent, measurement-based strategies are crucial for accurately assessing and effectively mitigating greenhouse gas emissions in the energy sector”.

Figures and Tables Figures

Figure 5a is

Axis labels in Figures 5b and 5c are unreadable; consider using ppm

The figure was revised.

Tables

Table 3: It is unclear whether inventory values apply to S1, S2, or combined Inventory values exceed measured values in several cases; this must be clarified and reconciled with conclusions

Revised

This study presents the measurements of an LNG facility in South Korea based on mobile transects and stationary measurements of methane concentration enhancements and Gaussian plume modeling to estimate methane flux rates. The paper is well written and the study design is well executed. The following comments should be considered and addressed to further improve the manuscript.

We deeply appreciate the reviewer’s positive evaluation of our study design and execution. We have carefully addressed all the constructive comments provided.

Consider reporting all quantification estimates in units of kg per hour rather than tonnes per hour. In addition, consider using no more than two significant figures in the reported quantification estimates, given the precision of the methods used here.

We agree with the reviewer’s suggestion regarding units. We have all emission rate throughout the manuscript from tons h^-1 to kg h^-1.

OTM-33A was previously tested and validated by Thoma et al. for specific methane emission rates, but field performance can vary depending on several factors, including meteorology, near-field obstructions, etc. Did the authors attempt a validation of the two methods here to better understand performance accuracy? If so, it would be great to report the results either in the main text or in the supplemental. Notably, the reported estimates of 2 tonnes per hour is well above the methane emission rates previously included in OTM-33A validation.

To ensure the reliability of our quantification in this complex environment, we applied strict Data Quality Indicators (DQIs) that meet or exceed the OTM-33A guidelines. The general description of DQI was added to the section 2.3: “Strict data quality indicators (DQIs) were applied to ensure the reliability of the OTM-33A estimates. While standard guidelines recommend a sampling duration of 15–20 minutes from the U.S. EPA, we extended the stationary measurement period to 30 minutes for sites characterized by complex terrain or dense vegetation. This extension was implemented to ensure sufficient plume event accumulation and to mitigate the impact of near-field obstructions on plume transport. Data were strictly screened based on the following criteria: (1) the plume center was captured within ±30° of the mean wind direction, (2) the in-plume methane concentration exceeded 0.1 ppm, and (3) the Gaussian curve fit exhibited an R2 > 0.8. We quantified the uncertainty of methane emission rates using a Monte Carlo simulation with 10,000 iterations. Given that the target LNG power plant is situated in a complex environment that deviates from the ideal conditions of flat terrain and homogeneous wind fields, we assigned robust uncertainty ranges to key input parameters. Specifically, uncertainties were set at ± 10% for the source-receptor distance; ± 20% for wind speed to account for surface roughness and turbulence; and ± 30% for the source height, reflecting the uncertainty in estimating vents from underground facilities”.

Include average distance from known source location in Table 3. The terminology “top-down” has come to mean observations of methane enhancements from above-ground sensors, such as remote sensing instruments on aircrafts and satellites. The use of the terminology here is technically correct in that the methane estimates account for atmospheric transport as opposed to “bottom-up” approaches that estimate emissions fluxes directly at the source. Nevertheless, the measurements are still conducted at the ground-level. To avoid confusion with true “top-down” approaches such as aircrafts and satellites, I’d suggest providing a clarification in the text that this method entails ground-based facility-scale measurements of methane enhancements and methane fluxes are estimated by accounting for atmospheric transport.

We have revised the text to explicitly use the term "ground-based top-down" throughout the manuscript.

It would be helpful to contextualize the emission rates reported here. That is, how do these emissions rates compare with typical facility-level emission rates at well pads or compressor stations or processing plants, where similar measurement methods are used?

We have expanded the results and discussions section by comparing them with flux-based studies of other natural gas facilities: “Jia et al. (2025) quantified fugitive methane emissions from typical natural gas infrastructure components (e.g., valves, flanges) using on-site measurement techniques. They reported that the total annual emissions from an entire LNG terminal were approximately 5,202 kg year-1 (averaging 0.6 kg h-1), with individual components emitting significantly less. In comparison, the methane emissions identified in our study at sources S1 (90 ± 9 kg h-1) and S2 (18 ± 2 kg h-1) are orders of magnitude higher than these typical component-level methane emissions. These results suggest that the emissions at S1 and S2 are caused by high-pressure fugitive leaks or compromised infrastructure, which is different from the minor component-level leaks that are typical of standard distribution stations. Furthermore, Zimmerle et al. (2015) reported comprehensive transmission and storage (T&S) sector measurements. Their study determined the mean methane emission rates to be approximately 670 Mg year⁻¹ (equivalent to ~76.5 kg h⁻¹) for transmission stations and 847 Mg year⁻¹ (equivalent to ~96.7 kg h⁻¹) for underground storage facilities. When placed in this context, the emission rate at S1 (90 ± 9 kg h⁻¹) is comparable to the average emission load of an entire US transmission station, indicating a substantial facility-level emissions. In stark contrast, the maximum emission rate at S3 (2,053 ± 283 kg h⁻¹) is exceptionally high, exceeding the average emissions of these large-scale midstream facilities by a factor of over 20. While S1 and S2 represent significant localized leaks, S3 falls into the category of extreme super-emitters, exhibiting discharge rates far surpassing typical operational emissions found in standard natural gas infrastructure”.