Pith. sign in

REVIEW 3 major objections 5 minor 3 references

Quantifying urban and landfill methane emissions in the United States using TROPOMI satellite data

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Satellite inversion finds US urban methane 80% above EPA inventory, with landfills as the main cause.

desk verdict The urban methane underestimate likely holds, but the most novel number—the factor-of-4 landfill underreporting from overestimated collection efficiencies—rests on a weak regression and needs stronger robustness checks before it can be cited. read the letter →

arxiv 2505.10835 v1 pith:JEDXP6KW submitted 2025-05-16 physics.ao-ph

Xiaolin Wang , Daniel J. Jacob , Hannah Nesser , Nicholas Balasus , Lucas Estrada , Melissa Sulprizio , Daniel H. Cusworth , Tia R. Scarpelli
show 3 more authors
Zichong Chen James D. East Daniel J. Varon
This is my paper · ORCID
classification physics.ao-ph
keywords methaneemissionsTROPOMIsatelliteinversionurbanlandfillsgreenhousegasinventorycollectionefficiencyinversemodeling
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Using 2022 TROPOMI satellite observations at 12 km resolution, the paper estimates methane emissions for 12 major US urban areas and compares them with the EPA Greenhouse Gas Inventory used as prior. It finds the inventory underestimates urban methane by 80% on average, with underestimates of 22% to 290% everywhere except Los Angeles and Cincinnati, where emissions are overestimated by 32% to 37%. It identifies landfills as the principal source of the bias: methane reported by landfills with gas collection systems using the recovery-first method is too low by a factor of four, because the reported gas collection efficiency (average 70%) is far above what the satellite inversion infers (average 38%). The result localizes the inventory gap to a single, actionable source type and shows that Los Angeles-level collection practices could cut urban landfill methane by a factor of four.

What carries the argument

The machinery is the Integrated Methane Inversion, a Bayesian analytical inversion of TROPOMI methane columns run at 12 x 12 km2 resolution over 3° by 4° domains for each city, using the blended TROPOMI+GOSAT product as observations and the gridded GHGI as prior. A transformation matrix W attributes posterior emissions to sectors and grid cells by prior proportions, and a set of selection criteria (averaging kernel sensitivity greater than 0.1, a single landfill exceeding 80% of prior cell emissions) isolates 53 individual landfills for comparison against GHGRP bottom-up models. The comparison hinges on equations (1) and (2) of the GHGRP recovery-first method, in which the collection efficiency CE is back-calculated from recovered methane; the paper recalculates CE from the posterior emissions, generating the 38% average that drives the factor-of-four bias.

What would settle it

Directly measure gas collection efficiency at a representative sample of the 44 non-Los Angeles landfills using tracer-release or eddy-covariance flux experiments; if the mean efficiency is near the reported 70% rather than the inferred 38%, the factor-of-four bias and the inventory correction derived from it would collapse.

Watch

Extended reading notes

Core claim

The central claim is a quantitative discrepancy: for 12 large US urban areas in 2022, the EPA GHGI underestimates annual methane emissions by 80% on average, and for most cities the underestimate ranges from 22% to 290%, while Los Angeles and Cincinnati are overestimated by 32% to 37%. The paper attributes the gap mainly to landfills, not to natural gas distribution as emphasized by earlier urban studies. Comparing the inversion with GHGRP facility reports for 53 landfills shows that landfills with gas collection and control systems using recovery-first reporting under-report emissions by a factor of about four; the implied mean collection efficiency is 38% (range 5–90%) against the 70% (range 40–87%) used in reports. Los Angeles landfills are the exception, with inferred collection efficiencies averaging 85% and near-consistent with their reported values, which the paper reads as evidence that operational practices there can inform wider mitigation.

Load-bearing premise

The load-bearing premise is that the 12-km posterior emissions for the 53 selected landfills are accurate enough to compare quantitatively with GHGRP reports, which depends on the selection thresholds and on the prior allocation of emissions to sectors within each grid cell.

Editorial extensions

If this is right

  • An EPA amendment that cuts assumed collection efficiencies by 10% starting in 2025 would still leave reported landfill emissions too low; the paper's mean implies a roughly 30% reduction is needed.
  • If non-Los Angeles landfills raised collection efficiency to the Los Angeles average of 85%, urban landfill methane emissions would fall by about a factor of four.
  • Downstream natural gas contributes less to the urban underestimate than previous ethane-based aircraft studies concluded, shifting mitigation attention toward waste management.
  • The sector-attribution method can be applied to future years and other cities as TROPOMI and its successors accumulate data, though product choice over dark surfaces changes results.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A straightforward extension would be to rescale all recovery-first GHGRP landfill reports outside Los Angeles by the inferred factor of about four, not just the 53 resolved landfills, which would raise national inventory totals.
  • The wide spread of inferred collection efficiencies (5% to 90%) suggests CE is not a single number but a function of cover management, weather, and maintenance, so reporting rules could benefit from empirical distributions rather than fixed cover-type defaults.
  • If landfill emissions are this dominant, city-level methane reduction pledges might be met more cheaply through gas collection upgrades at a small number of landfills than through pipeline replacement programs.
  • The dependence of the result on the TROPOMI product over dark, wet surfaces implies that applying the same method to other regions should include product intercomparison or independent validation before trusting sector splits.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper presents a 12x12 km2 inversion of TROPOMI methane column observations over 12 major US urban areas for 2022, using the Integrated Methane Inversion framework and the EPA GHGI as a prior. It reports that the GHGI underestimates total urban methane emissions by 80% on average, with city-specific underestimates of 22% to 290% and overestimates of 32% to 37% in Los Angeles and Cincinnati. It attributes most of the bias to landfills and further infers, from 53 individually resolved landfills, that recovery-first GHGRP reporting is too low by a factor of about 4 because the assumed gas collection efficiency (average 70%) is much higher than that inferred from the inversion (average 38%). Results are evaluated with independent surface-based observations in the Northeast Corridor and Los Angeles, and sensitivity is explored through an ensemble of six inversions and two satellite products.

Significance. If upheld, the finding that urban methane emissions are underestimated by 80% in the national inventory, and especially the factor-of-4 bias in recovery-first GHGRP landfill reporting, would have substantial policy relevance for US GHG monitoring, landfill regulation, and methane mitigation. The paper is methodologically strong in several respects: 12-km resolution is finer than most previous TROPOMI urban inversions, the IMI provides closed-form posterior error characterization, an ensemble of six inversions and two satellite products is used, independent tower and TCCON observations corroborate the posterior estimates, and the authors archive their urban emission data and shapefiles. The landfill-level quantitative claims, however, rest on weak per-cell averaging-kernel constraints and a high-scatter regression, so the central causal attribution requires additional evidence before it can be considered secure.

major comments (3)
  1. [Methods – Attributing posterior emissions to individual sectors and landfills; Fig. 4] The per-landfill factor-of-4 bias and the inferred mean collection efficiency of 38% rest on posterior emissions for grid cells selected with an averaging-kernel sensitivity threshold of only >0.1 and with one landfill contributing >80% of prior emissions. At sensitivities this low the posterior remains close to the prior, and since the prior is the GHGRP-derived inventory under test, the comparison in Fig. 4A is vulnerable to prior shrinkage. The reported R=0.29 (R2≈0.08) for the recovery-first regression indicates large scatter, and no per-landfill posterior uncertainties are shown, so the mean CE of 38% could be sensitive to a few high-leverage points. Please repeat the analysis at higher averaging-kernel thresholds (e.g., >0.3 and >0.5), report per-landfill posterior error standard deviations from the posterior covariance matrix, and provide leverage/influence diagnostics for the regression.
  2. [Results – Urban methane emissions; Methods Eq. (5)-(6)] Sector attribution in each 12-km grid cell is performed by applying the prior sectoral fractions through matrix W, so the headline conclusion that landfills are the principal cause of urban underestimates inherits the prior spatial allocation of sectors. Although posterior error correlations between landfill and other sectors are below 0.35, this does not test whether the within-grid-cell mix is observationally distinguishable. A sensitivity test using an alternative prior spatial distribution for the sectors, or a joint sector-state inversion, is needed to confirm that the posterior landfill attribution, and hence the 'landfills as principal cause' claim, is not an artifact of the prior apportionment.
  3. [Results – Urban methane emissions; Methods – IMI inversion framework] The headline 80% average underestimate and the city-specific range of 22% to 290% are reported without confidence intervals. The ensemble spread of six inversions is a useful sensitivity indicator but is not a statistical uncertainty, and the closed-form posterior error covariance computed by the IMI is not used to characterize urban-total uncertainties in the text. Please report posterior error standard deviations for each urban total and for the 12-city aggregate bias, in addition to the ensemble range, so that the significance of the claimed underestimate can be assessed.
minor comments (5)
  1. [Results, second paragraph] The sentence 'Our posterior estimate is 80% higher at 1.8 Gg a-1' should read '1.8 Tg a-1'; as written the unit is three orders of magnitude smaller than the stated prior total of 1.0 Tg a-1 and is internally inconsistent.
  2. [Main text and Methods, Eq. (1)] Equation numbering is duplicated: the GHGRP emission equation in the main text and the IMI cost function in Methods are both labeled Eq. (1). Renumber to avoid ambiguity.
  3. [Fig. 4A and associated text] Please specify that the reported R values are correlation coefficients from the reduced-major-axis regressions, state the number of points entering each regression (44 excluding Los Angeles for the recovery-first comparison), and provide confidence intervals for the slopes.
  4. [Fig. 4B and associated text] Clarify for the reader that the GHGRP average collection efficiency of 70% (range 40-87%) and the posterior-derived average of 38% (range 5-90%) both refer to the 44 landfills outside Los Angeles, not to the full set of 53 landfills.
  5. [Introduction and Results – comparison with previous studies] The discussion of disagreement with ethane-based aircraft studies would benefit from explicitly stating the different source attribution (natural gas vs. landfills) and the differing domain definitions in one sentence in the main text; currently the reader must infer these differences from the SI table.

Circularity Check

2 steps flagged · score 6.0 of 10

Urban totals are observationally constrained, but the landfill 'factor-of-4' and 38% collection-efficiency results reduce, by the paper's own W-matrix construction, to the prior sector shares rescaled by the grid-cell posterior correction; the causal attribution is therefore partially circular.

  1. self definitional [Results, 'Urban methane emissions'; Methods, 'Attributing posterior emissions to individual sectors and landfills', Eq. (5)-(6).]
    "We separate the contributions of individual sectors in our posterior estimates based on the relative contributions in the prior estimates for each 12-km grid cell. ... The relative contributions from the q individual sectors to the total posterior emissions in each grid cell are taken from the prior estimates to produce a matrix W (q×n). We then apply W to transform the posterior state vector (n×1) to the reduced state vector x_sector (q×1)."

    Equation (5), x_sector_post = W x_post, defines the posterior sector emissions as prior-relative sector shares times the posterior grid-cell totals. Consequently the posterior sector mix (62% landfill) and the statement that 'Landfills are the principal cause of urban emission underestimates' are not independently determined by TROPOMI; they are the prior's sectoral composition (59% landfill) rescaled by the total posterior correction. The causal attribution to landfills is a definitional consequence of the prior weighting matrix, not a first-principles prediction from the satellite observations.

  2. renaming known result [Results, 'Emissions from individual landfills'; Methods, 'Attributing posterior emissions to individual sectors and landfills'; Fig. 4.]
    "We limit our analysis to 53 landfills (43 with GCCS and 10 without) for which we have adequate sensitivity and which account for >80% of prior emissions in their 0.1°×0.1° GHGI grid cell. ... The recovery-first method is a factor of 3.7 lower than our estimate based on the regression slope (R = 0.29)."

    For a selected cell, the posterior landfill emission is (prior landfill fraction) × (posterior cell total), with prior landfill fraction >0.8. The comparison to GHGRP therefore reduces algebraically to the total cell posterior/prior ratio; the 'factor of 3.7' is the grid-cell TROPOMI correction relabeled as a facility-specific landfill underreport. The collection efficiencies in Fig. 4B are then obtained by solving Eq. (2) with that same ratio and assumed R, frec, OX, D, so the headline 'average 38%' collection efficiency is forced by the prior-ratio attribution rather than by an independent landfill-scale observation.

full rationale

The paper's central quantitative result that the 12 urban areas are 80% under the EPA GHGI is not circular: the posterior totals are observationally constrained by TROPOMI (DOFS 1.4-17, mean 7.9), and the improvement against independent Northeast Corridor tower and TCCON data provides external support. The circular content lies downstream, in attribution. The Methods define posterior sector emissions as W x_post, with W built from prior relative contributions, so the sector-level statements 'landfills are the principal cause' and '62% landfill' restate the prior's sectoral proportions under a total-emission rescaling. The individual-landfill factor-of-4 and the inferred collection efficiencies inherit this same construction: the >80% prior-dominance selection makes the landfill posterior essentially the total cell posterior, so the regression against GHGRP recovers the grid-cell correction under a landfill label. This does not make the total urban underestimate vanish, but it means the paper's principal causal attribution and the 38% collection-efficiency number are partly definitional rather than independently measured. Self-citations to Nesser et al. (7) and Balasus et al. (20) are used for context or data products and are not load-bearing to the derivation, so they do not raise the score further. Overall, one central attribution claim reduces by construction, giving partial circularity.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The central claims rest on the inversion setup (prior error choices, OX assumption) and on the spatial prior used for sector attribution. No new physical entities are introduced.

free parameters (4)
  • Prior geometric error standard deviation (sg) = 2, 3, 4 (ensemble members)
    Chosen by hand to build the inversion ensemble; affects posterior uncertainty, but the reported result is the ensemble mean.
  • Boundary condition error standard deviation (sb) = 5, 10 ppb (ensemble members)
    Chosen by hand for the inversion ensemble; affects posterior estimates and the reported uncertainty range.
  • Methane oxidation fraction (OX) = 0.1-0.35 (values from convention)
    Used in Eq. (1) and (2) to back-calculate collection efficiency from posterior emissions. The paper notes OX values reflect convention rather than site-specific measurements, and higher OX would lower the inferred CE. This directly affects the factor-of-4 and CE=38% claims.
  • Regularization factor (gamma) = 1
    Set to 1 because the state cost function was within the expected range; a modeling choice, not fitted to data.
assumptions (4)
  • domain assumption The blended TROPOMI+GOSAT retrieval is unbiased over the 12 urban areas after filtering of water and coastal pixels.
    The base inversion uses this product as the observation. The paper tests consistency against the operational TROPOMI product (Fig. S1) but cannot fully validate the machine-learning bias correction over all cities.
  • domain assumption GEOS-Chem at 12x12 km2 with GEOS-FP meteorology accurately represents urban methane transport and the sensitivity of columns to emissions.
    The inversion forward model relies on this; unresolved sub-grid transport could bias posterior emissions.
  • domain assumption The prior GHGI spatial distribution is accurate enough that sectoral attribution can be performed by applying prior fractions (W matrix) to posterior grid cells.
    The paper attributes posterior emissions to sectors using prior proportions; this is load-bearing for the 'landfills are the principal cause' claim. Error correlations for landfill vs gas are <0.35, but downstream gas vs wastewater are 0.45-0.87.
  • standard math Bayesian analytical inversion with lognormal priors yields the optimal posterior; the cost function and averaging kernel construction are standard.
    Underpins the IMI solution; not independently proved in this paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Quantifying urban and landfill methane emissions in the United States using TROPOMI satellite data." pith.science (2026). https://pith.science/paper/JEDXP6KW

@misc{pith2026250510835,
  author       = {Pith},
  title        = {Pith review of: Quantifying urban and landfill methane emissions in the United States using TROPOMI satellite data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JEDXP6KW}},
  note         = {Machine review of arXiv:2505.10835}
}
read the original abstract

Urban areas are major sources of methane due to population needs for landfills, natural gas distribution, wastewater treatment, and residential combustion. Here we apply an inversion of TROPOMI satellite observations of atmospheric methane to quantify and attribute annual methane emissions at 12x12 km2 resolution for 12 major US urban areas in 2022. The US Environmental Protection Agency Greenhouse Gas Inventory (EPA GHGI) is used as prior estimate. Our results indicate that the GHGI underestimates methane emissions by 80% on average for the 12 urban areas, with 22%-290% underestimations in most urban areas, except Los Angeles and Cincinnati where emissions are overestimated by 32%-37%. This is corroborated by independent surface-based observations in the Northeast Corridor and Los Angeles. Landfills are the principal cause of urban emission underestimates, with downstream gas activities contributing to a lesser extent than previously found. Examination of individual landfills other than in Los Angeles shows that emissions reported by facilities with gas collection and control systems to the Greenhouse Gas Reporting Program (GHGRP) and used in the GHGI are too low by a factor of 4 when using the prevailing recovery-first reporting method. This is because GHGRP-estimated gas collection efficiencies (average 70%, range 40-87%) are much higher than inferred from our work (average 38%, range 5-90%). Los Angeles landfills have much higher collection efficiencies (average 78% in GHGRP; 85% in our work) than elsewhere in the US, suggesting that operational practices there could help inform methane mitigation in other urban areas.

Figures

Figures reproduced from arXiv: 2505.10835 by the authors.

Figure 1
Figure 1. TROPOMI satellite observations of dry-column methane mixing ratios (XCH4) over US urban areas. Values are 2022 annual means from the blended TROPOMI+GOSAT product (20) on the 0.125° ×0.15625° (≈12 × 12 km2) inversion grid. White areas have no observations. Black rectangles outline the 3°×4° simulation domains used in the inversions for the 12 urban areas. These domains are shown in the lower panels with urban bounda… view at source ↗
Figure 2
Figure 2. Optimization of methane emissions in New York City with 12 × 12 km2 resolution by inversion of TROPOMI satellite observations for 2022. Panel A shows prior emission estimates for the inversion including both anthropogenic emissions from the gridded Greenhouse Gas Inventory (GHGI) (22) and natural sources. Also shown are contributions from urban sectors (B, C, D). Urban area totals are shown inset as the sum of grid … view at source ↗
Figure 3
Figure 3. Annual methane emissions in 12 US urban areas ordered by population (largest to smallest). Panel A shows urban population (grey bars) and per capita posterior emissions from our TROPOMI inversion for downstream gas and landfills. Panel B compares our posterior estimates for 2022 subdivided by sectors to the prior estimates from the US EPA GHGI and to previous studies. Vertical bars are error standard deviations of t… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Emissions from 53 urban landfills for which our TROPOMI inversion provides individual information, of which 43 are equipped with gas collection and control systems (GCCS). Panel A compares our posterior estimates for 2022 from the TROPOMI inversion to the GHGRP values …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

3 extracted references · 3 canonical work pages

  1. [15]

    J. D. Maasakkers, et al., Using satellites to uncover large methane emissions from landfills. Sci. Adv. 8, eabn9683 (2022). 16. D. P. Moore, et al., Underestimation of sector-wide methane emissions from United States wastewater treatment. Environ. Sci. Technol. 57, 4082–4090 (2023). 17. L. A. Estrada, et al., Integrated Methane Inversion (IMI) 2.0: an imp...

    work page 2022

  2. [30]

    Karion, et al., Methane Emissions Show Recent Decline but Strong Seasonality in Two US Northeastern Cities

    A. Karion, et al., Methane Emissions Show Recent Decline but Strong Seasonality in Two US Northeastern Cities. Environ. Sci. Technol. 57, 19565–19574 (2023). 31. J. R. Pitt, et al., Underestimation of Thermogenic Methane Emissions in New York City. Environ. Sci. Technol. (2024). https://doi.org/10.1021/acs.est.3c10307. 32. J. R. Pitt, et al., New York Cit...

    work page doi:10.1021/acs.est.3c10307 2023

  3. [46]

    Available at: https://ghgdata.epa.gov/ghgp/main.do [Accessed 15 April 2024]

    US Environmental Protection Agency, Facility Level Information on GreenHouse gases Tool (FLIGHT). Available at: https://ghgdata.epa.gov/ghgp/main.do [Accessed 15 April 2024]. 47. A. Karion, et al., Methane emissions estimates and related data sets for Washington DC and Baltimore, MD urban areas. National Institute of Standards and Technology. https://doi....

    work page doi:10.18434/mds2-3720 2024

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.