the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 29 Apr 2026
Methane fluxes from tropical wetlands of the Orinoco River Basin and their regional implications
Abstract. The Llanos del Orinoco, a vast tropical savanna floodplain in northern South America, plays a significant yet understudied role in the global methane (CH4) budget. This study synthesizes existing data to evaluate CH4 emissions from the region, highlighting the interplay between natural processes and anthropogenic influences. Top-down and bottom-up estimates for 2018 reveal annual CH4 emissions ranging from 3.27 ± 0.71 to 5.31 ± 2.50, with wetlands contributing 41–70 % of total fluxes. Seasonal variability follows precipitation patterns, with greater emissions occurring during the rainy season (April–October). However, discrepancies between global biogeochemical models and sparse field measurements underscore significant uncertainties, exacerbated by inconsistencies in inundation mapping and outdated local data. Anthropogenic activities, including oil extraction, livestock farming, and expanding rice cultivation, further modulate CH4 fluxes, though their impacts remain poorly quantified. Historical trends show declining precipitation and increasing temperatures, with models predicting more extreme weather events, potentially reducing wetland extent but favouring CH4 release during remaining inundated periods. Critical research gaps, including the need for updated field measurements, improved inundation mapping, and a better understanding of neglected habitats like peatlands and seasonal wetlands, are discussed. Addressing these gaps is essential for refining global and regional CH4 budgets and developing mitigation strategies. This work calls for integrated monitoring efforts to reconcile model disparities, assess land-use impacts, and predict responses to climate change, ensuring accurate representation of the Llanos del Orinoco in regional and global carbon cycle models.
- Preprint
(4725 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-2080', Anonymous Referee #1, 22 May 2026
-
AC1: 'Reply on RC1', Joao Henrique Fernandes Amaral, 27 Aug 2026
We thank all three reviewers for their careful reading of the manuscript and for their constructive comments. Below we provide a point-by-point response in bold to each comment. Changes to the manuscript are indicated in the revised text. Where comments from Reviewer 3 overlap with those of Reviewer 1, we refer to the corresponding Reviewer 1 response to avoid unnecessary repetition.
(RC1) Main comments
- Section 2.2, Dataset used
“Databases associated with lakes and wetlands were analyzed at 1 km resolution”: Could the authors clarify the temporal resolution of each dataset used?
Response: Temporal resolution of datasets used:
Thanks for the comment. There are important temporal differences between the datasets that we agree need clarification, as noted by the reviewer. We modified the text in the methods and discussion to reflect these temporal clarifications.
Text that was added to the methods section:
¨ We used a wide range of existing global and regional datasets that represent the wetland area extent, CH4 surface emissions, climate, land use and land management, as well as peat areas in the Llanos del Orinoco. (1) the Global Lakes and Wetland Database Level 3 (GLWD-3; Lehner & Döll, 2004), a static classification map representing wetland distribution based on early 2000s data sources with no temporal variability; (2) the Tropical and Subtropical Wetland Distribution Database v7.1 (TSWD; Gumbricht et al., 2017a), a climatological composite representing long-term average wetland distribution centered on 2011, without interannual resolution; (3) the Wetland Cover map from the Globcover Project (Arino et al., 2007; Bicheron et al., 2006), a single-epoch land cover classification based on MERIS imagery acquired during 2005–2006; (4) the Global Surface Water Extent (GSWE; Pekel et al., 2016), providing monthly surface water observations from 1984 to present derived from the full Landsat archive, for this analysis we used the maximum water extent layer for the period 1984-2021; (5) the Global Inundation Extent from Multi-Satellites (GIEMS; Fluet-Chouinard et al., 2015), a monthly inundation product covering 1993–2004, used here as a climatological reference for spatial inundation patterns; (6) the PEATMAP dataset (Xu et al., 2018) and the Tropical and Subtropical Peat Distribution v2.1 (TSPD; Gumbricht et al., 2017b), both static compilations of peat distribution with no temporal resolution. Only datasets that cover the entire Llanos del Orinoco region and bilinearly interpolated to 1 km resolution were included. Databases associated with lakes and wetlands were analyzed at 1 km resolution. We focused our analysis on 2018 as the reference year, as this is the most recent year for which the dynamic datasets overlap with available CH4 flux estimates. For static datasets, we treat their spatial extent as representative of contemporary wetland distribution, acknowledging that absolute wetland areas may differ from 2018 conditions due to long-term hydrological change or land-use conversion occurring after their respective baseline periods, particularly for GIEMS, whose coverage ends in 2007. It is also important to highlight that our main goal in this analysis is to contextualize the Llanos del Orinoco within a regional CH4 budget, disentangling the contributions of different sources, rather than exploring interannual variability in the computed fluxes.¨
Text added to Figure 4 captions:
“ Note that differences in panel G may reflect both methodological disagreement between products and real changes in surface water extent, as Globcover represents a single epoch (2005–2006) while GSWE captures dynamic inundation over multiple years.”
We added text in the results section with further clarification of wetcharts discrepancies associated with inundation:
“It is also worth noting that both inundation products differ not only in classification methodology but in their temporal baseline: GLWD-3 represents wetland distribution from the early 2000s and Globcover from 2005–2006, neither of which captures potential changes in inundated area up to 2018. As a result, the emissions magnitude difference between WetCHARTsX3 and WetCHARTsX4 likely reflects a combination of structural mapping differences and temporal misrepresentation of wetland extent, with the relative contribution of each source of uncertainty difficult to disentangle without a contemporaneous inundation product as reference. However, WetCHARTs seasonality is captured and reproduced by precipitation inputs (ERA5) used in the model, which is an important driver of inundation dynamics at the Llanos del Orinoco.”
(RC1). In addition, more recent datasets are now available and could improve the analysis or could be discussed:
GLWD v2: Lehner et al. (2024). Mapping the world’s inland surface waters: an update to the Global Lakes and Wetlands Database (GLWD v2). Earth System Science Data.
GIEMS-MethaneCentric: Bernard et al. (2024). The GIEMS-MethaneCentric database: a dynamic and comprehensive global product of methane-emitting aquatic areas. Earth System Science Data.
Response: We added text addressing the comment associated with more recent datasets:
Modified Text added (highlighted) to discussion section 4.2 regarding new available inundation mapping products:
“4.2 Research needs and challenges for understanding CH4 dynamics in the Llanos del Orinoco
4.2.1 Inundation mapping and seasonal dynamics.
We identified three key research areas that require further attention to improve CH4 emission estimates and dynamics. The first one is inundation mapping and seasonal dynamics. Inundation mapping is a central need for reducing uncertainties. A clear need for this is exemplified by the marked seasonal pattern of CH4 fluxes, driven by local precipitation, and consistently captured by both top-down and bottom-up estimates as well as field studies. Inconsistencies in mapped inundated areas currently represent the primary source of uncertainty in bottom-up CH4 emission estimates (Fig. 2). There are large inconsistencies in the inundation extent between products (Fig. 4), highlighting the need for improved inundation mapping for the Llanos del Orinoco region. Savannas and interfluvial wetlands have less predictable inundation dynamics, as it is largely governed by local rainfall and runoff (Fleischmann et al 2022). Furthermore, there are fundamental differences in sensor technology, coarse resolution among different products, the temporal representation, and the definition criteria for delimiting inundation areas among products that explain such contrast in inundation mapping area (see Fleischmann et al., 2022). GLWD-3 and TSWD provide inundation areas greater than 100,000 km2, but when the inundated mapped areas are compared between these two products, they only coincide in ~49,000 km2. Other available products (GlobCover and GSWE) for inundation mapping in the Llanos del Orinoco have a much lower inundation area, with ~20,000 km2. Comparisons between GlobCover and GSWE inundated mapped area coincide by only ~11,000 km2 (Fig. 4). As shown in Fig. 2, uncertainty in the extent of flooded areas directly translates into uncertainty in CH4 emissions, with inundation extent currently being the dominant factor limiting the accuracy of bottom-up CH₄ emission estimates. To assess whether this uncertainty persists in more recently available products, we extracted inundation extent estimates for our 2018 reference year using the GIEMS-MethaneCentric v2 database (Bernard et al., 2024), a dynamic product extending from 1992 to 2020, which reports a maximum annual inundated area of 91,507 km² for 2018 (compared to 111,524 km² for its full time-series maximum). This estimate, along with the 17,582 km² from GSWE (Pekel et al., 2016) already used in our analysis, falls within the same order of magnitude as the older products shown in Figure 4. This confirms that the uncertainty structure we document is not an artifact of using older datasets, but rather a fundamental challenge in satellite-based wetland mapping across this region.
Global models provide a broad overview of CH4 emissions; however, they often lack the detail necessary to accurately represent regional variations in CH4 emissions. None of the studies compiled from the literature have evaluated the impact of variations in inundation extent on CH4 fluxes. Mapping the frequency and duration of inundation for the different aquatic habitats and wetlands in the Llanos del Orinoco is crucial for properly upscaling CH4 emissions. These habitats represent the major CH4 flux in our budget, and seasonally inundated areas can act as a source or a sink depending on inundation and soil water saturation conditions (Smith et al., 2000; Rondon, 2000). We also note that CYGNSS (Cyclone Global Navigation Satellite System), a small satellite constellation with daily temporal resolution and the ability to penetrate cloud cover and dense vegetation, has recently demonstrated potential for inundation mapping in tropical regions (Jensen et al., 2018). Its sensitivity to surface soil moisture and inundation makes it particularly promising for the Llanos del Orinoco, where persistent cloud cover limits optical satellite observation, especially during the wet season (Fleischmann et al., 2022). While we have not applied CYGNSS data here, its integration with process-based models and atmospheric inversions represents a promising avenue for reducing wetland extent uncertainty in future regional CH₄ budget assessments. Taken together, these results reinforce the central message of this study: improving the accuracy and temporal resolution of wetland mapping in the Llanos del Orinoco is a prerequisite for better constrained CH₄ budgets. Achieving this will require not only the adoption of next-generation satellite products, but also local field measurements of inundation dynamics and mechanistic understanding of the flood pulse processes that drive the seasonal expansion and contraction of wetland areas in this system.”
( RC1). Bottom-up and top-down models used: 1) It appears that only one top-down product (CAMS) was used. How are uncertainties across different top-down inversion systems accounted for? 2) For bottom-up estimates, the analysis seems to focus primarily on wetland CH₄ models. How are uncertainties from other sectors (e.g., lakes, rivers, livestock, oil and gas, soil uptake) incorporated? 3) For the GCP-CH₄ dataset, which specific models, and sectors were used in the analysis?
Response: We agree that using a single top-down inversion system limits our ability to characterize cross-inversion uncertainty. However, the main objective of our analysis is to characterize the uncertainty range between bottom-up and top-down approaches and not only within inversions. It is important to note that the individual top-down inversions used in the global methane budget study are not open access, which further limits our ability to include additional top-down estimates. Now, more specifically about the open-access CAMS inversion (v22r2) used in this study, it employs TM5-MP as the atmospheric transport model with ERA5 meteorology, assimilating both NOAA surface observations and GOSAT satellite CH4 retrievals. Prior flux uncertainty is prescribed as 100% for wetlands, rice, and biomass burning, and 50% for anthropogenic sources, with a spatial error correlation length scale of 500 km. The small change from before posterior CAMS estimates in our analysis (5.3 ± 2.5 to 4.9 ± 2.3 Tg CH4 yr⁻¹) indicates that the region is not well constrained by the assimilated observations, which we now state explicitly in the revised text. We complement the CAMS inversion with the GCP-CH4 multi-model ensemble (Saunois et al., 2024), which provides an additional independent benchmark for the total budget and the wetland component specifically. Further details on the CAMS methane inversion framework can be found in Segers and Nanni (2023).
(R1C). For bottom-up estimates, the analysis seems to focus primarily on wetland CH₄ models. How are uncertainties from other sectors (e.g., lakes, rivers, livestock, oil and gas, soil uptake) incorporated?
Response: We thank the reviewer for this important clarification request. We've centered the discussions on wetlands, as this was the largest contributor to our estimates, and because of its global importance among natural sources. The non-wetland sectors in our budget are as follows: In the GCP-CH4 bottom-up estimate (Equation 2): livestock emissions average 0.78 ± 0.02 Tg CH4 yr⁻¹ (24% of total), oil and gas industry 0.75 ± 0.02 Tg CH4 yr⁻¹ (23%), rice 0.01 ± 0.01 Tg CH4 yr⁻¹ (0.4%), biomass burning 0.06 ± 0.08 Tg CH4 yr⁻¹ (1.9%), and a combined "other" category (freshwaters, termites, waste) of 0.50 ± 0.01 Tg CH4 yr⁻¹. Soil uptake is the only negative flux term, averaging -0.18 ± 0.01 Tg CH4 yr⁻¹. In the CAMS top-down estimates (Equation 1), the "other" category encompasses livestock, termites, waste, and soil sink, averaging 1.3 ± 0.1 Tg CH4 yr⁻¹ (posterior). In the CAMS framework, sector-specific breakdown for non-wetland anthropogenic sources is not separately optimized by the inversion; prior values from EDGAR v7.0 (anthropogenic) and GFAS (biomass burning) are used directly with prescribed uncertainties of 50% for anthropogenic sources and 100% for biomass burning and rice. Uncertainty in the non-wetland sectors therefore reflects prior inventory uncertainty rather than observational constraint, which we now state explicitly in the revised text. We have added these sector-specific values to the supplementary material and clarified the uncertainty basis for each.
(RC1). For the GCP-CH₄ dataset, which specific models and sectors were used in the analysis?
Response: The GCP-CH4 dataset used in this study (Saunois et al., 2024) provides multi-model ensemble means for the period 2000–2020. For our analysis, we extracted ensemble mean bottom-up estimates for 2018 for the Llanos del Orinoco region (Colombia and Venezuela combined), covering the following sectors: wetlands, livestock, oil and gas industry, rice cultivation, biomass burning, soil uptake, and a composite "other" category that includes freshwaters, termites, and waste. For the wetland sector, we used the ensemble mean across all process-based models included in the GCP-CH4 bottom-up ensemble (including LPJ-wsl, ORCHIDEE, CLM5, and others listed in Saunois et al., 2024, Supplementary Table S2). We did not use individual model outputs separately; the ensemble mean and standard deviation across models represent the central estimate and spread reported in Figure 2. We have added a clarifying sentence to the methods.
(RC1). Wetland uncertainties: In Fig. 3, how much of the uncertainty in seasonal CH₄ emissions can be attributed specifically to seasonal variability in wetland extent? In addition, does Fig. 4 represent annual maximum wetland extent, or another metric? Clarifying this would help interpretation.
Response: CAMS seasonality is similar among prior and posterior (Figure 3). In contrast, the wetland extent products used in this analysis for the WetCHARTs emissions are all static (GLWD-3, Globcover) or climatological (TSWD, GIEMS). None provide dynamic monthly inundation extent for 2018, but instead they impose seasonal variability using precipitation from ERA5 as a proxy for inundation extent. This is an important limitation we now state explicitly in the results section:
“The seasonal amplitude in WetCHARTs is driven by temperature and precipitation seasonality rather than inundation dynamics, which likely underestimates the true seasonal range given the strong wet-dry seasonality of the Llanos del Orinoco.”
(RC1). Field estimates of CH4 emissions: The differences between field-based estimates and model-based (top-down and bottom-up) estimates are substantial (~an order of magnitude). A clearer explanation of this discrepancy would be beneficial. In particular, since the compiled studies have different approaches and research focuses, it is unclear whether these estimates are directly comparable. Could the authors clarify: 1) Which emission sources each field study represents (e.g., soil-only, wetlands, open water, total flux including all sources?); 2) Whether fluxes include diffusive and/or ebullitive components? 3) What’s the spatial coverage of each study? 4) how did they upscale?
Response: We thank the reviewer for this important clarification request. The compiled field studies indeed differ substantially in scope, methodology, and spatial coverage, which limits direct comparison with top-down and bottom-up model estimates. However, as these are the only real field data available, we consider it essential to present the existing data, also to clarify how little information is available from actual field observations. We have added a structured clarification in the revised text and updated Table 1 where data are available. Text now reads:
“ Only a limited number of studies have examined CH₄ fluxes from seasonally or permanently inundated ecosystems in the Llanos del Orinoco floodplain. Our literature search identified only 7 relevant manuscripts that report field measurements for CH4 emissions (Table 1), although others (Etter et al., 2010; Rondón et al., 2006; San José & Montes, 2001) use the information to discuss the regional role of the Llanos de la Orinoquia in the carbon cycle. The field studies suggest that the Llanos del Orinoco floodplain is a small source of CH4 to the atmosphere compared to other wetlands in South America. Annual estimates of CH4 flux from the Llanos del Orinoco floodplain and upper delta (total area, 14.5 × 103 km2) in Venezuela are 0.17 Tg CH4 yr-1 or 113.6 mg CH4 m2 d-1 as reported by Smith et al. (2000), which is restricted to aquatic or inundated habitats. Castaldi et al. (2004) cover soil fluxes from terrestrial savanna habitats (including woodland, herbaceous savannas, open tree savannas, and cultivated pastures), and report that the Llanos del Orinoco in Colombia and Venezuela, for a total area of 357 × 103 km2, act as a source of CH4, contributing 0.13 Tg CH4 yr-1. A study at the Colombian portion of the Llanos del Orinoco calculated, for Colombia and Venezuela, annual CH4 emissions of 0.16 Tg CH4 yr-1, for an area of 120 × 103 km2 including estimates for the savanna ecosystem (soils under crops, pasture, gallery forests, and natural savannas, termites) together with burning and cattle (Rondon 2000).”
Upscaled regional estimates range from 0.13 to 0.17 Tg CH4 yr⁻¹. Approximately one order of magnitude lower than our model-based estimates (3.27–5.31 Tg CH4 yr⁻¹). This discrepancy is expected because: (i) all field studies predate 2005 and cover partial sectors of the total CH4 budget; (ii) none captures the full spatial extent of the Llanos del Orinoco used in our analysis; and (iii) the top-down and bottom-up models integrate wetland, anthropogenic, and all other sources over the entire region. The order-of-magnitude difference therefore reflects both differences in the area used for the spatial interpolation, and gaps in spatial coverage at sector representation, rather than a fundamental inconsistency in per-unit-area flux rates, which are broadly comparable across studies (Table 2).”(RC1). Discussion: The datasets or models used may miss small water bodies, such as small lakes/ponds/wetlands which could have high CH4 emissions (see papers below), and may explain the large discrepancies between field-based estimates and bottom-up or top-down estimates. Briefly discussing this limitation would strengthen the manuscript:
Holgerson et al. (2016). Large contribution to inland water CO2 and CH4 emissions from very small ponds. Nature Geoscience.
Li et al. (2026). The underappreciated importance of small wetlands in global methane emissions. Nature Climate Change.
Response: In section 4.1, we added text acknowledging the importance of small water bodies for CH4 budgets and their likely relevance for the Llanos del Orinoco. Paragraph now reads:
“ Our total CH4 budget for the Orinoco region ranges from 3.27 (+- 0.71, GCP-CH4) to 5.31 (+- 2.50, CAMS-Inversion prior), indicating that the region is a significant but not dominant CH4 source, compared to other important wetlands in South America (Table 2). On an annual basis, our estimate is at a similar magnitude as other seasonal savannas in the continent (Pantanal, Llanos de Moxos), but an order of magnitude smaller than the Amazon Basin (Table 2). Compared to other relevant known CH4 sources, our estimate, represents one- seventh of CH4 fluxes from boreal arctic wetlands and lakes (Table 2), one-fifth of global CH4 fluxes reported from reservoirs (25, range 13–65 Tg CH4 yr−1 ), one-sixth from rivers and streams (29 ±17 (±CI95 %) Tg CH4 yr−1 ), one-tenth of global emissions (53, range 19–86 Tg CH4 yr−1) from large lakes (>0.1 ha), small lakes and ponds (>0.1 ha), reported by the recent Global CH4 Budget (bottom-up approaches, 2010-2019) (Saunois et al., 2024). It is worth noting that global datasets and process-based models used in this study systematically underrepresent small water bodies such as ponds, small lakes, and wetlands, which can have disproportionately high CH4 emissions per unit area relative to larger water bodies (Holgerson et al., 2016; Li et al., 2026). In the Llanos del Orinoco, the mosaic of esteros, small oxbow lakes, and small terrain depressions, embedded in seasonally inundated savannas, is an example of small water bodies that are likely ignored by coarse-resolution products, which can partly explain the discrepancies between field-based estimates and our model-based budget. However, areal CH4 flux rates are similar, or within the range, reported for mean areal fluxes for important natural sources like the Amazon, other savanna ecosystems, or for global medians for northern wetlands (Table 2).”
(RC1) Minor comments
- Abstract Line 23: Add the unit Tg CH4 yr-1?
- Line 127: .”We used several approaches … CH4 concentrations” Seems this study did not analyze CH4 concentrations?
Response: Abstract line 23: We have added the unit as requested
Line 127: We have revised the sentence and deleted the mention of water concentrations. Now reads: " We used several approaches and available products to estimate CH4 emissions for the Llanos del Orinoco."
Citation: https://doi.org/10.5194/egusphere-2026-2080-AC1 - Section 2.2, Dataset used
-
AC1: 'Reply on RC1', Joao Henrique Fernandes Amaral, 27 Aug 2026
-
RC2: 'Comment on egusphere-2026-2080', Anonymous Referee #2, 15 Jun 2026
This is a clear and valuable study highlighting methane emissions from the understudied Llanos del Orinoco region. The topic is important because tropical wetlands are increasingly recognized as a major uncertainty in the global methane budget, and methane emissions in this region may be larger than previously expected. The authors do a great job collecting information from top-down and bottom-up sources and producing quantitative estimates for the region.
My main general comment is that the paper sometimes focuses more on pointing out what is missing than on what is known, even though the authors have done an excellent job collecting and synthesizing the available information. I think the manuscript would be stronger if some sections were smoothed out to first tell a clear story about what is currently known, and then more concisely identify what remains unknown. The message of the paper is important, so I would not want the writing or repetition to detract from it.
In general, the writing could be made smoother and more concise throughout the paper. There is some repeated information, especially around the region being understudied. Again, I feel that this is very important work, and could have greater impact if the writing was made more clear and concise. I provide a few specific suggestions for improving clarity below, starting with the Introduction.
Line 38: “Carbon Dioxide” should not be capitalized.
Line 40: “especially considering potential overshoot scenarios” is a bit vague. I would clarify what you mean here, or make the sentence more specific about why methane is especially important in this context.
Line 43: “approx.” should be changed to “approximately.”
Line 58: “at required spatial and seasonal scales” is unclear. Required by whom, or required for what purpose? Potential to clarify here.
Line 67: “understudied compared to other South American wetlands” repeats the idea already stated at the beginning of the paragraph. Consider removing or shortening this sentence to avoid repetition.
Section 3.1
Lines 161–162: “The small change from prior to posterior in the CAMS-Inversion suggests that the region is not well constrained by the assimilated data.” I think this sentence needs to be clarified. A small prior-to-posterior change could mean the region is not well constrained, but couldn’t it also mean that the CAMS prior is relatively accurate? Could this be diagnosed more quantitatively? For example, what is the observation density for the inversion over this region, or is there any posterior error reduction / sensitivity diagnostic available?
Line 164 and elsewhere: “Wetcharts” should be “WetCHARTs” for consistency with the product name.
Figure 2: Why is the CAMS wetland prior so much higher than the Global Methane Budget wetland fluxes? It could be useful to explain this more clearly in the text.
Section 4
Lines 280–281: “and a lack of coincidence between them” can be omitted because the sentence already says there are large inconsistencies between products.
Lines 283–285: The sentence beginning “Furthermore, there are fundamental differences…” should be clarified. I think the point is important, but the sentence is hard to follow.
Figure 4: This is really interesting. Have you looked at CYGNSS inundation extent as well? CYGNSS could be useful here because it provides satellite-derived, relatively high-resolution information and performs well over tropical regions, including Colombia. It would be great to add if it is not too difficult.
Line 316: “To date, no CH4 emissions data have been recorded from peatlands of the Llanos del Orinoco…” You could clarify this statement since you are using satellite/inversion information over the region to specify that no in situ or peatland-specific flux measurements have been reported.
Line 318: The paragraph beginning “Peatlands that are dependent on rainwater…” is really interesting. This type of mechanistic explanation is very helpful for methane scientists to understand why different wetland types may have different methane emissions. If you have sources to cite here that explain the distinction between rainwater-fed, nutrient-poor peatlands and more minerotrophic/productive peatlands, that would strengthen the paragraph.
Lines 353–354: “Cattle numbers in Colombia have increased by about 25% since 2016…” You could provide a source for this and clarify if this increase is for all of Colombia, or specifically for the Llanos del Orinoco region.
Lines 370–371: “Nathan et al. (2024) estimate national emissions for Venezuela of 7.5 Tg CH4 yr-1 in 2019, which is higher than our regional estimate for the entire Llanos del Orinoco.” I do not fully understand the point of this comparison, because presumably total Venezuelan emissions would be higher than emissions from only the Llanos del Orinoco. I would either clarify what this comparison is meant to show or replace it with a more direct comparison.
More generally, are there other broader TROPOMI regional studies that could be used to provide more top-down data for comparison? Other studies show some seasonal coverage over parts of the Orinoco region. Even if the satellite coverage is limited by clouds or surface conditions, it could be useful to include additional regional top-down context.
Overall, I appreciate the attention this paper brings to the Llanos del Orinoco, which is clearly an important and understudied region for the methane budget. Quantifying the wetland contribution here is especially important given the large differences among existing estimates and the uncertainty in inundation extent. I think the manuscript makes a valuable contribution by bringing together the available data, and I recommend publication after revisions that improve the flow, reduce repetition, and clarify a few points.
Citation: https://doi.org/10.5194/egusphere-2026-2080-RC2 -
AC2: 'Reply on RC2', Joao Henrique Fernandes Amaral, 27 Aug 2026
We thank all three reviewers for their careful reading of the manuscript and for their constructive comments. Below, we provide a point-by-point response to each comment in bold. Changes to the manuscript are indicated in the revised text.
(RC2). This is a clear and valuable study highlighting methane emissions from the understudied Llanos del Orinoco region. The topic is important because tropical wetlands are increasingly recognized as a major uncertainty in the global methane budget, and methane emissions in this region may be larger than previously expected. The authors do a great job collecting information from top-down and bottom-up sources and producing quantitative estimates for the region.
My main general comment is that the paper sometimes focuses more on pointing out what is missing than on what is known, even though the authors have done an excellent job collecting and synthesizing the available information. I think the manuscript would be stronger if some sections were smoothed out to first tell a clear story about what is currently known, and then more concisely identify what remains unknown. The message of the paper is important, so I would not want the writing or repetition to detract from it.
In general, the writing could be made smoother and more concise throughout the paper. There is some repeated information, especially around the region being understudied. Again, I feel that this is very important work, and could have greater impact if the writing was made more clear and concise. I provide a few specific suggestions for improving clarity below, starting with the Introduction.
We thank Reviewer 2 for their positive assessment of the manuscript and for their constructive suggestions to improve clarity, concision, and flow. We have carefully revised the manuscript to reduce repetition, improve the narrative structure, and address each specific point raised. We respond to each comment below.
(RC2). Line 38: “Carbon Dioxide” should not be capitalized.
Response: Corrected following reviewer suggestion.
(RC2). Line 40: “especially considering potential overshoot scenarios” is a bit vague. I would clarify what you mean here, or make the sentence more specific about why methane is especially important in this context.
Response: Corrected following reviewer suggestion. See L 39-42
“ As such, atmospheric CH4 plays an important role in short-term global warming. Its brief atmospheric lifetime becomes especially critical in overshoot scenarios, offering the most effective lever for rapidly reducing peak temperatures after exceeding the 1.5°C threshold.”
(RC2). Line 43: “approx.” should be changed to “approximately.”
Response: Corrected following reviewer suggestion.
(RC2). Line 58: “at required spatial and seasonal scales” is unclear. Required by whom, or required for what purpose? Potential to clarify here.
Response: Text was modified following the reviewer's suggestion. See L59:
“In addition, representative site-level CH₄ flux measurements at the spatial and temporal resolutions required to parameterize and validate bottom-up models are scarce (Melack et al., 2022).”
(RC2). Line 67: “understudied compared to other South American wetlands” repeats the idea already stated at the beginning of the paragraph. Consider removing or shortening this sentence to avoid repetition.
Response: Text was removed following the reviewer´s suggestion.
(RC2). Section 3.1
Lines 161–162: “The small change from prior to posterior in the CAMS-Inversion suggests that the region is not well constrained by the assimilated data.” I think this sentence needs to be clarified. A small prior-to-posterior change could mean the region is not well constrained, but couldn’t it also mean that the CAMS prior is relatively accurate? Could this be diagnosed more quantitatively? For example, what is the observation density for the inversion over this region, or is there any posterior error reduction / sensitivity diagnostic available?
Response: We thank the reviewer for this insightful observation. The comment touches upon a fundamental aspect of inverse modeling: the distinction between an accurate prior estimate and a region that is simply "sub-constrained" due to a lack of observational data. We agree that the small change between the prior (5.31 ± 2.50 Tg CH₄ yr⁻¹) and the posterior (4.97 ± 2.33 Tg CH₄ yr⁻¹) in the CAMS-Inversion for the Llanos del Orinoco suggests that the assimilated atmospheric signal is not contributing to significantly shift the model's initial values. While a small change could theoretically imply a highly accurate prior, the lack of local ground-based monitoring stations in the immediate region makes the "sub-constrained" hypothesis much more probable.
Diagnosing this quantitatively is a complex task that requires tools beyond the scope of this regional synthesis paper. To address this concern rigorously, one must perform detailed sensitivity experiments and posterior error reduction diagnostics. We are currently detailing these analyses in a separate manuscript (currently in preparation). Adding these extensive atmospheric transport diagnostics and sensitivity experiments to the current manuscript would fundamentally change its focus from a regional synthesis to an atmospheric modeling methodology paper.
(RC2).Line 164 and elsewhere: “Wetcharts” should be “WetCHARTs” for consistency with the product name.
Response: Revised and corrected following the reviewer's suggestion.
(RC2).Figure 2: Why is the CAMS wetland prior so much higher than the Global Methane Budget wetland fluxes? It could be useful to explain this more clearly in the text.
Response: We thank the reviewer for this comment. The CAMS-Inversion (v2022r2) specifically utilizes the LPJ-wsl model for its wetland prior. In contrast, the Global Methane Budget (GCP-CH4) is a synthesis derived from a broader ensemble of bottom-up process models. If the LPJ-wsl model assumes a larger or more persistent inundation area for the Orinoco region than the average of the GCP ensemble, the CAMS prior will be inherently higher. The higher CAMS wetland prior likely reflects a higher initial estimate of flooded area in the underlying LPJ-wsl model, which persists in the final results due to the lack of local ground-based measurements to provide a more accurate counter-constraint.
(RC2). Section 4
Lines 280–281: “and a lack of coincidence between them” can be omitted because the sentence already says there are large inconsistencies between products.
Response: Revised and corrected following the reviewer's suggestion
(RC2). Lines 283–285: The sentence beginning “Furthermore, there are fundamental differences…” should be clarified. I think the point is important, but the sentence is hard to follow.
(RC2).Figure 4: This is really interesting. Have you looked at CYGNSS inundation extent as well? CYGNSS could be useful here because it provides satellite-derived, relatively high-resolution information and performs well over tropical regions, including Colombia. It would be great to add if it is not too difficult.
Response: We chose not to include CYGNSS in Figure 4, as our four-product comparison was intended to serve as an illustrative example of the inconsistencies observed among widely used datasets. Nevertheless, in response to the reviewer’s suggestion, we have incorporated text into the Discussion section that explicitly recommends CYGNSS for future inundation mapping in the region, highlighting its daily temporal resolution and cloud-penetrating capabilities relative to optical products. This addition appears in the revised subsection 4.2.1 (Inundation mapping and seasonal dynamics), which also addresses the previous comment and is further detailed in our response to Reviewer 1.
(RC2). Line 316: “To date, no CH4 emissions data have been recorded from peatlands of the Llanos del Orinoco…” You could clarify this statement since you are using satellite/inversion information over the region to specify that no in situ or peatland-specific flux measurements have been reported.
Response: Text was modified to clarify the reviewer's comment.
“ We specifically highlight the total absence of in situ or habitat-specific field measurements, such as flux chambers or eddy covariance towers, for these newly identified peat-forming systems, though data from Amazonian peatlands suggest that they could be potent emitters (Winton et al. 2017, Teh et al. 2017, Soosar et al. 2022). While our regional top-down inversions provide a larger scale and coarser integrated estimate of methane emissions for the Llanos del Orinoco, it is important to clarify that these atmospheric signals represent a basin-wide average and cannot isolate the contribution of specific ecosystem types, such as peatlands.”
(RC2). Line 318: The paragraph beginning “Peatlands that are dependent on rainwater…” is really interesting. This type of mechanistic explanation is very helpful for methane scientists to understand why different wetland types may have different methane emissions. If you have sources to cite here that explain the distinction between rainwater-fed, nutrient-poor peatlands and more minerotrophic/productive peatlands, that would strengthen the paragraph.
Response: We modified the paragraph accordingly to the reviewer´s suggestions, now reads:
“ Peatlands that are dependent on rainwater (bogs) are typically poor producers and emitters of CH4 because the provision of fresh carbon inputs by plants is limited by their nutrient-poor setting (Abdalla et al. 2016; Finn et al. 2020). Peatlands of the Llanos del Orinoco, however, are likely to be of higher productivity and minerotrophic since they are often subject to pulses of nutrient-rich surface water during wet season floods and sustained by groundwater inputs during dry seasons. Although they may represent a small fraction of total wetland area, they may contribute disproportionately to regional CH4 budgets because of their potential for high perennial primary productivity and persistent soil saturation (Winton et al. 2025). However, field-based measurements are needed to ascertain their contribution to CH4 relative to other wetland types.”
(RC2). Lines 353–354: “Cattle numbers in Colombia have increased by about 25% since 2016…” You could provide a source for this and clarify if this increase is for all of Colombia, or specifically for the Llanos del Orinoco region.
Response: Text was modified to clarify the reviewer's comment. See L356: “Cattle numbers in all of Colombia have increased by about 25% since 2016 (Federación Colombiana de Ganaderos [Fedegán], 2026).”
Reference: Federación Colombiana de Ganaderos (Fedegán): Datos e Indicadores – Inventario Bovino, available at: https://www.fedegan.org.co/datos-e-indicadores, last accessed: 26 August 2026.
(RC2). Lines 370–371: “Nathan et al. (2024) estimate national emissions for Venezuela of 7.5 Tg CH4 yr-1 in 2019, which is higher than our regional estimate for the entire Llanos del Orinoco.” I do not fully understand the point of this comparison, because presumably total Venezuelan emissions would be higher than emissions from only the Llanos del Orinoco. I would either clarify what this comparison is meant to show or replace it with a more direct comparison.
Response: We opted to exclude the comparison, as we agree with the reviewer that the comparison was out of place.
(RC2). More generally, are there other broader TROPOMI regional studies that could be used to provide more top-down data for comparison? Other studies show some seasonal coverage over parts of the Orinoco region. Even if the satellite coverage is limited by clouds or surface conditions, it could be useful to include additional regional top-down context.
Response: We modified the paragraph to acknowledge the reviewer's comment, and now it reads:
“The anthropogenic share of the total fluxes compiled in our study is dominated by the oil and gas industry, which together contribute 0.75 Tg CH4 yr-1 (Fig. 2). The oil and gas industry in the Llanos del Orinoco is concentrated close to the Andean foothills, as well as in the southern-eastern portion of the Llanos del Orinoco (Fig. 1). A recent work in Venezuela (Nathan et al., 2024) reveals that CH4 emissions from Venezuela's oil production, particularly around Lake Maracaibo, remain alarmingly high despite a sharp decline in oil exploitation, suggesting leaks from abandoned or deteriorating infrastructure rather than active production. The region around Lake Maracaibo contributes 1.2 Tg CH4 yr-1, with approximately half of it being attributed to oil production, which is about the same magnitude of the oil and gas evasions reported in our study (Fig. 2). These findings highlight the urgent need for better monitoring of CH4 fluxes and leaks in oil-producing regions to mitigate climate impacts. Broader top-down context provided by recent TROPOMI and GOSAT inversions (Hancock et al., 2025) further underscores the regional significance of the Orinoco; such continental-scale assessments identify Colombia and Venezuela as two of the top five anthropogenic methane emitters in South America, primarily due to high methane intensities in their fossil fuel and livestock sectors. Nathan et al. (2024) also highlight limitations of using TROPOMI to obtain CH₄ flux estimates due to the region's complex topography, persistent cloud cover, and low albedo, making satellite observations difficult. Hancock et al. (2025) similarly note that TROPOMI observations over the Andes and northern South America are sparse, and they compensate by using GOSAT glint data offshore to improve constraints. TROPOMI's high resolution and daily coverage provide unprecedented data. Still, uncertainties remain due to limited observational constraints for northern South America, which can be improved with more ground measurements or atmospheric monitoring stations.”
(RC2). Overall, I appreciate the attention this paper brings to the Llanos del Orinoco, which is clearly an important and understudied region for the methane budget. Quantifying the wetland contribution here is especially important given the large differences among existing estimates and the uncertainty in inundation extent. I think the manuscript makes a valuable contribution by bringing together the available data, and I recommend publication after revisions that improve the flow, reduce repetition, and clarify a few points.
Response: We thank Reviewer 2 for their thorough reading of the manuscript and for raising several important points. We believe that the manuscript is improved in this new version, where we carefully reduced repetition and have clarified the points raised by the reviewer.
Citation: https://doi.org/10.5194/egusphere-2026-2080-AC2
-
AC2: 'Reply on RC2', Joao Henrique Fernandes Amaral, 27 Aug 2026
-
RC3: 'Comment on egusphere-2026-2080', Anonymous Referee #3, 22 Jun 2026
The work seems interesting based upon Abstract and Introduction, but it is presented as “this is what we are doing, and why we are doing in this way has not been provided”. As an academic work, providing the rationale of "why we are doing/selecting/developing this, rather than that..." is more important than "we are doing...". This is shown from data selection to methodology design. No detailed information of each selected data source, the function of each data source, spatial and temporal resolution and coverage, and rationale of data selection; no rationale of methodology/model selection, and temporal resolution of the model outputs, even no description of the listed two equations on page 5, and nothing in methodology is provided how to separate the wetland fluxes from others, and no validation of the model outputs. what is the benchmark of the estimations? Because of this, we cannot tell whether the estimations are correct or not. Figure 1 has not been called out in text and corresponding text has not been included. No real field data used to validate the estimations, even no reference dataset to validate the extent of wetlands in such a big area. Uncertainty might be huge because of no such dataset with detail level to quantify the heterogeneity of inundation frequency, wetland extent, and others.
Citation: https://doi.org/10.5194/egusphere-2026-2080-RC3 -
AC3: 'Reply on RC3', Joao Henrique Fernandes Amaral, 27 Aug 2026
We thank all three reviewers for their careful reading of the manuscript and for their constructive comments. Below we provide a point-by-point response in bold to each comment. Changes to the manuscript are indicated in the revised text. Where comments from Reviewer 3 overlap with those of Reviewer 1, we refer to the corresponding Reviewer 1 response to avoid unnecessary repetition.
(RC3). The work seems interesting based upon Abstract and Introduction, but it is presented as “this is what we are doing, and why we are doing in this way has not been provided”. As an academic work, providing the rationale of "why we are doing/selecting/developing this, rather than that..." is more important than "we are doing...". This is shown from data selection to methodology design. No detailed information of each selected data source, the function of each data source, spatial and temporal resolution and coverage, and rationale of data selection; no rationale of methodology/model selection, and temporal resolution of the model outputs, even no description of the listed two equations on page 5, and nothing in methodology is provided how to separate the wetland fluxes from others, and no validation of the model outputs. what is the benchmark of the estimations? Because of this, we cannot tell whether the estimations are correct or not. Figure 1 has not been called out in text and corresponding text has not been included. No real field data used to validate the estimations, even no reference dataset to validate the extent of wetlands in such a big area. Uncertainty might be huge because of no such dataset with detail level to quantify the heterogeneity of inundation frequency, wetland extent, and others.
Response: We thank Reviewer 3 for their reading of the manuscript and for raising several points. We agree that the original manuscript did not sufficiently justify the selection of each dataset or describe their spatial and temporal characteristics. We have substantially revised the methods section to include, for each dataset: (i) its spatial resolution, (ii) its temporal resolution and coverage, and (iii) the rationale for its inclusion. Specifically, we now distinguish between static climatological products (GLWD-3, TSWD, Globcover) and dynamic products (GSWE, GIEMS), and explicitly acknowledge the temporal mismatch between products and our 2018 reference year. We also discuss how this mismatch contributes to uncertainty in the wetland extent estimates that drive the WetCHARTs ensemble. An expanded description of the equations presented on page 5 was included. We note at the outset that several methodological concerns raised by Reviewer 3 overlap substantially with those of Reviewer 1, and we have addressed these in detail in our response to Reviewer 1.
The reviewer asks what benchmark is used to evaluate whether our estimates are correct. This is a legitimate claim but at the same time and given our findings is rather vague. What we show in our study is that most data products, from wetland extent to wetland methane emissions, disagree. Therefore, what could be a good benchmark in this context? Does the reviewer have any constructive suggestions for such a benchmark? Nevertheless, we acknowledge this and have added a dedicated paragraph in the discussion addressing it. That said, the reviewer should note that the Llanos del Orinoco is a severely data-sparse region for CH4 flux measurements. The absence of flux towers, site level chamber measurements and even large scale atmospheric in-situ monitoring stations, that at least could approximate a spatially comprehensive field evaluation dataset, is not an oversight of this study, it is precisely the knowledge gap that motivates it. Our paper's contribution is to synthesize available global and regional products to provide a first-order CH4 budget for this region, explicitly characterizing the uncertainty that arises from the lack of local observational constraints. We compare bottom-up (WetCHARTs) and top-down (CAMS inversion) estimates as a form of mutual benchmarking, and contextualize our results against the Global CH4 Budget (GCP-CH4, Saunois et al., 2024). We have revised the discussion to make this framing more explicit, and we argue that identifying the need for field campaigns in the Llanos is itself a key outcome of this work.
Lastly, we respectfully disagree with the comment that Figure 1 wasn’t cited in the manuscript text. We invite the reviewer to re-read Section 2.1 (Study Area), where it is referenced in the description of the geographical boundaries and land use context of the Llanos del Orinoco.
Finally, the reviewer correctly identifies that uncertainty may be large given the heterogeneity of inundation frequency and wetland extent in the region. We have expanded our uncertainty discussion, drawing on the ensemble spread between WetCHARTsX3 and WetCHARTsX4 as a measure of structural uncertainty in wetland extent, and on the divergence between bottom-up and top-down estimates as a measure of overall budget uncertainty. We explicitly acknowledge that without high-resolution inundation data and local flux measurements, the quantification of uncertainty remains incomplete, and frame this as a primary motivation for future observational investment in the region.
Citation: https://doi.org/10.5194/egusphere-2026-2080-AC3
-
AC3: 'Reply on RC3', Joao Henrique Fernandes Amaral, 27 Aug 2026
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 536 | 122 | 38 | 696 | 48 | 45 |
- HTML: 536
- PDF: 122
- XML: 38
- Total: 696
- BibTeX: 48
- EndNote: 45
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Methane emissions from the tropics play a significant role in the global methane cycle. While the Llanos del Orinoco represents a vast tropical savanna floodplain in northern South America, the sources, sinks, and processes driving regional methane dynamics remain understudied.
In this manuscript, the authors integrate top-down atmospheric inversion estimates (CAMS) and bottom-up process-based modeling (WetCHARTs, GCP-CH₄), together with a systematic compilation of sparse field measurements, to quantify regional CH₄ emissions and identify key uncertainties.
Overall, the manuscript is valuable and has the potential to advance our understanding of methane cycling and its uncertainty sources in the tropical Llanos del Orinoco region. I would recommend this manuscript for publication after addressing the following comments, which mainly relate to methodological clarification, uncertainties, and interpretation of results.
Main comments
“Databases associated with lakes and wetlands were analyzed at 1 km resolution”: Could the authors clarify the temporal resolution of each dataset used? In addition, more recent datasets are now available and could improve the analysis or could be discussed:
GLWD v2: Lehner et al. (2024). Mapping the world’s inland surface waters: an update to the Global Lakes and Wetlands Database (GLWD v2). Earth System Science Data.
GIEMS-MethaneCentric: Bernard et al. (2024). The GIEMS-MethaneCentric database: a dynamic and comprehensive global product of methane-emitting aquatic areas. Earth System Science Data.
2. Bottom-up and top-down models used: 1) It appears that only one top-down product (CAMS) was used. How are uncertainties across different top-down inversion systems accounted for? 2) For bottom-up estimates, the analysis seems to focus primarily on wetland CH₄ models. How are uncertainties from other sectors (e.g., lakes, rivers, livestock, oil and gas, soil uptake) incorporated? 3) For the GCP-CH₄ dataset, which specific models, and sectors were used in the analysis?
Minor comments