Satellite quantification of enhanced methane oxidation applied to the stratospheric plume following Hunga Tonga-Hunga Ha’apai eruption Authors: Maarten MJW van Herpen, Isabelle De Smedt, Daphne Meidan, Alfonso Saiz-Lopez, Matthew S Johnson, Thomas Röckmann, Jos de Laat Published: 2026 ABSTRACT AbstractMethane is a powerful greenhouse gas whose atmospheric sink remains uncertain, and emerging strategies to enhance its removal will require quantification and monitoring to verify any hypothetical future methane removal. Here we present satellite quantification of enhanced atmospheric methane oxidation, based on TROPOMI observations of a short-lived intermediate in methane oxidation, HCHO. We find a large HCHO enhancement of up to 12 ppb±10% at 30 km altitude, in the plume from the Hunga Tonga-Hunga Ha’apai eruption, persisting for ten days or more, and also explaining its low BrO levels. Total methane oxidation is 900 ± 220 Mg/day, suggesting at least 330 Gg of volcanic methane was injected into the stratosphere. The observed methane oxidation requires an estimated ongoing primary production of 2-5 Gg Cl per day that appears unexplained by known mechanisms. We show that chlorine production by iron-chloride photochemistry in sulfate-coated volcanic ash is a plausible mechanism, even outside the marine boundary layer. This method of measuring methane loss using formaldehyde can be sufficiently sensitive to quantify the impact of hypothetical future enhanced atmospheric methane oxidation approaches.Subject terms: Climate-change mitigation, Volcanology, Atmospheric chemistry Introduction Methane (CH4) is a powerful greenhouse gas, the most important after CO2, and currently responsible for 0.5 °C of warming1. The Paris agreement to stay below 2 °C of warming can now only be achieved by rapid and large reductions in methane emissions2. The worldwide increase in methane emissions originates from fossil fuel sources, waste treatment, rice production, and especially from agriculture (mainly dairy cows)3. At the same time, almost half of methane emissions come from natural sources such as wetlands3. The atmospheric sink of methane is also not well-quantified and leads to uncertainty regarding these methane sources calculated via mass balance arguments3. Methane is unique in that it is naturally broken down in the atmosphere within about 10 years, converting CH4 into CO2 and H2O. Because methane is 80 times as potent as CO2 over a 20-year timescale, its relatively fast atmospheric breakdown prevents methane from having an even greater climate impact. This also means that the current warming due to methane is caused only by recent emissions. Addressing methane emissions can therefore result in a reduction of global warming within a decade. Unfortunately, not all methane emissions can be mitigated; it is estimated that the maximum methane emission reduction if all technological options are used is 50%4, and natural methane emissions are rising due to global warming5,6. Despite global efforts to reduce methane emissions, such as through the Global Methane Pledge, atmospheric methane concentrations are rising at their fastest rate in over forty years, with record annual increases observed in 2020 and 2021 of 15.2 ± 0.5 ppb and 17.8 ± 0.5 ppb, respectively7. By 2023, the five-year growth rate of atmospheric methane reached its highest level on record7,8. In addition to methane emissions reduction, a new field of atmospheric methane removal is emerging that may reduce climate risks by artificially accelerating the natural breakdown of methane in the atmosphere9. Climate modeling has shown that large-scale atmospheric methane removal has the potential to reduce future temperatures by 0.5 °C10,11. If proven to be climate beneficial and cost-effective, open-air approaches likely have the largest potential scale and fastest time to scale compared to reactor-based approaches9. For example, researchers have studied the emission of chlorine to the open air, produced from conventional sea-water electrolysis technology12. The most studied approach involves iron-based particles that are lofted into the atmosphere to catalytically generate chlorine radicals that oxidize methane13. However, open-air approaches are inherently difficult to verify, and there are concerns about a higher risk of unintended consequences, necessitating strong governance approaches9. Such governance requires quantification and observations to verify any hypothetical future methane removal. Recently, a report by the National Academy of Sciences, Engineering, and Medicine14 considered the need and viable options for atmospheric methane removal. The report identified monitoring, reporting, and verification as a key challenge, and concluded that we currently lack tools for methane removal quantification. Here, we present a satellite-based quantification methodology that responds to this need. Satellite observations have proven their value for monitoring emission of air pollution and greenhouse gases15,16, e.g., supporting policy makers in the international drive to reduce global methane emissions. The capacity of satellites to map global concentrations cannot be matched by other means, e.g., the TROPOspheric Monitoring Instrument (TROPOMI) – currently one of the most advanced satellite instruments for atmospheric composition monitoring – is being used to monitor CH4, NO2, CO, HCHO, and SO2 from continental scales down to the scale of oil and gas infrastructure15,16. While it is possible to use local observations to quantify atmospheric methane removal17, satellite-based observations have been shown to be better suited to determining methane emissions16. However, due to the low albedo of water, satellites using reflected shortwave infrared cannot monitor methane over oceans where a substantial portion of natural methane oxidation occurs, and where proposed open-system approaches are envisioned to operate. Here we demonstrate that locally enhanced atmospheric methane oxidation can be detected using TROPOMI observations of the gas-phase species that are involved in methane oxidation, in particular formaldehyde (HCHO). These satellite measurements use UV wavelengths and can also be used over the oceans (unlike methane). Each oxidized methane molecule leads to the production of approximately one HCHO molecule, which is a short-lived intermediate yielding CO within a few hours. In the absence of local HCHO sources such as biomass burning, methane oxidation is the main source of HCHO, and the lifetime is limited to a few hours due to photolysis reactions and the reaction with OH and Cl radicals (see Methods). As a proof-of-concept demonstration, we will apply this method to the Hunga Tonga-Hunga Ha’apai (HTHH) volcanic eruption, in which chlorine activation was observed. Hunga Tonga-Hunga Ha’apai (HTHH), a submarine volcano in the South Pacific (20.54° S, 175.38° W) that was situated 150 meters below sea level, erupted violently on 15 January 2022. The area of the HTHH eruption is well-covered by satellites, including TROPOMI, EMI, CALIPSO, MSL, GOES-17, and Himawari-8. As a result, many satellite-based studies of the eruption have been conducted, using observations of species such as SO2, O3, BrO, CO, ClO, HCl, H2O, and aerosol optical depth18–25. What made the HTHH eruption exceptional is that it lofted material above 30 km to record-breaking heights of ∼55 km18,19. The blast released hundreds of Gg SO220–22, and injected an exceptionally large mass of H2O into the stratosphere, estimated to be 146 ± 5 Tg, or ∼10% of the total stratospheric burden21. The amount of SO2 was modest; for comparison, the Mt. Pinatubo (Philippines) eruption in 1991 injected ∼20 Tg of SO2 and reached 40 km at its highest point23. Recent work by Wu et al. showed that the low emissions of sulfur by HTHH are due to seawater–magma interactions that removed >93% of a total release of 18.8 Tg SO224. Chlorine activation within the stratospheric plume is evident from in-plume ClO enhancement and ozone depletion observed for about 10 days by the Microwave Limb Sounder (MLS) on board the Earth Observing System (EOS) Aura satellite25. Using a coupled chemistry-climate model (WACCM6), Zhu et al. found that a volcanic injection of 1.3 Gg of active Cl explained the observed O3 loss and ClO enhancement25. Results Satellite data analysis Note that we will report column density in units of molec/cm2 (with /), concentrations in cm−3, and rates in cm−3 s−1, to clearly distinguish between them. The main HTHH eruption occurred on 15 Jan 2022 at 17:00 local time, and is first visible in the Jan 16 TROPOMI overpass at 13:30 (20 hours later). Figure 1 shows the TROPOMI HCHO vertical column density (VCD)26 for Jan 16 compared with other observations taken around the same time by TROPOMI, VIIRS (on board of SNNP), MLS, and geostationary satellites. The Jan 16 observation shows a very strong HCHO enhancement over New Caledonia that is aligned with enhancements in SO2 (also detected by TROPOMI), aerosol optical depth (detected by VIIRS), as well as sulfate aerosol (SA, see Methods on how this is derived from EUMETSAT Volcanic Ash RGB). Correlation between HCHO, SO2, and aerosols within the Hunga Tonga-Hunga Ha’apai plume on 16 Jan 2022. A HCHO Vertical Column Density (VCD) (without cloud correction) compared with MLS measurements of ClO. MLS data with an X did not pass quality screening. B SO2 column density compared with MLS SO2 measurements. C Modified EUMETSAT Geostationary Ring Volcanic Ash RGB – Multimission (downloaded on 1 May 2025), showing only the green channel to highlight the detection of the sulfate aerosol plume, compared with MLS CO measurements. D VIIRS aerosol optical depth (AOD) at 550 nm compared with MLS O3 measurements. E Correlation between SO2 and HCHO showing different enhancement ratios for cloud C1 and C2. F Correlation between sulfate aerosol and HCHO shows a similar enhancement ratio for both clouds. Two clouds, C1 and C2, can be distinguished, with C1 at a slightly higher altitude (30–32 km) than C2 (27–29 km) as observed by CALIOP lidar27. The alignment in the location and movement of the HCHO, SO2, and particle enhancements implies that the HCHO enhancement is at the same high altitude, which is unusual for HCHO, which normally is mainly present in the troposphere. There is a very strong correlation between HCHO and SO2 for both C1 and C2 (r = 0.8), with a ∆HCHO/∆SO2 enhancement ratio that is 4.5 times higher for C1 (see Table 1, we will refer to observations as Cx_y, with x referring to the cloud 1/2 and y referring to the date in Jan-2022). However, while correlation is equally high with sulfate aerosol (r > 0.8), the enhancement ratio ∆HCHO/∆SA is more similar for C1 and C2. This suggests that the HCHO enhancement is more directly related to aerosols and more indirectly to SO2. We note that previous observations also found that SO2:SA ratios differ in different parts of the HHTH plume28. We found that SO2 spectral interference affected HCHO observations in C2 by +40 % on the initial day, but had a low impact of +10 % on observations in C1 (see Methods). By Jan 19, SO2 concentrations have dropped by an order of magnitude and no longer affect the spectral fits. Table 1Overview of HCHO enhancement ratios that could be confidently assessedObservation ***∆HCHO/∆SO2 (mmol/mol) *∆HCHO/∆SA (molec/cm2 per green value) **∆HCHO/∆AOD (molec/cm2 per AOD) *∆HCHO/∆BrO (mol/mol) *C1_164.5 ± 11% (r = 0.76)2.0E + 13 ± 12% (r = 0.81)1.3E + 15 ± 08% (r = 0.86)41.1 ± 17% (r = 0.62)C1a_1940.0 ± 07% (r = 0.57)invalid datainvalid data12.1 ± 29% (r = 0.17)C1b_20SO2 < DL6.8E + 12 ± 22% (r = 0.41)4.0E + 14 ± 12% (r = 0.54)19.0 ± 22% (r = 0.27)C2_161.0 ± 06% (r = 0.80)1.0E + 13 ± 08% (r = 0.85)invalid data5.5 ± 10% (r = 0.65)C2_171.1 ± 09% (r = 0.59)7.3E + 12 ± 12% (r = 0.61)2.2E + 14 ± 16% (r = 0.48)6.4 ± 08% (r = 0.62)C2a_202.2 ± 20% (r = 0.42)2.0E + 12 ± 37% (r = 0.32)1.7E + 14 ± 27% (r = 0.38)6.5 ± 23% (r = 0.37)C2b_21SO2 < DL5.3E + 12 ± 14% (r = 0.54)3.7E + 14 ± 14% (r = 0.53)21.8 ± 11% (r = 0.51)C2b_22SO2 < DL3.7E + 12 ± 29% (r = 0.33)1.9E + 14 ± 25% (r = 0.35)6.4 ± 32% (r = 0.20)C2b_25SO2 < DL7.3E + 12 ± 27% (r = 0.31)2.6E + 14 ± 15% (r = 0.39)10.1 ± 20% (r = 0.24)C3_23No correlation (r = -0.04)invalid dataNo correlation (r = -0.07)No correlation (r = -0.36)Showing ± standard error. SA Sulfate Aerosol, AOD aerosol optical depth, DL detection limit.* after compensating for cloud correction and air mass factor (AMF) error (factor of 4.85)** Only compensating for AMF error (factor of 4.85).*** Observations are referred to as Cx_y, with x referring to the observation and y referring to the date in Jan-2022. Coordinates are listed in Table S3. Overview of HCHO enhancement ratios that could be confidently assessed Showing ± standard error. SA Sulfate Aerosol, AOD aerosol optical depth, DL detection limit. * after compensating for cloud correction and air mass factor (AMF) error (factor of 4.85) ** Only compensating for AMF error (factor of 4.85). *** Observations are referred to as Cx_y, with x referring to the observation and y referring to the date in Jan-2022. Coordinates are listed in Table S3. We quantified the total HCHO enhancement (see Methods) by integrating the VCD over the total surface area, and subtracting background HCHO estimated north and south of the aerosol cloud. We also quantified the total HCHO enhancement by multiplying ∆HCHO/∆SO2 by the total integrated amount of SO2 and did the same with ∆HCHO/∆SA. The three quantification methods are in good agreement (see Table 2), showing a total integrated HCHO enhancement of 11.8 ± 2.7 × 106 mol of HCHO for Jan-16 (see Table 2, for clarity we use 1 mol = 6.02 × 1023 molec). Table 2Overview of total HCHO enhancements that could be confidently assessedObservation ***Total HCHO enhancement above baseline (mol)*Total HCHO via SO2 correlation (mol) *Total HCHO via SA correlation (mol) **Total HCHO via VIIRS AOD correlation (mol) **C1_166.6E + 06 ± 12%6.2E + 06± 11%8.4E + 06 ± 12%1.1E + 07± 08%C1a_19Invalid baseline6.6E + 06± 07%Invalid dataInvalid dataC1b_20Invalid baselineSO2 < DL>3.2E + 06 ± 22%>4.0E + 06± 12%C2_165.7E + 06 ± 08%5.0E + 06± 06%3.5E + 06 ± 08%Invalid dataC2_176.0E + 06 ± 12%5.3E + 06± 09%5.3E + 06 ± 12%Invalid dataC2a_20Invalid baseline>1.4E + 06± 20%>6.1E + 05 ± 37%>1.0E + 06 ± 27%C2b_21>3.3E + 06 ± 14%SO2 < DL>3.7E + 06 ± 14%>3.4E + 06 ± 14%C2b_22>2.7E + 06 ± 29%SO2 < DL>1.4E + 06 ± 29%>1.1E + 06 ± 25%C2b_25>5.4E + 06 ± 27%SO2 < DL>1.4E + 06 ± 27%>4.6E + 06 ± 15%C3_23No correlationNo correlationNo correlationNo correlationShowing ± standard error. SA Sulfate Aerosol, AOD aerosol optical depth, DL detection limit.* After compensating for cloud correction and air mass factor (AMF) error (factor of 4.85).** only compensating for AMF error (factor of 4.85).> Means only partial quantification was possible.*** Observations are referred to as Cx_y, with x referring to the observation and y referring to the date in Jan-2022. Coordinates are listed in Table S3. Overview of total HCHO enhancements that could be confidently assessed Showing ± standard error. SA Sulfate Aerosol, AOD aerosol optical depth, DL detection limit. * After compensating for cloud correction and air mass factor (AMF) error (factor of 4.85). ** only compensating for AMF error (factor of 4.85). > Means only partial quantification was possible. *** Observations are referred to as Cx_y, with x referring to the observation and y referring to the date in Jan-2022. Coordinates are listed in Table S3. The highest HCHO VCD enhancement that we observed was 1.6 × 1015 molec/cm2 ± 10% in cloud C1_16 (after sensitivity correction, see Methods). Using a 2 km layer thickness27 this means the peak HCHO concentration was 8 × 109 molec cm−3 ± 10% (12 ppb ± 10% at 20 mbar and 220 K). This stratospheric HCHO concentration is unusually high, with previous observations showing maximum HCHO below 0.1 ppb at 20 km altitude related to biomass burning emissions29. Figure 2 shows the HCHO photolysis rates calculated by the Tropospheric Ultraviolet and Visible (TUV) Radiation Model30, for 25 km altitude at the coordinates of the HCHO enhancement for Jan 15 to Jan 17, scaled to 58% of the TUV rate as modeled by Zhu et al.25 to represent conditions inside the plume. Based on this HCHO photolysis rate, its lifetime at around the TROPOMI overpass time (13:30) is ca. 2.5 hours, and any HCHO emitted by the eruption would have been 95% removed by the Jan 16 TROPOMI overpass, and by 99.95% for the Jan 17 overpass. However, the HCHO enhancement remained visible on the following days, indicating continuous HCHO production. HCHO in-plume photolysis rate as a function of time, starting from the moment of the eruption. The solid line shows the photolysis rate based on the Tropospheric Ultraviolet and Visible (TUV) Radiation Model30, scaled to 58% of the TUV rate as modelled by Zhu et al.25 to represent conditions inside the plume. At the TROPOMI overpass time (13:30, marked with a vertical line), a photolysis rate of 1.14 × 10−4 s−1 corresponds with a HCHO lifetime of 2.5 hours. The dashed line shows the percentage of HCHO remaining if it were injected by the eruption without in-plume formation, showing 95% HCHO reduction for the Jan 16 TROPOMI overpass, and 99.95% reduction for the second overpass. On Jan 17, cloud C2 is located between the Australian coast and New Caledonia (see Figs. 3 and S3), and has a moderately strong correlation with SO2 and sulfate aerosol (r = 0.6), despite interference from biomass burning HCHO in continental outflow. For C2, the total integrated HCHO enhancement and the ∆HCHO/∆SO2 ratio are similar to Jan 16, with a 30 % lower ∆HCHO/∆SA ratio. We did not quantify the HCHO enhancement for C1 because it is located directly above an Australian biomass burning region and has a high cloud fraction. Comparison of several observed HCHO enhancements in vertical column density (VCD) with coincident simultaneous MLS v5 observations. Showing HCHO (column 1), aerosol optical depth (AOD) (column 2), and MLS observations (columns 3 and 4) for clouds C2_17 (top), C2b_21 (middle), and C2b_25 (bottom). While initially MLS H2O enhancement is co-located with HCHO/AOD/HO2 enhancement, by Jan 25, the H2O enhancement becomes separated from the other enhancements. MLS data marked with an ‘X’ did not pass quality screening. We did not apply quality screening to MLS H2O, but the data can be trusted to show the location of the H2O enhancement21. On subsequent days, C1 and C2 disperse into elongated shapes that are too large for total integrated HCHO quantification due to interference by clouds and continental HCHO sources. However, we still quantified fractions of these clouds and determined their enhancement ratios (see Figs. S2–S19). On Jan 20, cloud C1 was identified over the ocean by the sulfate and AOD correlations, while SO2 was almost fully removed (Figures S8, S9). Section C1b_20 could be confidently quantified based on its correlation with AOD (r = 0.54), and compared to Jan 16, the ∆HCHO/∆SA and ∆HCHO/∆AOD enhancement ratios are 70% lower. Nevertheless, at least 36% of the total integrated HCHO enhancement remained compared to Jan 16. For C2 we were able to assess cloud fractions until Jan 25 (Figs. 3 and S18, S19), and found ∆HCHO/∆SA and ∆HCHO/∆AOD enhancement ratios that remained stable after Jan 17, while SO2 became totally removed. For Jan 23, we assessed an SO2 cloud C3 that was not associated with high AOD or sulfate aerosol, and was found not to contain any HCHO enhancement (Fig. S15). We analyzed the correlation between TROPOMI BrO and HCHO enhancement (see Table 1), and found a generally linear correlation with ∆HCHO/∆BrO varying from 6 to 40 mol/mol. The correlation between BrO and HCHO is generally weaker compared to the AOD/HCHO correlation, with a poor BrO/HCHO correlation for cloud C1. For cloud C2, the ∆HCHO/∆BrO enhancement ratio remained stable between Jan 16 and Jan 25. We compared the observed HCHO enhancements with coincident simultaneous MLS v5 observations of ClO, SO2, O3, CO, H2O, HO2 and HCl (Fig. 3). In the initial days, most of the MLS data did not pass the standard Quality Screening (MLS), which was attributed to extremely enhanced H2O at very high altitudes21. Fortunately, some of the clouds in Table 3 had suitable MLS observations, especially C2b_21 (Figs. 3 and S11). MLS H2O is enhanced within the HCHO plume, but the H2O data generally did not pass quality screening. According to Millán et al., the data can be trusted to show the location of the H2O enhancement, but the absolute values may have large errors21. MLS SO2 observations were in agreement with TROPOMI observations. HCHO enhancement was consistently associated with MLS ClO enhancement, but on the first days did not pass quality screening. However, for C2b_21 there was a high-quality MLS measurement for a confident HCHO enhancement (0.7 ppb ClO for C2b_21). We find an O3 depletion within the ClO/HCHO enhancement, in agreement with previous research25,31. For example, O3 decreased inside the HCHO enhancement from 6.6 to 5.9 ppm in C2_16, from 6.6 to 5.7 ppm in C2_17, and from 6.5 to 6.1 ppm in C2b_21. CO is elevated within the HCHO plume, with the CO enhancement decreasing over time. The highest enhancement was observed for C2_17, where CO increased from 30 to 130 ppb. CO was elevated by 30 ppb for C1b_20, and by 10 ppb for C2b_21. This was also observed by other researchers and may be due to the combination of an initial updraft of tropospheric air during the eruption, combined with continued CO production in the HTHH plume25. For HCl, we find a potential enhancement on Jan 16 and Jan 17, but this data did not pass quality screening. For C2b_21, we found no correlation between HCl and the HCHO enhancement. This is surprising, because other researchers found a slight depletion of HCl compared to the seasonal average31. HO2 within the HCHO enhancement is initially strongly elevated (for example, for C2_17 increased to 4.4 ppb from a baseline of 0.5 ppb; on this day, the HO2 data passed quality screening but should still be treated with caution), while later the enhancement is smaller (for example, for C2b_21 we find a 0.5 ppb increase in HO2 within the HCHO enhancement). Discussion The observed HCHO enhancement in the HTHH plume can only be explained by in-plume production. First, the HCHO photolysis lifetime, 2.5 hours at midday, is too short for HCHO emitted by the volcano to remain 20 hours later (see Fig. 2). Second, without in-plume formation, this would reduce HCHO concentrations by 100× between the Jan 16 and Jan 17 overpasses, while on Jan 17, we find a total integrated HCHO enhancement that is approximately the same as for Jan 16. Furthermore, we even detect HCHO enhancements up to Jan 25. We have observed a linear correlation between HCHO and aerosols (SA and AOD), which suggests that the HCHO lifetime at midday is mainly driven by photolysis (see Methods Eq. 1). If the HCHO lifetime were limited by OH or Cl produced by the aerosols, then the correlation would not be linear (see methods Eq. 2). To be consistent with this, HCHO loss to OH and Cl should be below 50%, resulting in a maximum concentration of 1 × 107 cm−3 for OH and 2 × 106 cm−3 for Cl, based on reaction rates (see Table S1 and Fig. S1). By dividing the HCHO enhancement with a 2.5-hour photolysis lifetime, we derive the total HCHO production rate on Jan 16: 4.7 ± 1.1 × 106 mol per hour at midday, with peak values of 9 × 105 cm−3 s−1 ± 10% (5 ppb/hour ± 10%). This estimation excludes (minor) HCHO loss to OH and Cl and is therefore a lower estimate. In the absence of other HCHO sources, the CH4 oxidation rate is approximately equal to this amount, amounting to 75±18 Mg CH4/hour at midday. For Jan 20 and Jan 21, we found HCHO enhancements of up to 0.2 × 1015 molec/cm2 (after sensitivity correction), which corresponds to a concentration of 1 × 109 cm−3 (1.6 ppb HCHO), and an HCHO production rate of 1.2 × 105 cm−3 s−1 (0.7 ppb/hour). We calculated a total CH4 oxidation of 900 ± 220 Mg/day (with local peak values of 60 ppb/day on Jan 16 and 8 ppb/day for Jan 20 and 21) in the volcanic plume by assuming the Cl is produced by a photochemical source, using the NO2 photolysis rate to scale the hourly oxidation rate (see Fig. 4). Considering that CO lifetime inside the HTHH plume is reduced to a few days by elevated OH25, the continued observation of 10-100 ppb CO enhancement in the MLS observations provides further evidence of in-plume production, and fits well with the observed 8 – 60 ppb/day CH4 oxidation. Our observations did not show a decline in the total integrated HCHO enhancement, which suggests this rate of CH4 oxidation may have continued for at least 10 days. This is surprisingly high in view of typical stratospheric CH4 background concentrations of 1 ppm, and suggests that CH4 concentrations in the HTHH plume were elevated. Quantification of total CH4 oxidation in the Hunga Tonga-Hunga Ha’apai (HHTH) plume. Showing calculated hourly (solid line) and cumulative (dashed line) CH4 oxidation in the HHTH plume, based on the observed HCHO production. We calculated the minimum required methane elevation by combining the observed methane oxidation rate with the maximum possible OH and Cl concentrations, and with the known reaction rates for methane oxidation by Cl and OH (see Table S1). On Jan 16, the area of the HCHO enhancement is approximately 7.2 × 106 km2 and the thickness is 2 km27. This yields an average CH4 oxidation rate of 5.5 × 106 cm3 s−1. If 100 % of this is due to a maximum OH enhancement of 1 × 107 cm−3, the CH4 concentration was at least 95 ppm compared to a background value of 1 ppm (an enhancement of at least 2300 Gg CH4). If instead 90 % is due to a maximum Cl enhancement of 2 × 106 cm−3, the CH4 concentration was at least 14 ppm (330 Gg CH4). For explosive eruptions the CO2:SO2 molar ratio can be up to 3032. Using an estimated 18.8 Tg of SO2 emission before ocean uptake24 leads to an estimated CO2 emission of 390 Tg CO2. The range of measured volcanic CO2/CH4 emission ratios is 10 to 105, with higher values for higher volcanic activity33, which leads to an estimated CH4 emission of 4 – 40,000 Gg, and fits our estimates based on either OH or Cl. We examined ACE-FTS data34 for evidence of CH4 enhancement in the HTHH plume. The earliest ACE-FTS encounter with the HTHH plume occurred 3 weeks after the eruption, on 6 Feb 202235. Unfortunately, due to pointing jumps caused by high aerosol extinction, CH4 observations are invalid for this day. In the monthly average CH4 profiles for 2022, we did not find evidence of a CH4 enhancement that is above the natural swings of 0.05 ppm CH4 with varying altitude (see Fig. S20), while a H2O enhancement of 7.4 ppm can be seen at 26 km altitude in the Feb 2022 average. The absence of a CH4 enhancement in the ACE-FTS profiles, therefore, implies that the H2O:CH4 ratio in the HTHH emission was less than 150, suggesting the eruption emitted less than 1000 Gg CH4 to the stratosphere. This rules out the possibility that the HCHO enhancement is due to OH (it would mean at least 2300 Gg CH4 emission, which would be clearly visible using ACE-FTS). However, the emission of 330 Gg CH4 due to Cl enhancement is realistic and would indeed not have been detectable with ACE-FTS in Feb 2022. Thus, the observed HCHO enhancement is due to an increase in Cl, combined with an average methane concentration enhancement of at least 14 ppm in the Jan 16 volcanic cloud. This corresponds to a CH4 vertical column density enhancement of 0.18 × 1019 molec/cm2, which is around 4% of a typical background measured by TROPOMI, of 4.2 × 1019 molec/cm2. By Jan 20, the CH4 concentration enhancement is expected to have dispersed by an order of magnitude, in line with the observed lower HCHO concentration enhancement. At this later time, the methane enhancement is therefore also too low to detect with TROPOMI. The observed HCHO production is sufficiently high that the majority of HCHO could only have been produced by CH4 oxidation (the main precursor for HCHO in the stratosphere), and not from non-methane VOCs (NMVOCs) emitted by the volcano. NMVOCs are known to be emitted by volcanoes, but only at trace concentrations that are at least an order of magnitude less than methane36. In addition, the seawater concentrations of DMS (1–10 nM)37 and dissolved organic carbon (maximum 100 uM)38 are too low to cause a substantial injection of carbon through the 146 Tg stratospheric H2O injection. Zhu et al.25 used the Whole Atmosphere Community Climate Model version 6 to analyze the chemistry leading to ozone depletion inside the HTHH plume during the first days following the eruption, by constraining the model with MLS observations. For Jan 20 during the daytime, they were able to explain the O3 depletion in the HTHH plume using a mechanism of HOCl uptake resulting in [Cl] = 6 × 10−14 (mixing ratio) = 4 × 104 cm−3, and a rate of the Cl + CH4 reaction of 900 cm−3 s−1, which is 130× lower than our inferred rate of 1.2 × 105 cm−3 s−1 (we discuss this mechanism later in the discussion). We also note that the Zhu model found 80 ppt HO2, while MLS observed HO2 concentrations around 500 ppt in the HTHH plume (6× more) on Jan 21. The problem is to explain why the observed HCHO and HO2 concentrations are significantly higher than in the Zhu model output. We propose that the elevated HCHO and HO2 concentrations arise from the injection of volcanic CH4 into the stratosphere by the HTHH eruption. Increased CH4 causes more Cl to react with CH4 instead of O3, producing HCl and HCHO. This terminates the chain reactions causing O3 depletion in the HTHH plume, such as cycles involving ClO + O, ClO + NO2, and ClO + HO2. Another effect is that enhanced HCHO production leads to enhanced HO2 production, which can produce O3 via reaction with NO2. A higher primary production of active chlorine is also required to explain the simultaneous loss of O3 and production of HCHO. Based on previous research39–41 we estimate 0.2 g CH4 is oxidized per g primary Cl produced – accounting for secondary impacts on OH formation and radical chain length. By primary Cl production, we mean the additional Cl that is added to the atmosphere. This leads to an estimated primary Cl production of 375 ± 90 Mg Cl/hour at midday and 4.5 ± 1 Gg Cl/day based on our Jan 16 observations. Peak midday rates are 2.2 × 106 cm−3 s−1 ± 10 % for Jan 16 and 0.3 × 106 cm−3 s−1 for Jan 21 (using 0.4 CH4 molecules per Cl atom). This amount may seem similar to the injection of 1.3 Gg ClO used by Zhu25, but the difference is that to explain both HCHO production and O3 loss, this amount needs to be injected daily. Bromine chemistry is a key mechanism for chlorine activation in a typical volcanic plume42,43. This is a catalytic cycle in which Br activates Cl while depleting ozone. According to Zhu, bromine chemistry cannot explain the observed Cl production in the HHTH plume because it implies a much stronger ozone depletion than was observed25. In addition, BrO was observed at a different time during the HTHH eruptions and reached a lower altitude, 8-15 km, where the different wind direction spread the BrO in the opposite southeastward direction compared to the plume that we investigate here20. Bromine catalytic cycling is constrained by the Br + HCHO reaction that forms HBr43. This shifts bromine speciation towards HBr within our observed strong HCHO enhancements, possibly explaining the relatively low observed BrO compared to SO2 in the high-altitude stratospheric HHTH plume that we investigate, and limiting bromine chemistry as a Cl source. This also means bromine emissions might have been higher than current BrO-based estimates. Despite these arguments, we still observe a modest BrO enhancement, and it is correlated with HCHO (see Table 1, and Figs. S2–S19). We calculated the maximum rate of Cl production through bromine chemistry by calculating the rate of formation and reactive uptake of HOBr using observed values for BrO, HO2, and aerosol surface area for cloud C2b_21 (see Supplemental Information Text). We find that the maximum Cl production is 1.5 × 104 cm−3 s−1, while our observed value is an order of magnitude larger at 3 × 105 cm−3 s−1. We therefore conclude that Br activation of Cl cannot explain the majority of our observed Cl production. Previous studies attributed the Cl chemistry to an initial volcanic injection of active Cl, followed by chlorine recycling, especially via ClO + HO2, forming HOCl (see Table S1)25,31. Under normal conditions, these recycling mechanisms lead to ozone depletion in which chlorine is catalytic, and it is constrained because there is no production of chlorine to compensate for the loss of active chlorine to CH4 + Cl. However, under the conditions of high aerosol surface area in the HTHH plume, it is possible for chlorine recycling to amplify the total amount of active chlorine (see Fig. S21). The main pathway for this starts with 1 chlorine atom forming HOCl, followed by the reactive uptake reaction of HOCl + HCl that forms Cl2, which photolyzes, yielding two chlorine atoms31. We calculated that this mechanism could theoretically reach rates that are high enough to explain some of our observed chlorine production (see Supplemental Information Text). However, the mechanism is driven by the strong dependence of the reactive uptake probability γ on H2O concentration, which decreases substantially during the days covered by our observations. This mechanism may therefore explain why we observe relatively higher enhancement ratios on Jan 16, but does not explain our observation that the ∆HCHO/∆AOD enhancement ratios remained stable after Jan 17. We also note that by Jan 25 the H2O enhancement becomes partly separated from the ClO/HO2 enhancement (see Fig. S19), which does not fit with a Cl source that depends on H2O concentration, and is a strong argument for why HOCl reactive uptake cannot explain the long-term primary chlorine production implied by our observed HCHO enhancement. Iron photochemistry in volcanic ash as a chlorine source We propose that another possible chlorine source could be iron photochemistry, similar to chlorine production by mineral dust aerosols mixed with sea spray over the North Atlantic39. It is estimated that the HTHH eruption released up to 32 kt of iron into the South Pacific Ocean44. However, most of the emitted iron is deposited close to the volcano, as is evident from the resulting phytoplankton bloom45, with only a small fraction reaching the stratosphere. In contrast to initial studies that concluded fine volcanic ash particles were rapidly washed out27, more recent studies found that fine volcanic ash particles were more likely to remain in suspension, and were difficult to distinguish from more chemically pristine sulfate particles due to a sulfate coating that gave them sulfate-like absorbing properties46–49. Romeo et al. used observations and modeling to estimate that between 1.2 and 3.8 × 1011 g of fine ash reached the stratospheric cloud48. Using an estimated Fe mass fraction of 2–8%50 leads to an estimated 2.4–30 Gg Fe emission. If we assume 2% of the iron is photoactive (similar to mineral dust)39, the observed chlorine production of 4.5 ± 1 Gg Cl2 per day implies a production rate between 7 and 94 g Cl2 per g photoactive Fe per day. This fits well with the observed value for mineral dust in the marine boundary layer of 70 g Cl2/g Fe per day, especially considering that conditions in the stratosphere are very different, and that sulfate is known to reduce chlorine production in iron photochemistry by up to 40%51,52. Using the observed aerosol surface area density of 2.9 × 10−6 cm2 cm−3 for Jan 2131 combined with an estimated average particle size of 1 μm for coated ash46 (based on a mix of pristine sulfate aerosols of 0.5 μm and course volcanic ash of 4.6 μm), leads to an estimated aerosol mass of 110 μg/m3, which is reasonable for a translucent plume that is visible on true color satellite images (see Figure S11). Using 50% volcanic ash by weight46, and again using 2 – 8% Fe mass fraction of which 2% is photoactive, our observed Cl production rate of 0.3 × 106 cm−3 s−1 for Jan 21 implies a Fe catalytic cycling rate of 1.1–4.5 per hour. Considering the different stratospheric conditions and the presence of sulfate, this corresponds well with the observed value of 11 per hour for mineral dust39 and values ranging from 6-78 hr−1 in laboratory studies53. Based on the above, we conclude that iron photochemistry is a plausible source for active chlorine in the HTHH plume. This analysis suggests that iron–chloride photochemistry may be active in the stratosphere, but confirmation will require dedicated modeling and laboratory studies (e.g., a global or plume-resolving model including iron photochemistry and methane injection). The iron-chloride photochemistry mechanism may not be as significant in other volcanic eruptions, because the HHTH eruption provided unique conditions favorable to iron-chloride photochemistry. This includes the exceptionally large seawater injection that also injected a large amount of sea salt needed for the mechanism. At the same time, the SO2 emission was relatively modest (reducing potential inhibition by sulfate). Application of methane removal as a quantification method We present a methodology for satellite quantification of enhanced atmospheric methane oxidation based on satellite quantification of HCHO, a short-lived intermediate in the CH4 oxidation mechanism. A key advantage is that this approach is especially sensitive to CH4 oxidation, and it works over ocean surfaces where satellite-based CH4 measurements are limited. The use of the methodology is limited by interference from local HCHO sources, but this can be partly overcome through correlations with additional observations such as aerosol optical depth. When we applied the methodology to the stratospheric plume from the HTHH eruption, we found the highest HCHO enhancement ever recorded in the stratosphere (up to 12 ppb at 30 km altitude), and that the HCHO enhancement persisted for weeks and possibly months. We attributed the HCHO enhancement to a total CH4 oxidation of 900 ± 220 ton/day, with a peak rate of 60 ppb/day on Jan 16. Such a large amount of CH4 oxidation implies that the HTHH eruption must have injected elevated levels of CH4 into the stratosphere. Meidan et al.41 modelled local emission of iron for atmospheric methane removal over the ocean and found 25 Gg Cl per hour removed 3.1 Gg CH4 per hour, reducing global radiative forcing by 0.04 W m−2 within 10 years. This removal amount is much higher than our observed HHTH removal of 75 ±18 Mg CH4 per hour at midday, which was clearly detectable. Therefore, the sensitivity of our methodology can be sufficient for quantification in hypothetical future enhanced atmospheric methane oxidation approaches to help address future global warming. Methods Methodology Our method is to use TROPOMI HCHO observations to quantify enhanced methane oxidation in the HTHH plume, and to use this to derive the required Cl production. Our hypothesis is that locally enhanced atmospheric methane (CH4) oxidation is revealed using TROPOMI observations of gas-phase species that are involved in CH4 oxidation, in particular formaldehyde (HCHO). CH4 oxidation is mainly initiated by the hydroxyl (OH) radical reaction. In addition, a few percent of CH4 reacts with chlorine atoms (Cl)54. Each oxidized CH4 molecule leads to the production of approximately one additional HCHO molecule. HCHO is an intermediate species that breaks down to form CO within a few hours. CO is a stable intermediate species with a lifetime of about 30 days in the troposphere that eventually oxidizes into CO2. In the absence of local HCHO sources such as biomass burning, CH4 oxidation is the main source of HCHO. The main sinks for HCHO are two photolysis reactions and its reactions with OH and Cl radicals (see Table S1 and S2). Using a steady-state approximation, the concentration of HCHO is given by Eq. 1 in which kCH4+OH and kCH4+Cl are the rate coefficients for the reactions of methane with OH and Cl, kHCHO_OH and kHCHO_Cl are the rate coefficients for the reaction of HCHO with Cl and OH, and j1 and j2 are the photolysis rate coefficients for HCHO with one channel generating radical (H and CHO) and one molecular (H2 and CO) species. In typical marine boundary layer conditions, photolysis is the main sink for HCHO, with j1 + j2 below 10−4 s−1 40, depending on season, cloudiness and altitude, while the loss to OH is not insignificant (5 × 10−5 s−1 for [OH] of 6 × 106 cm−3). This implies that for small chlorine enhancements, the concentration of HCHO will increase linearly with the rate of methane oxidation LCH4. For higher chlorine concentrations (above 2 × 106 cm−3), the concentration of HCHO no longer increases linearly, due to increasing loss by reaction with Cl. For extremely high Cl concentrations (above 107 cm−3), both the source and loss of HCHO are dominated by Cl. Here, the concentration of HCHO can be approximated using Eq. 2 with [HCHO] stabilizing at a fixed value that depends on methane concentration instead of oxidation rate. For example, using 1940 ppb CH4, Eq. 2 yields 2.75 ppb HCHO for the marine boundary layer. Sulfate aerosol (SA) detection Sulfate aerosol is represented by the green channel of the EUMETSAT Volcanic Ash RGB – that is based on analysis of the infrared channel of several geostationary satellites, and was shown to be correlated with sulfate aerosol in the HTHH plume27. In our analysis, we extracted the green channel value, resulting in a number between 0 and 255 that scales with sulfate load. HCHO quantification from TROPOMI observation To quantify HCHO enhancement (Etot) in the volcanic plume, we follow this protocol: Quantification by enhancement relative to baseline: in this method, we estimate the enhancement along north-south or east-west transects, by estimating the baseline outside the enhancement region: Etot=∑i=1N[HCHO]i−[HCHO]i,baselineAi, in which [HCHO]i is the HCHO VCD for pixel i, [HCHO]i,baseline is the baseline value, Ai is the surface area represented by pixel i, and N is the number of pixels in the area of interest. Quantification by enhancement relative to baseline: in this method, we estimate the enhancement along north-south or east-west transects, by estimating the baseline outside the enhancement region: Etot=∑i=1N[HCHO]i−[HCHO]i,baselineAi, in which [HCHO]i is the HCHO VCD for pixel i, [HCHO]i,baseline is the baseline value, Ai is the surface area represented by pixel i, and N is the number of pixels in the area of interest. Quantification by correlation: in this method, we determine the ∆HCHO/∆X enhancement ratio, in which X may be SO2, SA, or aerosol optical depth. The total enhancement is then the sum of the enhancement in X multiplied by the enhancement ratio: Etot=ΔHCHOΔX∑i=1NXi−Xi,baselineAi, where Xi,baseline is 0 for SO2 and AOD. This methodology makes it easier to quantify the HCHO enhancement in the presence of local HCHO sources (that don’t emit species X), and provides a more accurate quantification by combining additional data. Quantification by correlation: in this method, we determine the ∆HCHO/∆X enhancement ratio, in which X may be SO2, SA, or aerosol optical depth. The total enhancement is then the sum of the enhancement in X multiplied by the enhancement ratio: Etot=ΔHCHOΔX∑i=1NXi−Xi,baselineAi, where Xi,baseline is 0 for SO2 and AOD. This methodology makes it easier to quantify the HCHO enhancement in the presence of local HCHO sources (that don’t emit species X), and provides a more accurate quantification by combining additional data. Correction for air mass factor (AMF) in stratospheric HCHO observation Vertical column density (VCD) represents the amount of a trace gas integrated straight down through the atmosphere, while slant column density (SCD) is the amount measured along the actual, typically angled, light path through the atmosphere. Because the slanted path is longer than the vertical path, the SCD is usually larger than the VCD for the same atmospheric state. The air mass factor (AMF) relates the two by VCD = SCD/AMF. The value of the AMF depends on the length of the light path, the vertical distribution of absorbing trace gases in the atmosphere, the reflectivity (albedo) of the earth’s surface, the presence of aerosols and clouds, etc. The original air mass factor (AMF) that was used to calculate the VCD of HCHO in Figs. 1 and 3 assumed that HCHO is present in the troposphere. However, the enhancement is present in the stratosphere, where the TROPOMI sensor is more sensitive to HCHO, requiring a modified analysis. Instead of an AMF of 1.3 for tropospheric background HCHO, we used the altitude-resolved air mass factors of the HCHO retrievals55, and took a value for a stratospheric layer around 25 km of 6.3 (correction factor 4.85). In addition, the stratospheric HCHO enhancement is above the clouds, which means it should not be corrected for clouds. To compensate for this, we first determine the HCHO enhancement using cloud-corrected HCHO VCD (as shown in Figs. 1 and 3), and then we apply an AMF correction factor (6.3/1.3) and reverse the cloud correction. Role of spectral and aerosol interference for HCHO observations To check for interference by SO2 in the spectral fits for HCHO we used data from Jan 16 (with the highest SO2 signal), and compared the HCHO slant column density (SCD) fits with and without SO2 (see Fig. S22). Including SO2 in the fits reduces the peak HCHO SCD in C2_16 by about 2 × 1015 molec/cm2, and about 1 × 1015 molec/cm2 in C1. Using an AMF of 6.3, this gives a difference in the observed peak VCD of 0.3 × 1015 molec/cm2 in C2_16, compared to an observed peak enhancement of 0.8 × 1015 molec/cm2 (40%). There is a difference of 0.15 × 1015 molec/cm2 in C1_16, compared to an observed enhancement of 1.6 × 1015 molec/cm2 (10%). The influence of aerosols on HCHO sensitivity (AMF) is very complicated and depends on aerosol type and altitude relative to the HCHO plume. Both cloud and aerosols can have a shielding effect if the cloud is located above the HCHO plume, but also a reflecting effect if the cloud is located under the plume (in the same way as a bright surface). In addition, sensitivity is generally increased when scattering aerosols (like sulfate aerosols) are vertically collocated with HCHO, while sensitivity is decreased for absorbing aerosols (like volcanic ash) (56). Which effect will dominate depends on the cloud/aerosol properties, and this leads to uncertainties in the stratospheric HCHO observation of the HTHH volcanic plume with an estimated magnitude of ± 20%56. Despite these uncertainties, the detected signal is clear. The observation that ∆HCHO/∆SA and ∆HCHO/∆AOD enhancement ratios remain stable over multiple days, while aerosol scattering/absorbing properties are changing due to SO2 oxidation, provides additional confidence in the detected signal. In our analysis, we removed the cloud correction for the observed stratospheric HCHO, assuming the stratospheric HCHO is located above clouds and above the majority of the aerosols. This can lead to a slight underestimation of HCHO. Supplementary information Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Supplementary Data 2 Transparent Peer Review file Acknowledgements We thank Chris Boone of the ACE-FTS team for his valuable support in the interpretation of the ACE-FTS data. Part of the study was funded by Spark Climate Solutions (grant numbers AMR-24-007: JdL, AMR-24-009: ID, AMR-24-010: MvH, AMR-23-001: MSJ, AMR-23-002: TR). Author contributions Hypothesis: Mv.H. & Jd.L. Study design: Mv.H., I.D., D.M., A.S.L., M.S.J., T.R., Jd.L. Model simulations: Mv.H. Model analysis: Mv.H. Satellite data analysis: Mv.H., I.D., Jd.L. TROPOMI sensitivity analysis & corrections: ID. Writing – original draft: Mv.H. Writing – review & editing: Mv.H., I.D., D.M., A.S.L., M.S.J., T.R., Jd.L. Peer review Peer review information Nature Communications thanks Luke Surl and the other, anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Data availability The source data used for Tables 1 & 2 (Supplementary Data 1) and MLS data (Supplementary Data 2) used in the analysis are provided with the paper. The Sentinel 5 P HCHO, SO2, BrO, and AOD data are publicly available through the S5P-PAL Data Portal (https://data-portal.s5p-pal.com/) and the Copernicus Data Space Ecosystem (CDSE) (https://dataspace.copernicus.eu/). MLS v5 data and VIIRS AOD data are available from the NASA Goddard Space Flight Center Earth Sciences (GES) Data and Information Services Center (DISC) (https://disc.gsfc.nasa.gov/). ACE-FTS data is available through the DatabACE homepage (https://databace.scisat.ca/). Competing interests A patent application was filed about the application of satellite-based methane removal quantification to interventions that enhance atmospheric methane removal, which is distinct from the natural atmospheric methane removal enhancement studied here (patent applicant: MvH, inventors: MvH, application number: PCT/NL2026/050017, status: application, receiving office: Netherlands Patent Office / Netherlands). All other authors declare they have no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-72191-4. References Associated Data Supplementary Materials Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Supplementary Data 2 Transparent Peer Review file Data Availability Statement The source data used for Tables 1 & 2 (Supplementary Data 1) and MLS data (Supplementary Data 2) used in the analysis are provided with the paper. The Sentinel 5 P HCHO, SO2, BrO, and AOD data are publicly available through the S5P-PAL Data Portal (https://data-portal.s5p-pal.com/) and the Copernicus Data Space Ecosystem (CDSE) (https://dataspace.copernicus.eu/). MLS v5 data and VIIRS AOD data are available from the NASA Goddard Space Flight Center Earth Sciences (GES) Data and Information Services Center (DISC) (https://disc.gsfc.nasa.gov/). ACE-FTS data is available through the DatabACE homepage (https://databace.scisat.ca/).