Satellite quantification of enhanced methane oxidation applied to the stratospheric plume following Hunga Tonga-Hunga Ha’apai eruption

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 × 10⁷cm⁻³ for OH and 2 × 10⁶cm⁻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 × 10⁵cm⁻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 × 10⁹cm⁻³ (1.6 ppb HCHO), and an HCHO production rate of 1.2 × 10⁵cm⁻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.

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 × 10⁶km² and the thickness is 2 km²⁷. This yields an average CH₄ oxidation rate of 5.5 × 10⁶cm³s⁻¹. If 100 % of this is due to a maximum OH enhancement of 1 × 10⁷cm⁻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 × 10⁶cm⁻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 × 10⁴cm⁻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 × 10⁵cm⁻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 + NO_2, 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 research^39,40,41 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 × 10⁶cm⁻³ s⁻¹ ± 10 % for Jan 16 and 0.3 × 10⁶cm⁻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 plume^42,43. 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, HO_2, and aerosol surface area for cloud C2b_21 (see Supplemental Information Text). We find that the maximum Cl production is 1.5 × 10⁴cm⁻³ s⁻¹, while our observed value is an order of magnitude larger at 3 × 10⁵cm⁻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 Cl_2, 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 properties^46,47,48,49.

Romeo et al. used observations and modeling to estimate that between 1.2 and 3.8 × 10¹¹g 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 × 10⁻⁶cm²cm⁻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 × 10⁶cm⁻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.