1. Both water injection and evaporating ice determine the water vapor remaining in the stratosphere
Here we use the Whole Atmosphere Community Climate Model (WACCM) to simulate the persistent water vapor enhancement from 10 to 50 hPa observed by MLS version 4 (Figure 1a) from February to April. Zonal averages of simulated water vapor on March 1 shows good agreement with MLS observations (Figure S2). Figure 1a shows both the model (solid line) and MLS (dashed line) display a positive water vapor anomaly of 6 - 8 ppmv peaking at 30 hPa from February to April. The anomaly is averaged from 30˚S to 0˚ and calculated using February, March, and April profiles minus the profile on January 5. The simulation has wider vertical extent in March and April. Both the model and MLS (Figure 1a) show that the positive water vapor anomaly slowly ascends from February 10 to April 1, which is largely related to the ascending branch of Brewer-Dobson circulation 18(Figure S2b).
To be consistent with the MLS water anomalies4, we need to inject ~ 150 Tg of water in an area of ~ 7105 km2, which is about three times larger than the anvil size observed by NOAA's Geostationary Operational Environmental Satellite 17 on January 15, 2022. This large area is needed because the residual amount of water is determined by the ice vapor pressure curve as shown in Figure S3. If we inject the water into a smaller area, the model forms too much ice and cannot retain enough water at 30 hPa as observed. It is not unreasonable to inject in a larger area. MLS observations on Jan 16 (a day after the eruption) show the latitudinal spreading of water vapor anomaly is more than 10 degrees4. The simulated plume does not spread as fast as observed because it is hard for the global model to reproduce the vertical wind shear as observed due to the limited spatial resolution, especially for a tropical volcanic plume spreading in the first couple of days19. In the model, the injection is over 6 hours on January 15, 2022 between 25.5 km and 35 km with a majority of the water injected between 25.5 and 30 km (see methods for details). The model simulation (Figure 1b) shows that 10 Tg of ice falls out and 10 Tg of ice evaporates and contributes to the residual water vapor (~140 Tg) until April. In reality, it is possible more H2O was injected to form ice but these ice particles fell out because of their large sizes or forming aggregates with ash particles7.
2. The water injection significantly shortens the SO2 lifetime by providing abundant OH
An accurate volcanic SO2 lifetime is key for predicting the particle sizes in the volcanic cloud. With shorter lifetimes, the resulting aerosol concentration is higher, leading to more rapid coagulation and larger sulfate particles20. In the stratosphere, the volcanic SO2 lifetime is mainly determined by its reaction rate with OH, as well as the heterogeneous reaction on ash19. The depletion of OH in SO2-rich plumes slows the SO2/OH reaction8,19,21–23. However, during the HTHH eruption, the significant amount of water vapor injected rapidly increased OH (Figure S4) and shortened the SO2 lifetime as shown in Figure 2. Clegg & Abbatt24 demonstrated that SO2 uptake on the ice surfaces is insignificant, so this mechanism is not included in the model. The SO2 lifetime (the e-folding time) of the SO2only case (red line) is 28 days, while the SO2 lifetime of the SO2_H2O case (blue solid line) is 12 days. Zhu et al.19 showed that a considerable amount of SO2 might be undetectable because it falls below the detection limits of the instruments as it spreads through the atmosphere. However, the blue dashed line in Figure 2 suggests that this effect is small for this eruption for the first ten days.
3. The impact of enhanced water vapor on stratospheric aerosol optical depth and radiation
We compare the stratospheric aerosol optical depth (sAOD) between OMPS-LP data and two model cases in Figure 3. Without the water injection, the SO2only case has a very small sAOD in the first two weeks after the eruption, because SO2 converts slowly to H2SO4 with a lifetime of ~1 month (Figure 2). Both OMPS and the SO2_H2O case show almost immediate formation of a large amount of sulfate aerosol. OMPS LP sAOD retrieval during the first couple of days has large uncertainties when the volcanic clouds are optically thick and localized26. The backscatter coefficient at visible wavelengths for the SO2_H2O case and CALIOP shows a similar peak value of 0.005 to 0.01 km-1sr-1 (Figure S5). In addition, compared to the SO2only case, sulfate particles in the SO2_H2O case doubled the sAOD in February because sulfate forms faster in the SO2_H2O case, and the more abundant particles coagulate to larger sizes. Also, sulfate particles swell to be a little larger because of the enhanced background water vapor (Figure S6). Generally, OMPS-LP observes a faster spreading of the plume to the northern hemisphere than is simulated (Figure 3) and has higher optic values in the northern hemisphere in March (Figure S7). Sellitto et al.27 indicate the initial fast spreading of the HTHH aerosol plume is highly related to the strong cooling inside the plume. Our model is nudged to the observed meteorology, which doesn’t capture this strong local cooling. Also, as we discuss in Result 1, the global model cannot reproduce the vertical wind shear as observed due to the limited spatial resolution. On March 1, the SO2_H2O case and the OMPS observation are consistent regarding the vertical extent of the plume from 20-30 km, and the extension to 40˚S (Figure S7).
Because of the increase in the burden of sulfate, the volcanic plume creates a negative radiative effect of about -1 to -2 W/m2 in the perturbed areas in January and February (Figure 4). The global mean radiative effect at the top of the atmosphere in February is -0.12 W/m2 in the SO2only case and -0.20 W/m2 in the SO2_H2O case; the surface radiative effect is -0.16 W/m2 for the SO2only case and -0.21 W/m2 for the SO2_H2O case. These values are typical for middle-sized volcanic eruptions28. Even though the sAOD is approximately doubled in the SO2_H2O case, it only results in slightly more negative radiative forcing due to enhanced positive radiative forcing of the enhanced water vapor 29,30 in the SO2_H2O case.
4. Persistent volcanic water vapor and sulfate may impact stratospheric ozone chemistry
Here we focus our study on the transport of volcanic water vapor and sulfate in the Southern Hemisphere. We conduct three sets of 3-member ensemble simulations to explore the transport of volcanic water vapor and volcanic sulfur from January to October. The simulations are nudged to the observed meteorology until the end of March and then are free-running until October. Figure 5 shows that both H2O and sulfur are slowly transported to the south from January to May. After mid-June, the majority of H2O and sulfur reside between 30˚S and 60˚S. This is because the strong Antarctic vortex during June-July-August prevents volcanic materials from entering the polar cap. The black contours in Figure 5 are the zonal wind showing the polar vortex starts to build up in April and remains through October between 50˚S and 60˚S. Figures 5c and 5d show the percentage increase compared to the background levels of H2O and sulfur. Even though the majority of volcanic material is outside the polar vortex, we see a ~10% increase of sulfur at 90˚S starting from April and the sulfur mass almost doubles in October. The sulfate aerosol provides extra surface area density (SAD) for heterogeneous reactions affecting ozone chemistry. On the other hand, water increases by about 30% at the edge of the vortex (~ 60˚S) after June. However, due to polar stratospheric cloud formation and dehydration, the positive water anomaly inside the polar vortex does not pass the Student’s t-test at a 90% significance level.
The simulations predict an increase in water vapor and aerosol surface area density (SAD) near the vortex. These changes are expected to impact polar ozone because heterogeneous reactions on polar stratospheric clouds and volcanic particles convert inactive chlorine (ClONO2 and HCl) into photochemically active chlorine31,32. Figures 6a and 6b show that water increases by about 2-3 ppmv and SAD increases by about 2 to 4 µm2/cm3 near the edge of the vortex by the end of September. Water vapor increases between 50 to 10 hPa, while the aerosol increase is between 150 hPa to 30 hPa. This difference in altitude occurs because of aerosol sedimentation during transport. Figure 6c shows the HCl+ClONO2 heterogeneous reaction rate as a function of temperature for different amounts of water vapor. The reaction probability increases about one order of magnitude as we increase water by 3 ppmv above 194 K.