The Los Chocoyos (LCY, 14.6∘ N, 91.2∘ W) supereruption (Kutterolf et al., 2016) of M=8 (calculated by Pyle, 2013), dated to ∼84 kyr before the present (Rose et al., 1999), is known to be one of the largest volcanic eruptions of the past 100 000 years (Drexler et al., 1980). The eruption formed the current stage of the large Atitlán caldera in present-day Guatemala. LCY released more than ∼1100 km3 of tephra, and the eruption is used as a widespread key stratigraphic marker during that time (Cisneros et al., 2020; Kutterolf et al., 2016). The ash layers can be found in marine deposits from offshore Ecuador to Florida over an area of more than 107 km2 (Kutterolf et al., 2016). Hardly anything is known about the climate impacts of this eruption from proxy records, but LCY emitted large amounts of climate-relevant and environmentally relevant gases including sulfur, chlorine, and bromine compounds (Krüger et al., 2015; Kutterolf et al., 2015, 2016; Metzner et al., 2014).

The sulfur gases emitted by volcanoes have a strong direct climate impact through the formation of sulfuric acid aerosols, which block incoming sunlight and cool the surface (Robock, 2000). Halogen compounds, such as chlorine and bromine, contribute to catalytic ozone depletion in the stratosphere (Brasseur and Solomon, 2005; Solomon, 1999). There is well-documented petrological evidence that volcanic supereruptions have emitted environmentally significant amounts of chlorine and bromine (Cadoux et al., 2015, 2018; Krüger et al., 2015; Kutterolf et al., 2013, 2015; Vidal et al., 2016). Furthermore, recent atmospheric observations revealed that even relatively small volcanic eruptions can inject significant amounts of halogen compounds in the stratosphere (for review and overview discussions, see von Glasow et al., 2009; Krüger et al., 2015; WMO, 2018). This means that a sulfur- and halogen-rich eruption is expected to cool the Earth's surface and potentially damage the stratospheric ozone layer, with further impacts on the surface environment through the change in the atmosphere's transparency to harmful ultraviolet (UV), particularly shortwave UVB, radiation (i.e., Brenna et al., 2019). Potential volcanic iodine injections to the stratosphere (Schönhardt et al., 2017) would have even a larger ozone depletion potential than chlorine and bromine (Solomon et al., 1994). However, no direct iodine erupted mass measurements are available for the LCY eruption.

Metzner et al. (2014) used the general circulation model (GCM) MAECHAM5-HAM coupled with a modal aerosol microphysics scheme, together with the CLIMBER-2 (an Earth system model of intermediate complexity) to study the climate impact of LCY. Based on the former published mass estimate of 687 Mt SO2 (343.5 Mt S), they simulated a peak cooling of 3.1 ∘C from their LCY eruption scenario. Toohey et al. (2011, 2013) investigated the atmospheric physical processes of LCY, using 700 Mt SO2 (350 Mt S) injections with the model MAECHAM5-HAM, and revealed the important effects of the different seasons of that eruption on the aerosol evolution, transport, and deposition of sulfate to the ice cores and then compared them to weaker eruptions strengths.

Even though there is little literature about LCY, the climate impact of supereruptions has been discussed in the scientific literature since at least the early 1990s. Early studies argued that the eruption of Toba (73 kyr ago) could have initiated a glacial period (Rampino and Self, 1992, 1993; Zielinski et al., 1996). In addition, there is evidence that human populations went through a genetic bottleneck (i.e., most of the population died) at approximately the same time as the eruption of Toba (Ambrose, 1998; Haslam and Petraglia, 2010; Williams et al., 2009), but this is now considered unlikely (Timmreck et al., 2010, 2012).

A thorough investigation of the climatic and environmental impacts of extremely large to supervolcanic eruptions (M: 7–8) requires the use of a global climate model or, ideally, an Earth system model (ESM). There are several studies published with different model complexities that mostly focus on the Toba eruption and its sulfur impact on the atmosphere and climate.

Bekki et al. (1996) used a two-dimensional chemistry transport model with internally generated atmospheric circulation to study the Toba eruption impact on the atmosphere. Their simulations indicate that Toba could have caused a long-lasting atmospheric response due to the interactions between chemistry and aerosol microphysics.

Later, Jones et al. (2005) used a coupled atmosphere–ocean general circulation model (AOGCM) to study the Toba eruption impact on climate. In this study, they forced the model by linearly scaling the observed aerosol optical depth (AOD) from the 1991 eruption of Pinatubo by a factor of 100, resulting in peak cooling of ∼11 K in the second post-eruption year, followed by an initial recovery that took a decade. Surface cooling larger than 1 K persisted for more than 20 years. The volcanic aerosol forcing only lasted 5 years, so the response needed to be maintained by feedback in other components of the Earth system through, e.g., sea ice–ocean feedback sustaining the short atmospheric forcing to longer (decadal to centennial) timescales (Miller et al., 2012; Stenchikov et al., 2009; Zhong et al., 2011).

In a similar study, Robock et al. (2009) used three different AOGCMs to study the Toba eruption's effect on climate and ozone. In their study, both the linear scaling AOD approach and directly injecting sulfur into a model with an interactive bulk aerosol module were utilized. The resulting magnitude and length of the cooling was similar to what was published by Jones et al. (2005) across the different model versions and forcing methods. The scenarios representing a 100-times Pinatubo forcing resulted in ∼12 K peak cooling with multidecadal recovery times. In a 300-times Pinatubo scenario simulation including atmospheric chemistry effects, Robock et al. (2009) found slightly stronger (∼1 K) and much more long-lasting surface cooling (length of >10 K cooling extended by ∼5 years) compared to the similar forcing scenario without chemistry, due to the depletion of atmospheric hydroxyl (OH) that limits the speed of aerosol formation, which leads to a longer-lasting forcing. In addition, they simulated an increase in global column ozone, which attributed to the reduction of reactive hydrogen oxides in the atmosphere.

Another model study of Toba, presented in Timmreck et al. (2010, 2012) and Zanchettin et al. (2014), simulated a smaller climate impact with peak cooling of 3.5 K that lasted up to 10 years from the injected sulfur compared to the analogous simulations in Robock et al. (2009). A key difference between Timmreck et al. (2010) and previous AOGCM simulations of Toba is the inclusion of online aerosol microphysics in a modal aerosol scheme and the OH limitation mechanism, which leads to larger aerosol sizes, lower peak AOD (∼4) values, and thus lower climate impacts in Timmreck et al. (2010). The inclusion of interactive OH chemistry in the formation of aerosol is important because the availability of OH controls the speed of SO2 oxidation into sulfate (Bekki, 1995).

Concentrating on the atmospheric processes, English et al. (2013) used a sectional aerosol microphysics model coupled with a chemistry climate model with prescribed sea surface temperatures (SSTs) to study the aerosols and atmospheric impacts of the Toba eruption. In their model setup, by neglecting aerosol radiative effects, they simulate even lower peak AOD values (∼2.6) due to the fact that their sectional aerosol module creates larger aerosols compared to Timmreck et al. (2010).

In the climate modeling literature on the Toba supereruption, there is a progression from larger to smaller climate (and environmental) impacts as model complexity develops over time. In the more recent climate models one key reason seems to be that climate effects are self-limiting for larger eruptions due to an increase in aerosol growth, which reduces peak AOD (English et al., 2013; Pinto et al., 1989; Timmreck et al., 2010). In addition, the role of atmospheric chemistry and OH limitation on sulfuric acid aerosols is continuously under discussion in the literature (Bekki, 1995; Mills et al., 2017; Niemeier et al., 2019; Robock et al., 2009; Timmreck et al., 2003).

Investigating the effects of the Toba eruption on the Earth system, such as hydrology and terrestrial net primary production, reveals a substantial reduction of precipitation globally, which leads to the reduction of tree cover, increase in grass cover, and decreased net primary production (NPP) but with large regional and intermodel variability (Robock et al., 2009; Timmreck et al., 2010, 2012). While tropical deciduous trees and broadleaf evergreen trees virtually disappear in the simulations of Robock et al. (2009), Timmreck et al. (2012) find much more muted impacts on tree cover, particularly in the tropics. Timmreck et al. (2012) even simulate an increase in NPP in the tropical rain forest regions of South America and Africa.

Recent studies proposed a sea ice–ocean mechanism, which prolongs the volcanic-induced short, abrupt surface cooling and sea ice increase due to longer timescales (decadal), with the ocean sustaining the signal by buffering and transporting the cooling poleward (Miller et al., 2012; Zhong et al., 2011). In addition, Zanchettin et al. (2014) simulated an interhemispheric response to different volcanic forcings with Pinatubo to Toba strength by using Arctic sea ice expansion for all cases and only using an Antarctic sea ice expansion and subsequent contraction for the supereruptions.

We are not aware of studies of the supersized eruption effects on the El Niño–Southern Oscillation (ENSO), whereas the effects of large to very large (M: 5–6) volcanic eruptions have been widely discussed in the literature. There is an ongoing debate (Stevenson et al., 2017) about whether tropical volcanic eruptions can lead to La Niña-like responses in the same year (Anchukaitis et al., 2010; Li et al., 2013) or the (5) following years (Zanchettin et al., 2012) or to El Niño-like responses in up to the 2 following years (e.g., Adams et al., 2003; Handler, 1984; Khodri et al., 2017; Ohba et al., 2013; Predybaylo et al., 2017; Stevenson et al., 2016). Discussions include the significance and mechanism of the results as well as the eruption characteristics, latitude, season, and strength.

Simulations of halogen-rich (and sulfur-rich) eruptions show that these can have serious, long-lasting impacts on the ozone layer, with implications for the surface environment through the increase in surface ultraviolet radiation. In a 2D chemical transport model (CTM) study of the late Bronze Age eruption of Santorini (M=7), Cadoux et al. (2015) simulated the decadal ozone depletion, mainly in the Northern Hemisphere, with a peak depletion of 20 %–90 % depending on the degassing budget. In another study using a 2D CTM (Klobas et al., 2017), (hypothetical) volcanic halogens were included with the sulfur injection of Pinatubo, showing the ozone depletion of 20 % lasting a few years under different future emission scenarios. Using the Community Earth System Model version 1 (CESM1) coupled with the Whole Atmosphere Community Climate Model (WACCM) and a comprehensive coupled chemistry–climate model (CCM) with prescribed volcanic aerosols and SSTs, Brenna et al. (2019) simulated an average Central American volcanic arc (CAVA) eruption with M=6.4. They found ozone depletion up to 20 % globally and lasting up to 10 years, which was most pronounced over the Northern Hemisphere (NH) and dropped to below present-day ozone hole conditions over Antarctica and the tropics. Consequently, surface UV radiation increased by >80 % over the 2 years, with potential impacts on human health, agriculture, and marine life.

We are not aware of supereruption studies taking the combined effect of sulfur and halogen injections in a fully coupled aerosol–chemistry–climate model ESM into account.

In this study we use the recently released CESM2 coupled with WACCM6 as the atmospheric component, which allows us to newly investigate the coupling and the feedback between volcanic aerosols, chemistry, radiation, climate, and the Earth system after a sulfur- and halogen-rich supervolcanic eruption. The primary goal of this paper is to investigate the combined effect of the sulfur- and halogen-rich LCY supereruption on climate and environment. In particular, we study the impacts of LCY by varying the eruption composition and size on the following: (i) atmospheric burden of volcanic gases and aerosols; (ii) ozone and UV; (iii) climate and environment; and (iv) ENSO. Finally, we compare our findings with other model studies before we give a summary and conclusion. In a forthcoming paper, we will investigate the impacts on stratospheric circulation in the tropics.

Our simulations reveal very large climate impacts from the LCY sulfur- and halogen-rich supereruption, which are larger than other recent simulation studies of supereruptions. In Fig. 7 we show scatterplots comparing our simulations of LCY to other simulations of supervolcanic eruptions using sulfur-only injections (English et al., 2013; Jones et al., 2005; Metzner et al., 2014; Robock et al., 2009; Timmreck et al., 2010).

Compared to other model studies with interactive aerosols of volcanic eruptions of M>7, our simulations show very large maximum AODs and thus maximum surface climate impacts for a given sulfur injection (Fig. 7a, b). The largest climate cooling for a supereruption is achieved when using linearly scaled AOD values based on Pinatubo (Jones et al., 2005; Robock et al., 2009), but this approach is simplified since there are several feedbacks (i.e., self-limiting, scattering, and removal of aerosols) that make the relationship between sulfur injection, aerosols, radiative forcing, and climate highly nonlinear (i.e., Bekki, 1995; Metzner et al., 2014; Pinto et al., 1989).

Limiting our comparison to model studies that use sulfur injection to generate self-consistent AOD estimates, we see that our model experiments show longer aerosol lifetime, larger radiative impacts, and larger surface cooling per injected sulfur mass to the stratosphere than those studies (English et al., 2013; Metzner et al., 2014; Timmreck et al., 2010). A model intercomparison for the Tambora eruption revealed that version 5 of WACCM also has the longest aerosol lifetime among compared models (Marshall et al., 2018). The differences (Fig. 7) could be caused by different model top levels, aerosol microphysics (bulk vs. modal vs. sectional modules), radiation, advection, and depositions schemes (see discussions by English et al., 2013; Marshall et al., 2018), as well as atmospheric chemistry (OH, ozone, H2O) and climate/ESM model differences (coupling, resolution, clouds). Our comparison is limited by the fact that the simulations were not part of a coordinated model intercomparison yet; thus, the model experiments are all different when related to eruption strength, date, location, and injection altitude. Volcanic aerosol climate model intercomparisons are in progress now (see Timmreck et al., 2018; Zanchettin et al., 2016) and should include extremely large to supersized eruptions, where the model spread is even larger (Fig. 7) but observational evidence is lacking.

Even though the halogen injection efficiency for a supereruption like LCY is highly uncertain, we expect that the effects of injected halogens would similarly be qualitatively independent of the magnitude of the injection as our model results for 10 % and 1 % halogen injections reveal. Injecting additional volcanic halogens into the stratosphere leads to ozone depletion (this study; Brenna et al., 2019), and the interaction with the OH availability impacts the aerosol formation which leads to smaller maximum AOD and hence weaker surface cooling. Including the volcanic release of halogens as well as sulfur should be part of future model intercomparisons focusing on volcanic impacts on climate and ozone.

Figure 7c shows a clear, nearly linear relationship between peak surface cooling and peak AOD. This is consistent with previous studies (Hansen et al., 1980; Metzner et al., 2014; Timmreck et al., 2012).

When atmospheric temperatures drop after a volcanic eruption, changes in the energy balance of the climate system lead to decreased global mean precipitation (Iles et al., 2013; Robock and Liu, 1994). The global mean precipitation response to the supervolcanic eruptions follows a nearly linear relation with temperature (Fig. 7d) because larger cooling leads to larger negative precipitation anomalies through a weakening of the global hydrological cycle, since lower temperatures leads to lower relative humidity in the troposphere. Our modeled southward shift of the ITCZ towards the southern tropics is accompanied by increased precipitation across northern Africa and the Middle East, which is also partly simulated in Robock et al. (2009) and Timmreck et al. (2010, 2012), but the area experiencing wetting is larger in our simulations. The wetting of northern Africa and the Middle East in our simulations leads to a strong increase in NPP in this area, which likely leads to a greening of the Sahara. Timmreck et al. (2012), with only half the volcanic forcing that we simulate, show NPP maps (vegetation impacts are simulated using an offline vegetation model), and here there is very little change over the first 3 post-eruption years throughout this region. While we cannot compare our NPP field directly to the changes in the vegetation types presented in Robock et al. (2009), we note that they simulate an increase in grass cover throughout northern Africa and parts of the Middle East where there is very low vegetation cover in their control run, which would imply an increase in NPP in this region as well.

We simulate pronounced El Niño conditions in our LCY supereruption during the first 3 post-eruption years, which is superposed on a strong surface cooling signal. El Niño conditions may be favored as discussed in more detail by Emile-Geay et al. (2008) due to the uniform solar dimming that leads to a thermostat mechanism (Clement et al., 1996) initiating air–sea interaction in the equatorial Pacific. Our simulations of the sulfur- and halogen-rich LCY supereruption in the northern tropics (14.6∘ N) during January add another puzzle piece to the ongoing discussion of volcanic eruption impacts on ENSO (see Introduction). A coordinated model intercomparison study would help to shed more light on the different model response and mechanism.

Atmospheric circulation changes at high latitudes (i.e., stratospheric polar vortices, annular modes) are expected to be significant as was investigated by Toohey et al. (2011, 2013). A follow-up LCY paper will investigate the impacts on the stratospheric circulation in the tropics, the QBO, in more detail.

Using a fully coupled ESM with interactive aerosols and atmospheric chemistry is currently the best possible way to simulate the impacts of supervolcanoes on the Earth system. Our model setup takes into account the interactive coupling between most of the components of the Earth system, including ocean, sea ice, biogeochemistry, land surface and vegetation interactions. In addition, the inclusion of interactive aerosols and atmospheric chemistry is crucial to correctly simulate the feedback between the chemical composition of the atmosphere, aerosols and radiation. That said, there is still considerable uncertainty in the impacts of volcanic sulfur injections, particularly in the conversion of SO2 into sulfate aerosol, aerosol size and the lifetime of the radiative perturbation. The uncertainty in the Earth system's reaction to a given volcanic aerosol radiative forcing seems to be smaller (cf. Fig. 7c). Two recent studies suggest that LCY might have been the eruption of the last 100 kyr with the largest climate impact. Cisneros et al. (2020) report a new and higher sulfur-mass estimate for LCY. Meanwhile, Toba is estimated to be less sulfur-rich than previously assumed (Chesner and Luhr, 2010). In order to compare these two supereruptions and petrological estimates, other archives such as ice core records would be needed. However, no tephra has been identified in Greenland and Antarctica ice cores for both eruptions up to now (Abbott et al., 2012; Svensson et al., 2013). This model study together with a new examination of the LCY eruption date and a higher mass estimate (Cisneros et al., 2020) will hopefully stimulate upcoming studies finding corresponding paleoproxies in ice cores, climate, and archeological archives with high temporal resolution and precision.

We simulated the Los Chocoyos (LCY) eruption of Atitlán under 1850 preindustrial conditions with 523 Mt sulfur, 1200 Mt chlorine, and 2 Mt bromine emissions and assumed 10 % stratospheric injection efficiency for the halogens. The model results may have been similar for LCY 84 000 years ago, as we did not set up the simulations with observed initial conditions and there are uncertainties in the volcanic emissions. As expected, if there are large halogen emissions, the climate and environment response is different than if the volcano only emits sulfur into the stratosphere. Overall, we evaluate our model results to show a lower estimate of the possible climate and environment response given the likely low estimates for our petrologically derived volcanic emissions.

Our comprehensive aerosol chemistry Earth system model (ESM) study shows that a sulfur- and halogen-rich tropical supereruption like LCY has massive impacts on global climate and the environment that last at least 20 to 30 years.

In the model, enhanced volcanic sulfate burdens and aerosol optical depth (AOD) persists for 5 years, while the halogens stay elevated for ∼14 years. Under preindustrial conditions, the eruption leads to a global collapse of the ozone layer (80 % decrease) with global mean values of 50 DU and increasing surface UVB by 550 % globally over the first 5 years after the eruption. (In high latitudes the increase is >1000 %.) The ozone layer takes 15 years to recover. The simulated volcanic eruption, at 14.6∘ N in January, shows an asymmetric-hemispheric response with enhanced AOD, ozone, UV, and climate signals over the Northern Hemisphere (NH).

The eruption cools the global climate, which lasts more than 30 years with the peak AOD of >6 that leads to surface cooling >6 K, and precipitation and terrestrial net primary production decrease up to 30 % in the first 2 years. Locally, a wetting (>100 %) and strong increase in NPP (>400 %) over northern Africa that is simulated in the first 5 years is also related to a southward shift of the ITCZ to the southern tropics. The global sea ice area almost doubles, and the long-lasting surface cooling is sustained by an increase in Arctic ice area, followed by a decrease of poleward ocean heat transport at 60∘ N from year 3. Both changes last up to 20 years. The ocean responds with pronounced El Niño conditions in the first 3 years by maximizing at 12∘ S, which is coherent with the southward shift of the ITCZ.

In contrast, simulations of LCY including sulfur, but neglecting halogens, reveal a larger sulfate burden and maximum AOD (∼8); hence, there is a larger radiative forcing with more pronounced surface climate cooling (>7 K) and reduced precipitation (25 %) globally, even though spatial patterns of changes are similar to the simulations including volcanic sulfur and halogens. The environmental impacts reveal the opposite signal with a short-lived increase in column ozone of 100 DU (>30 %) and decrease in UV (>60 %) that lasts up to 3 years.

LCY is one of the largest volcanic eruptions over the past 100 kyr, and we predict large impacts on the biosphere and thus any human populations at that time. Finding paleoproxies that show the impact of LCY on climate and the environment should be possible, given the large long-lasting impact from our ESM simulations, but it will require high (subdecadal) temporal-resolution archives that use the eruption as a time marker.