1. Introduction
Large explosive volcanic eruptions are a potential source of uncertainty in the evolution of the climate system on the time scale of a few years, as they cannot be predicted in advance, causing short-term climatic impacts at both local and global scales (Kirtman et al. 2013; Illing et al. 2018). How large eruptions impact Earth’s climate is quite well understood: volcanic sulfur emissions change the atmospheric radiation balance, leading to surface cooling and stratospheric warming. Nevertheless, their regional impacts on various time scales have remained uncertain, as is the precipitation response (Marshall et al. 2022; Timmreck 2012). The volcanic impact on tropical climate variability, in particular, El Niño–Southern Oscillation (ENSO), tropical precipitation, and their combined effect, has been heavily discussed in the recent scientific literature. Several authors investigated the role of eruption parameters (location and season) for the tropical precipitation response (e.g., Zuo et al. 2019; Zhuo et al. 2021; Jacobson et al. 2020) and for the ENSO response to volcanic forcing (Zuo et al. 2018; Pausata et al. 2020). Other studies investigated the influence of the state of the tropical ocean before the eruption on the volcanic impact on ENSO (e.g., Predybaylo et al. 2020; Pausata et al. 2020) and tropical precipitation (Zuo et al. 2021). Further studies addressed how both tropical sea surface temperature (SST) variability and volcanic forcing impact monsoon precipitation (Paik et al. 2020; Singh et al. 2020). However, a clear separation of the individual contributions is complicated. Hence, Paik et al. (2020) suggested that large initial condition ensembles would be a suitable way forward to better isolate the model response from internal variability. Single-model initial-condition large ensembles (i.e., ensemble simulations with the same model and radiative forcing scenario but varying initial conditions) have become a valuable tool in recent years to disentangle forced and internal variability (e.g., Maher et al. 2019; Deser et al. 2020).
Ward et al. (2021) used the historical simulations of the Max Planck Institute Grand Ensemble (MPI-GE; Maher et al. 2019) to study the impact of the largest three volcanic eruptions of the past 50 years (Agung in 1963, El Chichón in 1982, and Pinatubo in 1992) on the ENSO response. They found the volcanically induced displacement of the intertropical convergence zone (ITCZ) to be a key mechanism that drives the ENSO response to volcanic eruptions in their simulations. However, the historical eruptions differ in their strength, geographical location, and season of the eruption, which makes it difficult to draw more generally robust conclusions about the emergence of the volcanic signal from the tropical variability, which is important for climate projections. Here, we use a more idealized large ensemble setup of tropical Northern Hemisphere (NH) summer eruptions of different sulfur emission strengths to investigate whether there is a linear relationship between forcing and response and at which eruption strength the signal might emerge from internal variability. To do so, we use the EVA-ENS (Azoulay et al. 2021), 100-member ensembles of the MPI-ESM-LR with volcanic forcing for idealized equatorial eruptions of different eruption strengths and an additional 100-member ensemble without volcanic forcing. EVA-ENS has already been successfully used to study the NH polar vortex response (Azoulay et al. 2021), stratospheric water vapor changes (Kroll et al. 2021), and regional monsoon impacts (D’Agostino and Timmreck 2022) to volcanic forcing of different emission strengths. The present paper is structured as follows: In section 2, we explain the applied experimental setup and the analysis tools followed by the presentation of main results in section 3. We first focus on the significance of the ensemble mean and the linearity of the signal (sections 3a and 3b) before we will look at spatial anomalies one year after the eruption (section 3c) and historical volcanic forcing (section 3d). Discussions and conclusions are presented in section 4.
2. Method
a. Model and experiments
The EVA-ENS is branched off from the historical simulations of the Max Planck Institute Earth System Model Grand Ensemble (MPI-GE; Maher et al. 2019) in January 1991. The MPI-GE, which is a compilation of 100-member ensembles of several CMIP5 experiments, is based on the low-resolution version of the Max Planck Institute Earth System Model (MPI-ESM1.1-LR), an intermediate version between its CMIP5 version (Giorgetta et al. 2013) and its CMIP6 version (Mauritsen et al. 2019). The MPI-ESM1.1-LR couples the atmospheric general circulation model ECHAM6.3 with a horizontal resolution of T63 (200 km) configuration and 47 vertical levels up to 0.01 hPa (Stevens et al. 2013), with the ocean–sea ice model MPIOM with a nominal ocean resolution of 1.5° and with 64 vertical levels (Jungclaus et al. 2013). The land surface is modeled using JSBACH3.0 (Reick et al. 2013), and HAMOCC5.2 (Ilyina et al. 2013) simulates the ocean biogeochemistry component of the MPI-ESM. The MPI-ESM has been widely used to study the impacts of large volcanic eruptions on climate, including in paleoclimate studies (e.g., Timmreck et al. 2021; Schurer et al. 2019; Fang et al. 2022, 2023) and decadal predictions (e.g., Illing et al. 2018; Timmreck et al. 2016) and in a large ensemble framework (Bittner et al. 2016; Azoulay et al. 2021; Ward et al. 2021; D’Agostino and Timmreck 2022). Multimodel intercomparison studies (Zanchettin et al. 2022; Hermanson et al. 2020) reveal that the MPI-ESM shows good agreement with the other climate models on the global and hemispheric scales concerning the surface climate responses. For example, VolMIP Pinatubo simulations (Zanchettin et al. 2022) show that global and hemispheric-mean temperature anomalies from the HadCRUT5 dataset (Morice et al. 2021) are well within the range of simulated anomalies by the various VolMIP models (including the MPI-ESM) during the post-Pinatubo period and the global peak cooling in late 1992 compares well to the model responses. Simulated precipitation anomalies compare more poorly to observations than near-surface air temperature anomalies.
b. Volcanic forcing/EVA and experimental setup
Volcanic aerosol forcing is prescribed in the MPI-ESM by monthly zonal mean wavelength-dependent optical properties that are interpolated linearly in time for the radiative transfer calculations. The volcanic aerosol forcing, which is used in the historical MPI-GE simulations, is an extended version of the PADS dataset (Stenchikov et al. 1998; Schmidt et al. 2013), which was originally derived from satellite observations for the Pinatubo episode and then extended to the historical period from 1850 to 1999. To study the climate response of volcanic forcing of different strengths, we replace the prescribed volcanic forcing of the June 1991 Mt. Pinatubo eruption in the MPI-GE historical simulations by monthly and zonal mean forcing fields of idealized tropical eruptions of different strengths, compiled with the EVA forcing generator (Toohey et al. 2016). The EVA forcing generator calculates zonal and monthly mean aerosol extinction, single scattering albedo and asymmetry factor, and their spatial and temporal evolution as a function of the volcanic sulfur emission, eruption season, and location and wavelength. Volcanic forcing datasets have been compiled with the EVA tool specifically for idealized tropical June eruptions with sulfur emissions of 2.5, 5, 10, 20, and 40 Tg S. Our upper value of 40 Tg S equals approximately the mean of the stratospheric sulfur emissions of the top six eruptions from the past 2500 years (41.7 Tg S; Toohey and Sigl 2017) including the 1257 Samalas eruption (59.4 Tg S) and the 1815 Tambora eruption (28.1 Tg S). The lower value of 2.5 Tg S is often used in the literature as lower bound for climatic relevant large volcanic eruptions (e.g., Timmreck 2012). The June 1991 Pinatubo eruption sulfur emission is estimated to a range of 5–10 Tg S (Timmreck et al. 2018).
We have performed MPI-ESM 100-member ensemble simulations for these five idealized tropical eruptions as sensitivity experiments to the MPI-GE historical ensemble simulations, while all other historical forcings remained unchanged. Hereinafter, we will denote the individual 100-member ensemble of an idealized tropical NH summer eruption with X Tg S emission as EVAX. In addition, a 100-member ensemble without volcanic forcing has been performed (EVA0; 0 Tg S). Previous studies have shown that a 100-member ensemble is by far sufficient to detect the forced global-mean cooling response after a volcanic eruption (Milinski et al. 2020), as well as regional temperature changes (Pausata et al. 2015). All EVA ensembles were initialized from one of the 100 members of the MPI-GE historical ensemble in January 1991 and were run for at least three years. EVA0, EVA10, EVA20, and EVA40 were run for two additional years (until the end of 1995).
Figure 1a shows the zonal mean stratospheric aerosol optical depth at 0.55 μm (AOD) for the years between 1991 and 1994 exemplary for an idealized tropical June eruption with an emission of 10 Tg S in comparison with the temporal evolution of the Pinatubo volcanic forcing, which is based on satellite observations in the historical MPI-GE simulations. The bulk of the aerosol cloud in the idealized EVA ensemble is centered around the equator and remains there for three years. Transport to the extratropics of both hemispheres occurs in line with a change in the atmospheric circulation from NH summer to winter circulation. In comparison with the MPI-GE Pinatubo forcing dataset, the aerosol in the EVA-ENS dataset moves faster to the NH extratropics and also disappears faster. In EVA, the stratospheric transport is calculated in a very simplistic way among only three boxes (the tropics and two extratropical regions), so, for example, the wintertime stratospheric transport vortex barrier is ignored, as well as the potential influence of different emission heights. In general, in the EVA-ENS dataset the aerosol is more confined to the tropics than to the extratropics. In the Southern Hemisphere (SH) extratropics the MPI-GE dataset has more aerosol in the first months after the eruption relative to EVA-ENS, because the dataset includes not only the Pinatubo eruption but also the historical eruption of Cerro Hudson (45°S) in August 1991. The zonal mean AOD distributions for the other idealized tropical eruptions with different emission strengths resemble each other although with different magnitude (Fig. S1 in the online supplemental material). Despite these subtle differences, our idealized volcanic forcing distributions for a NH summer eruption broadly reflect a Pinatubo-like volcanic eruption with different sulfur emission strengths.
(a) Monthly and zonal mean stratospheric aerosol optical depth at 0.55 μm (AOD) for (top) an idealized equatorial eruption of 10 Tg S (EVA10) and (bottom) the 1991 Pinatubo eruption as described in the historical simulation of the MPI-GE (Maher et al. 2019). (b) Time series of global mean AOD for all five idealized experiments and the MPI-GE Pinatubo forcing. (c) As in (b), but for the relative AOD divided by the amount of sulfur emission (Tg S). The MPI-GE value for Pinatubo is divided by 10 for comparison.
Citation: Journal of Climate 37, 8; 10.1175/JCLI-D-23-0408.1
The temporal evolution of global mean AOD (Fig. 1b) and relative global mean AOD (Fig. 1c) depict, however, some differences in the timing of the peak and the decay time between the different EVA forcing fields. By definition, the AOD at 0.55 μm compiled by the EVA forcing generator depends linearly on the injected sulfur amount below a certain threshold. Modeling studies, however, imply that for very large eruptions, the maximum AOD produced is not a linear function of the injected sulfur anymore as the aerosol particle size distribution is shifted toward larger particle sizes (e.g., Timmreck et al. 2009, 2010). Large aerosol particles could limit the climatic response in a nonlinear way as they are removed more rapidly from the atmosphere than the smaller ones, and they have a reduced light-scattering efficiency. Hence, for eruptions greater in magnitude than the 1815 Tambora eruption, a two-thirds power-law relationship between AOD and injected sulfur amount is implemented in EVA as suggested by Crowley and Unterman (2013) and supported by aerosol general circulation model simulations (Metzner et al. 2014). In our case the two-thirds power-law relationship is applied to the 40 Tg S case, which reduces its radiative forcing relative to the other ones that use the linear relationship. Therefore, the temporal evolution of the global mean AOD for EVA40 differs substantially in comparison with the other experiments with a very slow decay after the peak loading (e-folding time ∼52 months) until summer 1992 followed by a much faster decay (e-folding time ∼15 months) for the following year before slowly approaching the background state. All other forcing fields show a similar behavior as prescribed for the Pinatubo forcing in the MPI-GE with a smooth decay to stratospheric background conditions however with different paces. The slowest decay after the peak loading in NH winter 1991/92 is found for the smaller eruptions with e-folding times of ∼30 months until the end of 1993.
c. Analysis methods
Near-surface temperature and precipitation anomalies for EVA40–EVA2.5 are calculated with respect to the ensemble mean of the volcanically unperturbed experiment (EVA0) for the corresponding year. This anomaly method assumes that the unperturbed climate is characterized by uncorrelated white noise, which adds to the forced response and smears out climate variations on seasonal or longer time scales that may have already been in progress before the perturbation experiments started (Zanchettin et al. 2022). Anomalies for the selected tropical eruptions of the historical MPI-GE are calculated with respect to the ensemble mean of an unperturbed reference period prior to the eruption. The reference periods for the historical MPI-GE eruptions are listed in Table S1 in the online supplemental material and are taken from Bittner et al. (2016).
In the following, we determine the significance of volcanic forced changes by assessing whether the ensemble mean anomalies lie outside of 2 standard deviations as calculated by the spread of individual members of the EVA0 ensemble. We define emergence of the signal in an individual simulation when the anomaly exceeds the range of 2 times the standard deviations as calculated by the spread of individual members of the EVA0 ensemble. The emergence is investigated for all ensemble members and those of specific ENSO states and then compared with the variability of EVA0 members with the same ENSO state. The fraction of individual ensemble members that exceed 2 standard deviations is indicated in percent of the total ensemble members for the composite. Emergence is calculated only if the number of realizations meeting this criterion (i.e., have the respective ENSO state required) is larger than 10, and the results indicate how likely an observed year is to show an anomalous climate.
ENSO variability is characterized by the ocean Niño index (ONI), which is defined as a 3-month running mean over the same region as the Niño-3.4 index (5°N–5°S, 170°–120°W). According to its operational definition used by NOAA, a full-fledged El-Niño or La Niña exists if the ONI exceeds +0.5°C or −0.5°C for at least five consecutive months. For computing the ONI, we calculate SST monthly mean anomalies with respect to the monthly mean climatology of the previous 30 years (1961–90). We also use relative SSTs to ensure that our results are not biased by the volcanic cooling signal. Relative SSTs are defined as the residual signal after removing the mean tropical (20°N–20°S) SST anomalies at each time step (Khodri et al. 2017). To distinguish the ONI based on raw SST from the one based on relative SST, we mark the latter one with a subscript R (ONIR).
3. Results
a. Linearity of the signal
1) Large-scale average response
Near-surface temperature and precipitation ensemble mean anomalies dependent on the sulfur emission strength are shown as global, tropical (30°S–30°N), NH (30°–90°N), and SH (30°–90°S) extratropical means in Fig. 2. As expected, the cooling in the aftermath of the eruption increases with emission strength. Strongest anomalies are found in the NH extratropics in summer 1992 for EVA40 with a peak value of 2.3°C. For all regions a significant cooling response relative to 2 standard deviations of internal variability as measured using the standard deviation across ensemble members for the EVA0 case (see analysis methods) is found for the ensemble mean for eruptions stronger than 20 Tg S. Note that the ensemble mean for EVA0 is significantly different from that of EVAX for a larger forcing range due to our large ensemble size, but here we focus on a comparison with internal climate variability that is indicative of the signal strength expected in observations. At the end of 1994, the ensemble mean cooling of EVA20 is within the internal variability range (2 standard deviations of monthly temperature for EVA0), while significant anomalies are still found in EVA40 in 1995 in all regions except the SH extratropics. Extratropical temperature anomalies are modulated by the seasonal cycle, with stronger cooling occurring in late autumn. In the tropics, the seasonal cycle is only weakly pronounced in the near-surface temperature anomalies, whereas it is more evident in the precipitation anomalies. Local maxima in the tropical precipitation decrease appear in spring and in autumn with a maximum reduction of 8% in early spring 1992 for EVA40.
Time series of monthly mean (left) near-surface temperature (°C) and (right center) precipitation (%) anomalies for the first 4.5 years after the eruption: (a),(c) globally, (e),(g) for the tropics, (i),(k) the NH extratropics, and (m),(o) the SH extratropics. The gray shaded areas indicate the range of internal variability as measured using the standard deviation across ensemble members for the EVA0 case. (b),(d),(f),(h),(j),(l),(n),(p) As in the left and right-center columns, but anomalies are linearly scaled by 1/X, where X is the volcanic sulfur emission (Tg S) for the respective experiment.
Citation: Journal of Climate 37, 8; 10.1175/JCLI-D-23-0408.1
Significant ensemble-mean precipitation anomalies relative to a single realization of internal variability occur only for volcanic emission strengths larger or equal to 20 Tg S in the global and tropical mean, while for the NH and SH extratropics the ensemble-mean precipitation anomalies for all experiments are in the range of internal variability. To determine how unusual the climate anomalies are likely to be following an eruption we can assess the number of individual ensemble members that lie outside the bounds of what can normally be expected for a typical year. Table 1 shows that for large eruptions large-scale temperature is almost certain to be outside 2 standard deviations and would be considered very unusual, as is global and tropical precipitation. The likelihood of an unusual year decreases with the eruption size, but with more than expected by chance (5%) even for relatively small eruptions (EVA5 for global and tropical averages).
Number of realizations in the different EVA-ENS ensembles for which annual mean near-surface temperature and precipitation anomalies exceed 2 standard deviations of internal climate variability for annual mean changes in 1992.
To test whether near-surface temperature and precipitation response change linearly with the eruption strength, we scale the anomalies by 1/X, where X is the amount of sulfur (Tg S) injected in the respective experiment. We find that global and large-scale hemispheric mean temperature and precipitation anomalies are in general scalable with the sulfur emission above a certain emission strength ≥10 Tg S (Fig. 2). The scaled global and hemispheric mean ensemble anomalies for the eruptions with sulfur emission smaller than 10 Tg S are dominated by internal variability, in particular for precipitation. However, the time series of global, tropical, and extratropical mean scaled temperature anomalies for EVA2.5 and EVA5 look similar to the other experiments. Interestingly, the scaled tropical mean near-surface temperature anomalies for EVA40 show less relative cooling when compared with EVA20 and EVA10 while the tropical mean scaled precipitation anomalies of the three experiments look similar. This behavior is related to ENSO dynamics and will be discussed in section 3b).
2) Regional scale
The time series of the ensemble-mean near-surface temperature and precipitation anomalies (Fig. 2) indicate a linear response on global and hemispheric scale. To test whether this linear response also holds at regional scale, we investigate the patterns of scaled seasonal and ensemble mean near-surface temperature and precipitation anomalies for EVA5–EVA40 for the year 1992. The scaling allows us to extrapolate a spatial fingerprint of the volcanic forcing on temperature and precipitation and makes the spatial distribution of the anomalies more comparable among the experiments.
It also leads to a more prominent appearance of such anomalies not attributable to the linear aspects of the forcing itself, but instead results from differences in dynamical responses that behave nonlinearly as a function of the forcing. Overall, we find that the patterns of scaled seasonal and ensemble mean near-surface temperature and precipitation anomalies resemble each other (Figs. 3 and 4). The near-surface temperature anomaly patterns (Fig. 3) are similar among the experiments with strong cooling over the NH continents throughout the year with seasonal varying regional minima. The strongest cooling is visible in boreal summer over the northern midlatitudes.
Seasonal mean and ensemble mean near-surface temperature anomaly (°C) for 1992. Mean anomalies and their variance are linearly scaled by 20/X where X is the amount of sulfur (Tg S) injected in the respective experiment that would render linear responses equal. Stippled anomalies are statistically not significant from zero at the 95% confidence level according to a t test prior to scaling.
Citation: Journal of Climate 37, 8; 10.1175/JCLI-D-23-0408.1
As in Fig. 3, but for precipitation.
Citation: Journal of Climate 37, 8; 10.1175/JCLI-D-23-0408.1
In boreal winter, significant warming anomalies are found at the 95% level over Eurasia in all experiments. This phenomenon is known as NH winter warming pattern and is induced by atmospheric circulation changes (e.g., Wunderlich and Mitchell 2017; DallaSanta et al. 2019; Azoulay et al. 2021; DallaSanta and Polvani 2022). It occurs in NH post-eruption winters as a dynamic response to the stratospheric volcanic aerosol layer and displays a high spatial variability (Shindell et al. 2004), which is also evident in our experiments. Interestingly, the warming appears also in spring, although with smaller values and is only significant in EVA40 and EVA20 (Fig. 3). In all experiments, an El Niño–type warm anomaly appears off the coast of South America from boreal spring to autumn, although this anomaly is not significant in most cases. Almost no temperature change is found south of 50°S. Most deviations to the other experiments are found for the EVA5 experiment. The reduction in solar radiation and surface cooling leads to pronounced circulation changes, which modulate the global precipitation response to volcanic eruptions at regional scales, with some dry regions showing a tendency to become wetter while on average tropical wet regions become drier (Fig. 4). Ensemble-mean precipitation anomalies are less significant than the corresponding temperature anomalies. Precipitation hardly changes significantly in the EVA5 ensemble mean and even in EVA40 precipitation anomalies are not significant over large parts of the oceans. Nevertheless, several significant spatial anomaly patterns are found across the EVA experiments (e.g., a drying of the Maritime Continent throughout the year). As discussed in D’Agostino and Timmreck (2022), these eruptions also lead to significant and substantial monsoon changes, and some regions (such as northern and southern Africa, South America, and South Asia) are much more sensitive to this kind of forcing than others.
3) Pattern correlations
To quantify the linearity of the anomaly patterns for the different forcing strengths, we show correlations among seasonal mean patterns of the ensemble mean near-surface temperature and precipitation anomalies of EVA5, EVA10, and EVA40 with the corresponding seasonal and ensemble-mean EVA20 anomalies (Fig. 5). We choose EVA20 as our reference as it represents the middle of those experiments that show a clear global and large hemispheric mean linear signal. The very high pattern correlation for both near-surface temperature and precipitation anomalies with values (≥0.8) in 1992 highlights the consistent patterns over eruption strengths and shows that linearity exists not only in large-scale averages but also in global spatial patterns. The season with the highest correlation, however, varies with the region: the highest correlation occurs predominantly in the extratropics earlier in 1992 than in the tropics. Interestingly, global and tropical precipitation pattern correlation coefficients show a distinct minimum between autumn 1993 and summer 1994 (Figs. 5e,f), which is most prominent in EVA40 and also seen in its tropical temperature correlation (Fig. 5b). The correlation coefficients for EVA5 are weaker, mostly below 0.5 due to the smaller signal-to-noise ratio relative to EVA10 and EVA40 with correlation coefficients larger than 0.6, except for precipitation in the NH and SH extratropics in 1994 and 1995 (Figs. 5g,h).
Pattern correlations of (a)–(d) seasonal mean near-surface temperature anomalies and (e)–(h) seasonal mean precipitation anomalies of EVA5, EVA10, and EVA40 with respect to seasonal mean EVA20 anomalies for selected regions: (left) global, (left center) tropical, (right center) NH extratropics, and (right) SH extratropics. The white areas in the plots indicate a lack of simulations (and not a correlation of 0).
Citation: Journal of Climate 37, 8; 10.1175/JCLI-D-23-0408.1
The prominent surface warming in NH winter over the Eurasian continent (Fig. 3) is the cause of the very high correlation for EVA40 and EVA10 in NH winter 1991/92. The relatively low correlation in EVA5 can be explained by the smaller signal-to-noise ratio in particular in NH winter with a high internal variability and the fact that the winter warming pattern is shifted farther south and east relative to the other experiments and that Alaska warms instead of cools (Fig. 3). The alternating pattern correlation coefficients in the tropics might be related to the high tropical hydroclimate variability, in particular ENSO. In the next section, we investigate how different ENSO states impact our results.
b. Role of ENSO
Large tropical eruptions not only induce widespread cooling but also modulate tropical variability (i.e., ENSO). ENSO itself impacts the tropical precipitation distribution by influencing atmospheric circulation patterns and thereby altering the pattern of dry and wet regions (e.g., Rasmusson and Carpenter 1982; Rasmusson and Wallace 1983). In the aftermath of a large volcanic eruption, the dynamical ENSO signature is often masked by tropical radiative cooling, and hence in many cases the temperature anomaly pattern shows only a weak ENSO-like warming pattern in the tropical east Pacific. The ONI index for the EVA-ENS indicates for tropical SST a nonlinear dependence on the sulfur emission strength, with EVA10 being the experiment with the warmest ENSO signal in the mean, which is even larger (≥+0.5) for a couple of months in 1992. In contrast, ONI in EVA40 stays within the range of weak to neutral ENSO state.
To isolate the intrinsic/dynamical ENSO signature from the volcanically induced surface cooling we followed Khodri et al. (2017), by using instead the relative SST anomalies, rather than raw SST anomalies for the computation of the ONI. The ONIR (Fig. 6b) shows an El Niño–like response to volcanic forcing in 1992 in our EVA-ENS ensemble, in agreement with most climate model results (e.g., Paik et al. 2020; McGregor et al. 2020; Pausata et al. 2023) while recent observational studies do not support the model-based hypothesis that tropical eruptions drive an ENSO response (Dee et al. 2020; Zhu et al. 2022).
Temporal evolution of (a) the standardized 3-month running mean ONI index and (b) the standardized 3-month running mean ONIR index. The standardized ONIR index is calculated for relative SSTs according to Khodri et al. (2017). Dashed horizontal lines indicate the thresholds for weak La Niña (−0.5) and El Niño (0.5) events. A black vertical solid line marks the date of the eruption. Solid color lines indicate the ensemble mean of each experiment, and the shaded regions show the standard error.
Citation: Journal of Climate 37, 8; 10.1175/JCLI-D-23-0408.1
The simulated response is a function of the sulfur emission strength and it is strongest and longest lasting in EVA40. The strong positive ENSO response in EVA40 counteracts the radiative cooling leading to relatively weak tropical cooling in comparison with the other experiments (Fig. 2f). The maximum ONIR index is reached in autumn 1992 and then slowly declines toward neutral and La Niña–like conditions in 1994 and 1995.
To study in more detail the influence of ENSO on the spatial distribution of volcanically induced changes, we focus on the NH summer in 1992, which has the largest near-surface temperature and precipitation anomalies (see Figs. 2–4) and also a very high ONIR index (Fig. 6b). The percentage values of all ENSO-like composites for all experiments in July–August (JJA) 1992 are listed in Table 2. For EVA40 and EVA20, the two experiments with the strongest volcanic forcing, around three-quarters of the ensemble members show a positive ONIR, with just one member in EVA40, and six members in EVA20 depicting a negative ONIR. For the unperturbed case 40% of the ensemble members are in the neutral phase with the positive and negative ENSO-like phases more or less equally distributed. The transition into a preference for a positive ENSO-like regime starts with EVA10, where almost all ensemble members are either in the neutral or in the positive ENSO-like phase.
Percentage of ENSO-like states in the different EVA-ENS ensembles in JJA 1992 separated by their ONIR index with ENSO+ ≥ 0.5 and ENSO− ≤ −0.5.
We show the seasonal and ensemble mean precipitation anomalies and near-surface temperature anomalies for all states and the corresponding ENSO composites for EVA40 to EVA5 based on their relative ONIR index in NH summer 1992 (Fig. 7, along with Fig. S2 in the online supplemental material). More than 70% of all ensemble members are in a positive ENSO-like phase in JJA 1992 in EVA40 and EVA20. Consequently, their ensemble mean precipitation anomalies of all states reflect the spatial anomaly distribution for positive ENSO-like events. On the other hand, for EVA5 the spatial precipitation anomaly pattern of the ensemble mean of all states reflects the anomaly pattern of the neutral state, with very weak anomalies, as the ensemble members with positive and negative ENSO-like states offset each other. This highlights the importance of the state of the tropical ocean for regional precipitation anomalies as they might alter the spatial pattern substantially.
Seasonal mean precipitation anomalies in NH summer 1992 for EVA5–EVA40 for the ensemble mean of all states and the ENSO composites, separated by their relative ONIR index with ENSO+ > 0.5 and ENSO− < −0.5. The number of ensemble members for each composite is indicated in the upper-right corner of each panel.
Citation: Journal of Climate 37, 8; 10.1175/JCLI-D-23-0408.1
In addition, the ensemble mean anomaly pattern reflects a different spatial distribution than individual realizations and might therefore not be representative for regional volcanically induced precipitation anomalies, in particular for eruptions of similar or smaller size than the Pinatubo eruption. This is also true for the near-surface temperature anomalies (Fig. S2 in the online supplemental material) where the ensemble mean of all states does not reflect the positive temperature anomalies over the Indian subcontinent and the Sahel region, which are seen in more than 50% or 34% of all realizations in EVA10 or EVA5, respectively.
c. Emergence of the signal
The distinct seasonal mean precipitation and near-surface temperature anomaly patterns (Fig. 7, along with Fig. S2 in the online supplemental material) for the different ENSO composites in the large ensemble context offer the opportunity for a better disentangling of the volcanic signal from internal variability and show how the appearance of a significant signal may change between observed cases if there is no consistent ENSO response.
Figures 8 and 9 show the emergence of the volcanically induced signal in the JJA 1992 seasonal mean near-surface temperature and precipitation anomalies for all members and for members with specific ENSO states (e.g., neutral or positive). The emergence is defined as a signal larger or smaller than 2 standard deviations of the corresponding composites of the EVA0 ensemble as calculated by the spread of individual members of the corresponding composites of the EVA0 ensemble. The fraction of individual ensemble members that exceed 2 standard deviations is also indicated in percent of the total ensemble members for the corresponding composite.
Emergence of seasonal mean near-surface anomalies in NH summer 1992 in the EVA-ENS experiments as percentage of the ensemble members exceeding the threshold of 2 standard deviations of seasonal EVA0 temperatures with the respective ENSO states in JJA 1992. Blue color indicates emergence of cooling; red color indicates emergence of warming. The fraction of individual ensemble members who exceed 2 standard deviations is indicated in percent of the total ensemble number (upper-right corner). Calculations are done only if the total number of ensemble members in each composite is larger than 10. The pink lines mark the regions where the ensemble mean exceeds 2 standard deviations.
Citation: Journal of Climate 37, 8; 10.1175/JCLI-D-23-0408.1
Emergence of seasonal mean precipitation anomalies in NH summer 1992 in the EVA-ENS experiments as percentage of the ensemble members exceeding the threshold of 2 standard deviations of seasonal EVA0 temperatures with the respective ENSO states in JJA 1992. Green color indicates the emergence of more precipitation (moistening), and brown shows the emergence of less precipitation (drying). The fraction of individual ensemble members that exceed 2 standard deviations is indicated in percent of the total ensemble number (upper-right corner). Calculations are done only if the total number of ensemble members in each composite is larger than 10. The purple lines mark the regions where the ensemble mean exceeds 2 standard deviations.
Citation: Journal of Climate 37, 8; 10.1175/JCLI-D-23-0408.1
Significant near-surface temperature and precipitation anomalies emerge only for EVA40, EVA20, and EVA10 although for the latter experiment the number of significant locations is drastically reduced. No significant anomalies that exceed the 95% range can be found for EVA5. As most ensemble members for EVA40 and EVA20 have an ONIR larger than 1 the ensemble mean of all states reflects the warm ENSO-like composites, while for the neutral case only in a few locations a significant signal occurs, in particular for precipitation. We find a few regions where no significance is found for the ensemble mean states but where already 20%–40% of all realizations are outside the range of internal variability. If we lower the threshold to 1 standard deviation, more regions over land are visible now and also occur in EVA5 where the ensemble mean and/or individual ensemble members are outside of the 68% confidence interval (Figs. S3 and S4 in the online supplemental material).
In general, the number of regions where the signal exceeds the threshold increases when subdivided into ENSO composites. In summary, significant volcanic cooling appears generally for EVA40 and EVA20 on a large hemispheric scale. In contrast, the emergence of the precipitation response is more localized and mainly confined to the tropical and subtropical ocean.
d. Historical volcanic forcing
An important aspect of our experimental design is that we varied the strength of the stratospheric sulfur emission but kept the season and geographical location the same for all experiments. The high degree of scalability of the surface climate response will most likely be altered if we consider more realistic volcanic forcing distributions, as has been derived for the large tropical eruptions of the historical period. To see how robust the results of the idealized experiments are in comparison with tropical historical eruptions, we calculate correlation coefficients for the seasonal mean near-surface temperature and precipitation anomalies of large tropical eruptions of the historical MPI-GE with the corresponding anomaly pattern of the idealized EVA20 ensemble. The historical eruptions we consider are Krakatau in August 1883, Agung in March 1963, El Chichón in April 1982, and Pinatubo in June 1991, which all show a different spatial volcanic forcing distribution than our idealized ensemble (Fig. 1; see also Figs. S1 and S5 in the online supplemental material). The zonal stratospheric AOD distributions for the Krakatau and the Pinatubo eruptions with a tropical maximum over the first months and meridional transport to both hemispheres resembles the applied idealized tropical forcing of the EVA-ENS, in contrast to the other two eruptions. Although Agung (6°S) and El Chichón (15°N) are both located in the tropics, the forcing patterns of their recent large eruptions are asymmetric with almost all aerosol located in the NH for El Chichón in 1982 and in the SH for the Agung 1963 eruption.
The seasonal mean pattern correlation values reflect the difference in the forcing distribution (Fig. 10). The strongest correlation coefficients are found for the more equatorial symmetric eruptions (Krakatau and Pinatubo) while smaller correlation values are depicted for Agung and El Chichón. In general, the seasonal mean pattern correlation decreases when different volcanic forcing distributions are considered. The effect is stronger for the extratropical seasonal averages and most evident for NH extratropical precipitation anomalies. Correlation coefficients higher than 0.8 are only seen for Agung and El Chichón in rare cases. For the two more symmetrical tropical eruptions Pinatubo and Krakatau high correlation coefficients are found in the second NH autumn/winter but the correlations strongly decrease 1.5–2 years after the eruption depending on the region and the variable considered. In the second year after the eruption, the reference period also becomes relevant for the pattern correlation. We have calculated the near-surface temperature and precipitation anomalies for Pinatubo in two ways; one comparable to EVA20 with respect to EVA0 (E-Pin) and the other similar to the other three historical eruptions with respect to a reference period prior to the eruption (GE-Pin; Table S1 in the online supplemental material). While for both cases the seasonal mean pattern correlation coefficients are almost identical in the first two years after the eruption, E-Pin correlation coefficients are clearly higher than in GE-Pin for the following years, suggesting that low-frequency variability plays a role.
Pattern correlations of (a)–(d) seasonal mean near-surface temperature and (e)–(h) seasonal mean precipitation anomalies simulated after large historical tropical eruptions in the MPI-GE: Krakatau in August 1883 (Kra), Agung in March 1963 (Agu), El Chichón in April 1982 (EIC), and Pinatubo in June 1991 (Pin) with respect to the seasonal mean EVA20 anomalies for selected regions: (left) global, (left center) tropical, (right center) NH extratropics, and (right) SH extratropics. For the Pinatubo eruption, we calculated the anomalies with respect to EVA0 (E-Pin) and analogous to the other historical eruptions to a reference period prior to the eruption (GE-Pin). The reference periods for the historical MPI-GE eruptions are listed in Table S1 in the online supplemental material. To ensure comparability all eruptions are aligned to 1991 as the eruption year.
Citation: Journal of Climate 37, 8; 10.1175/JCLI-D-23-0408.1
4. Discussion and conclusions
In this work, we investigate how large tropical eruptions shape tropical hydroclimate and to what extent the impacts are a function of the emission strength in a suite of sensitivity experiments within the MPI-ESM Grand Ensemble framework. Our results reveal a high degree of linearity and scalability for both near-surface temperature and precipitation anomalies in the aftermath of a tropical volcanic eruption with sulfur emissions between 10 and 40 Tg S. In an accompanying paper (Freychet et al. 2023) we have looked into the volcanic impact on extreme events finding a large change in the probability of extreme temperature and precipitation events. That paper also found, to first order, that results for extreme indices can also likely be linearly scaled (provided that the volcanic forcing distribution is similar). However, when considering uncertainties represented by ensemble spreads, it is clear that lower emission cases are poorly detectable. Previous model studies (e.g., Timmreck et al. 2009, 2010) have shown that the climate response is not linear in the case of extremely large volcanic eruptions (∼100 Tg S) as the radiative forcing is not a linear function of the sulfur emission anymore. Although the temporal development of the AOD for the EVA40 differs from the other experiments, the deviations are still very small (Fig. 1). For larger eruptions as considered here another parameter than the sulfur emission strength e.g., AOD might be better useful to investigate the linearity of the climate response to increasingly strong volcanic eruptions.
In summary, the large ensembles of idealized tropical eruptions show the following:
-
Global and large-scale hemispheric mean near-surface temperature and precipitation anomalies are scalable for sulfur emission between 10 and 40 Tg S. This does not hold for the tropical near-surface temperature mean for a sulfur emission strength of 40 Tg S due to the nonlinear development of ENSO.
-
Seasonal and ensemble mean pattern correlation of near-surface temperature and precipitation anomalies are highly correlated in particular for larger emission strengths in the tropics. However, the pattern correlation is critically dependent on the reference period, whose importance increases with time as the forcing weakens. This suggests a potential role for low-frequency variability even in an ensemble mean of 100 members.
-
Tropical hydroclimate anomalies are modulated by ENSO. The tendency for a warm ENSO-like state increases with eruption strength for the first post volcanic year. Although we did not investigate here mechanisms for tropical volcanic eruption triggering a prevalent El Niño state, our results would further support recent findings by Pausata et al. (2023), in which warm ENSO-like state are more likely after a very strong tropical or NH eruption.
-
A stratospheric emission of 20 Tg S, which is about 2 times the upper estimate of the 1991 Pinatubo eruption, is identified as a threshold where the signal exceeds the range of internal variability on global and hemispheric scales.
-
Emergence of the volcanic signal occurs more often for smaller eruption strengths when looking at ENSO composites rather than the mean of all states. The emergence of cooling appears on a hemispheric scale, while the precipitation response is more localized and mainly confined to the tropical and subtropical ocean.
Overall, our results suggest that there is a potential of predictability of near-surface temperature and precipitation anomaly patterns on the seasonal scale if the volcanic forcing pattern is similar to the equatorial symmetric one of the 1991 Pinatubo eruption. However, the relevance of our results for real world climate predictions in the case of a potential future volcanic eruptions should not be overstated. Although hindcasts with and without volcanic aerosols (e.g., Timmreck et al. 2016; Ménégoz et al. 2018; Hermanson et al. 2020) have demonstrated that in general the multiyear to decadal prediction skill has increased if volcanic radiative forcing is included, this is not the case for all regions, in particular over the tropical Pacific the multiyear forecast skill is degraded when volcanic forcing is taken into account (Wu et al. 2023; Bilbao et al. 2023). The tropical Pacific is a very crucial and sensitive region with large implications on global to regional hydroclimate changes. Therefore, the reliability of climate models for simulating tropical hydroclimate must be considered when discussing simulated volcanic impacts on it. Furthermore, our results show that the signal-to-noise ratio is relatively low and volcanically induced anomalies exceed the range of internal variability only for eruptions with sulfur emission ≥20 Tg S with some spatial variability. These eruptions are very rare in time; only 10 of them occurred in the last 2500 years (Toohey and Sigl 2017). However, despite low signal-to-noise ratio, the risk of very cold anomalies is higher in the volcanically forced simulations than in the ones without forcing (see Schurer et al. 2019).
If a future eruption happens the largest uncertainties stem from the compilation/simulation of its radiative forcing based on the estimated sulfur emission from satellite observations. An aerosol model intercomparison study for the Pinatubo eruption (Quaglia et al. 2023) reveals for example not only a large intermodel spread but also large differences between the model and satellite observations related to atmospheric transport and precursor chemistry.
The model response to external forcing can also be very different on regional scales, which is in particular the case for ENSO (e.g., Beobide-Arsuaga et al. 2021; Zanchettin et al. 2022) and monsoons (e.g., Paik et al. 2020). It would therefore be desirable to perform a similar analysis in a multimodel ensemble framework to identify general surface climate response patterns in the aftermath of a large tropical eruption. This could be of major importance not only for seasonal to decadal forecasts in the case of a potential future eruption but also to better understand the response of the coupled atmosphere–ocean system to external forcing.
Acknowledgments.
The authors are grateful to Alon Azoulay and Hauke Schmidt for providing the original EVA-ENS ensemble data. Author Claudia Timmreck is funded by the German National funding agency (DFG) research unit FOR 2820: Revisiting the Volcanic Impact on Atmosphere and Climate—Preparations for the Next Big Volcanic Eruption (VolImpact, Project Number: 398006378). Author Dirk Olonscheck received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement 820829 (CONSTRAIN), and author Shih-Wei Fang received funding from the German Federal Ministry of Education and Research (BMBF), research programme “ROMIC-II,ISOVIC” (FKZ: 01LG1909B). The Edinburgh authors were funded by the GloSAT project (NE/S015698/1). Computations and analysis were performed on the computer of the Deutsches Klima Rechenzentrum (DKRZ) using resources granted by its Scientific Steering Committee (WLA) under project identifier bb1093.
Data availability statement.
Primary data and scripts used in the analysis and other supplemental material that may be useful in reproducing this work are archived by the Max Planck Institute for Meteorology (https://hdl.handle.net/21.11116/0000-000D-4B1F-E).
REFERENCES
Azoulay, A., H. Schmidt, and C. Timmreck, 2021: The Arctic polar vortex response to volcanic forcing of different strengths. J. Geophys Res. Atmos., 126, e2020JD034450, https://doi.org/10.1029/2020JD034450.
Beobide-Arsuaga, G., T. Bayr, A. Reintges, and M. Latif, 2021: Uncertainty of ENSO-amplitude projections in CMIP5 and CMIP6 models. Climate Dyn., 56, 3875–3888, https://doi.org/10.1007/s00382-021-05673-4.
Bilbao, R., and Coauthors, 2023: Impact of volcanic eruptions on CMIP6 decadal predictions: A multi-model analysis. Earth Syst. Dyn., https://doi.org/10.5194/esd-2023-36, in press.
Bittner, M., H. Schmidt, C. Timmreck, and F. Sienz, 2016: Using a large ensemble of simulations to assess the Northern Hemisphere stratospheric dynamical response to tropical volcanic eruptions and its uncertainty. Geophys. Res. Lett., 43, 9324–9332, https://doi.org/10.1002/2016GL070587.
Crowley, T. J., and M. B. Unterman, 2013: Technical details concerning development of a 1200 yr proxy index for global volcanism. Earth Syst. Sci. Data, 5, 187–197, https://doi.org/10.5194/essd-5-187-2013.
D’Agostino, R., and C. Timmreck, 2022: Sensitivity of regional monsoons to idealised equatorial volcanic eruption of different sulfur emission strengths. Environ. Res. Lett., 17, 054001, https://doi.org/10.1088/1748-9326/ac62af.
DallaSanta, K., and L. M. Polvani, 2022: Volcanic stratospheric injections up to 160 Tg(S) yield a Eurasian winter warming indistinguishable from internal variability. Atmos. Chem. Phys., 22, 8843–8862, https://doi.org/10.5194/acp-22-8843-2022.
DallaSanta, K., E. P. Gerber, and M. Toohey, 2019: The circulation response to volcanic eruptions: The key roles of stratospheric warming and eddy interactions. J. Climate, 32, 1101–1120, https://doi.org/10.1175/JCLI-D-18-0099.1.
Dee, S. G., K. M. Cobb, J. Emile-Geay, T. R. Ault, R. L. Edwards, H. Cheng, and C. D. Charles, 2020: No consistent ENSO response to volcanic forcing over the last millennium. Science, 367, 1477–1481, https://doi.org/10.1126/science.aax2000.
Deser, C., and Coauthors, 2020: Insights from Earth system model initial-condition large ensembles and future prospects. Nat. Climate Change, 10, 277–286, https://doi.org/10.1038/s41558-020-0731-2.
Fang, S.-W., C. Timmreck, J. Jungclaus, K. Krüger, and H. Schmidt, 2022: On the additivity of climate responses to the volcanic and solar forcing in the early 19th century. Earth Syst. Dyn., 13, 1535–1555, https://doi.org/10.5194/esd-13-1535-2022.
Fang, S.-W., M. Sigl, M. Toohey, J. Jungclaus, D. Zanchettin, and C. Timmreck, 2023: The role of small to moderate volcanic eruptions in the early 19th century climate. Geophys. Res. Lett., 50, e2023GL105307, https://doi.org/10.1029/2023GL105307.
Freychet, N., A. P. Schurer, A. P. Ballinger, L. Suarez-Gutierrez, and C. Timmreck, 2023: Assessing the impact of very large volcanic eruptions on the risk of extreme climate events. Environ. Res. Climate, 2, 035015, https://doi.org/10.1088/2752-5295/acee9f.
Giorgetta, M. A., and Coauthors, 2013: Climate and carbon cycle changes from 1850 to 2100 in MPI-ESM simulations for the Coupled Model Intercomparison Project phase 5. J. Adv. Model. Earth Syst., 5, 572–597, https://doi.org/10.1002/jame.20038.
Hermanson, L., and Coauthors, 2020: Robust multiyear climate impacts of volcanic eruptions in decadal prediction systems. J. Geophys. Res. Atmos., 125, e2019JD031739, https://doi.org/10.1029/2019JD031739.
Illing, S., C. Kadow, H. Pohlmann, and C. Timmreck, 2018: Assessing the impact of a future volcanic eruption on decadal predictions. Earth Syst. Dyn., 9, 701–715, https://doi.org/10.5194/esd-9-701-2018.
Ilyina, T., K. D. Six, J. Segschneider, E. Maier-Reimer, H. Li, and I. Núñez-Riboni, 2013: Global ocean biogeochemistry model HAMOCC: Model architecture and performance as component of the MPI-Earth system model in different CMIP5 experimental realizations. J. Adv. Model. Earth Syst., 5, 287–315, https://doi.org/10.1029/2012MS000178.
Jacobson, T. W. P., W. Yang, G. A. Vecchi, and L. W. Horowitz, 2020: Impact of volcanic aerosol hemispheric symmetry on Sahel rainfall. Climate Dyn., 55, 1733–1758, https://doi.org/10.1007/s00382-020-05347-7.
Jungclaus, J. H., and Coauthors, 2013: Characteristics of the ocean simulations in the Max Planck Institute Ocean Model (MPIOM) the ocean component of the MPI-Earth system model. J. Adv. Model. Earth Syst., 5, 422–446, https://doi.org/10.1002/jame.20023.
Khodri, M., and Coauthors, 2017: Tropical explosive volcanic eruptions can trigger El Niño by cooling tropical Africa. Nat. Commun., 8, 778, https://doi.org/10.1038/s41467-017-00755-6.
Kirtman, B., and Coauthors, 2013: Near-term climate change: Projections and predictability. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 953–1028.
Kroll, C. A., S. Dacie, A. Azoulay, H. Schmidt, and C. Timmreck, 2021: The impact of volcanic eruptions of different magnitude on stratospheric water vapor in the tropics. Atmos. Chem. Phys., 21, 6565–6591, https://doi.org/10.5194/acp-21-6565-2021.
Maher, N., and Coauthors, 2019: The Max Planck Institute Grand Ensemble: Enabling the exploration of climate system variability. J. Adv. Model. Earth Syst., 11, 2050–2069, https://doi.org/10.1029/2019MS001639.
Marshall, L. R., E. C. Maters, A. Schmidt, C. Timmreck, A. Robock, and M. Toohey, 2022: Volcanic effects on climate: Recent advances and future avenues. Bull. Volcanol., 84, 54, https://doi.org/10.1007/s00445-022-01559-3.
Mauritsen, T., and Coauthors, 2019: Developments in the MPI-M Earth System Model version 1.2 (MPI-ESM1.2) and its response to increasing CO2. J. Adv. Model. Earth Syst., 11, 998–1038, https://doi.org/10.1029/2018MS001400.
McGregor, S., M. Khodri, N. Maher, M. Ohba, F. S. Pausata, and S. Stevenson, 2020: The effect of strong volcanic eruptions on ENSO. El Niño Southern Oscillation in a Changing Climate, Geophys. Monogr., Vol. 253, Amer. Geophys. Union, 267–287, https://doi.org/10.1002/9781119548164.ch12.
Ménégoz, M., R. Bilbao, O. Bellprat, V. Guemas, and F. J. Doblas-Reyes, 2018: Forecasting the climate response to volcanic eruptions: Prediction skill related to stratospheric aerosol forcing. Environ. Res. Lett., 13, 064022, https://doi.org/10.1088/1748-9326/aac4db.
Metzner, D., S. Kutterolf, M. Toohey, C. Timmreck, U. Niemeier, A. Freundt, and K. Krüger, 2014: Radiative forcing and climate impact resulting from SO2 injections based on a 200,000-year record of Plinian eruptions along the Central American volcanic arc. Int. J. Earth Sci., 103, 2063–2079, https://doi.org/10.1007/s00531-012-0814-z.
Milinski, S., N. Maher, and D. Olonscheck, 2020: How large does a large ensemble need to be? Earth Syst. Dyn., 11, 885–901, https://doi.org/10.5194/esd-11-885-2020.
Morice, C. P., and Coauthors, 2021: An updated assessment of near-surface temperature change from 1850: The HadCRUT5 data set. J. Geophys. Res. Atmos., 126, e2019JD032361, https://doi.org/10.1029/2019JD032361.
Paik, S., S.-K. Min, C. E. Iles, E. M. Fischer, and A. P. Schurer, 2020: Volcanic-induced global monsoon drying modulated by diverse El Niño responses. Sci. Adv., 6, eaba1212, https://doi.org/10.1126/sciadv.aba1212.
Pausata, F. S. R., A. Grini, R. Caballero, A. Hannachi, and Ø. Seland, 2015: High-latitude volcanic eruptions in the Norwegian Earth System Model: The effect of different initial conditions and of the ensemble size. Tellus, 67B, 26728, https://doi.org/10.3402/tellusb.v67.26728.
Pausata, F. S. R., D. Zanchettin, C. Karamperidou, R. Caballero, and D. S. Battisti, 2020: ITCZ shift and extratropical teleconnections drive ENSO response to volcanic eruptions. Sci. Adv., 6, eaaz5006, https://doi.org/10.1126/sciadv.aaz5006.
Pausata, F. S. R., Y. Zhao, D. Zanchettin, R. Caballero, and D. S. Battisti, 2023: Revisiting the mechanisms of ENSO response to tropical volcanic eruptions. Geophys. Res. Lett., 50, e2022GL102183, https://doi.org/10.1029/2022GL102183.
Predybaylo, E., G. Stenchikov, A. T. Wittenberg, and S. Osipov, 2020: El Niño/Southern Oscillation response to low-latitude volcanic eruptions depends on ocean pre-conditions and eruption timing. Commun. Earth Environ., 1, 12, https://doi.org/10.1038/s43247-020-0013-y.
Quaglia, I., and Coauthors, 2023: Interactive stratospheric aerosol models’ response to different amounts and altitudes of SO2 injection during the 1991 Pinatubo eruption. Atmos. Chem. Phys., 23, 921–948, https://doi.org/10.5194/acp-23-921-2023.
Rasmusson, E. M., and T. H. Carpenter, 1982: Variations in tropical sea surface temperature and surface wind fields associated with the Southern Oscillation/El Niño. Mon. Wea. Rev., 110, 354–384, https://doi.org/10.1175/1520-0493(1982)110<0354:VITSST>2.0.CO;2.
Rasmusson, E. M., and J. M. Wallace, 1983: Meteorological aspects of the El Niño/Southern Oscillation. Science, 222, 1195–1202, https://doi.org/10.1126/science.222.4629.1195.
Reick, C. H., T. Raddatz, V. Brovkin, and V. Gayler, 2013: Representation of natural and anthropogenic land cover change in MPI-ESM. J. Adv. Model Earth Syst., 5, 459–482, https://doi.org/10.1002/jame.20022.
Schmidt, H., and Coauthors, 2013: Response of the middle atmosphere to anthropogenic and natural forcings in the CMIP5 simulations with the Max Planck Institute Earth System Model. J. Adv. Model. Earth Syst., 5, 98–116, https://doi.org/10.1002/jame.20014.
Schurer, A. P., G. C. Hegerl, J. Luterbacher, S. Brönnimann, T. Cowan, S. F. B. Tett, D. Zanchettin, and C. Timmreck, 2019: Disentangling the causes of the 1816 European year without a summer. Environ. Res. Lett., 14, 094019, https://doi.org/10.1088/1748-9326/ab3a10.
Shindell, D. T., G. A. Schmidt, M. E. Mann, and G. Faluvegi, 2004: Dynamic winter climate response to large tropical volcanic eruptions since 1600. J. Geophys. Res., 109, D05104, https://doi.org/10.1029/2003JD004151.
Singh, M., and Coauthors, 2020: Fingerprint of volcanic forcing on the ENSO–Indian monsoon coupling. Sci. Adv., 6, eaba8164, https://doi.org/10.1126/sciadv.aba8164.
Stenchikov, G. L., I. Kirchner, A. Robock, H.-F. Graf, J. C. Antuña, R. G. Grainger, A. Lambert, and L. Thomason, 1998: Radiative forcing from the 1991 Mount Pinatubo volcanic eruption. J. Geophys. Res., 103, 13 837–13 857, https://doi.org/10.1029/98JD00693.
Stevens, B., and Coauthors, 2013: Atmospheric component of the MPI-M Earth System Model: ECHAM6. J. Adv. Model. Earth Syst., 5, 146–172, https://doi.org/10.1002/jame.20015.
Timmreck, C., 2012: Modeling the climatic effects of large explosive volcanic eruptions. Wiley Interdiscip. Rev.: Climate Change, 3, 545–564, https://doi.org/10.1002/wcc.192.
Timmreck, C., S. J. Lorenz, T. J. Crowley, S. Kinne, T. J. Raddatz, M. A. Thomas, and J. H. Jungclaus, 2009: Limited temperature response to the very large AD 1258 volcanic eruption. Geophys. Res. Lett., 36, L21708, https://doi.org/10.1029/2009GL040083.
Timmreck, C., H.-F. Graf, S. J. Lorenz, U. Niemeier, D. Zanchettin, D. Matei, J. H. Jungclaus, and T. J. Crowley, 2010: Aerosol size confines climate response to volcanic super-eruptions. Geophys. Res. Lett., 37, L24705, https://doi.org/10.1029/2010GL045464.
Timmreck, C., H. Pohlmann, S. Illing, and C. Kadow, 2016: The impact of stratospheric volcanic aerosol on decadal-scale climate predictions. Geophys. Res. Lett., 43, 834–842, https://doi.org/10.1002/2015GL067431.
Timmreck, C., and Coauthors, 2018: The Interactive Stratospheric Aerosol Model Intercomparison Project (ISA-MIP): Motivation and experimental design. Geosci. Model Dev., 11, 2581–2608, https://doi.org/10.5194/gmd-11-2581-2018.
Timmreck, C., M. Toohey, D. Zanchettin, S. Brönnimann, E. Lundstad, and R. Wilson, 2021: The unidentified eruption of 1809: A climatic cold case. Climate Past, 17, 1455–1482, https://doi.org/10.5194/cp-17-1455-2021.
Toohey, M., and M. Sigl, 2017: Volcanic stratospheric sulfur injections and aerosol optical depth from 500 BCE to 1900 CE. Earth Syst. Sci. Data, 9, 809–831, https://doi.org/10.5194/essd-9-809-2017.
Toohey, M., B. Stevens, H. Schmidt, and C. Timmreck, 2016: Easy volcanic aerosol (EVA v1.0): An idealized forcing generator for climate simulations. Geosci. Model Dev., 9, 4049–4070, https://doi.org/10.5194/gmd-9-4049-2016.
Ward, B., F. S. R. Pausata, and N. Maher, 2021: The sensitivity of the El Niño–Southern Oscillation to volcanic aerosol spatial distribution in the MPI grand ensemble. Earth Syst. Dyn., 12, 975–996, https://doi.org/10.5194/esd-12-975-2021.
Wu, X., S. G. Yeager, C. Deser, N. Rosenbloom, and G. A. Meehl, 2023: Volcanic forcing degrades multiyear-to-decadal prediction skill in the tropical Pacific. Sci. Adv., 9, eadd9364, https://doi.org/10.1126/sciadv.add9364.
Wunderlich, F., and D. M. Mitchell, 2017: Revisiting the observed surface climate response to large volcanic eruptions. Atmos. Chem. Phys., 17, 485–499, https://doi.org/10.5194/acp-17-485-2017.
Zanchettin, D., and Coauthors, 2022: Effects of forcing differences and initial conditions on inter-model agreement in the VolMIP volc-pinatubo-full experiment. Geosci. Model Dev., 15, 2265–2292, https://doi.org/10.5194/gmd-15-2265-2022.
Zhu, F., J. Emile-Geay, K. J. Anchukaitis, G. J. Hakim, A. T. Wittenberg, M. S. Morales, M. Toohey, and J. King, 2022: A re-appraisal of the ENSO response to volcanism with paleoclimate data assimilation. Nat. Commun., 13, 747, https://doi.org/10.1038/s41467-022-28210-1.
Zhuo, Z., I. Kirchner, S. Pfahl, and U. Cubasch, 2021: Climate impact of volcanic eruptions: The sensitivity to eruption season and latitude in MPI-ESM ensemble experiments. Atmos. Chem. Phys., 21, 13 425–13 442, https://doi.org/10.5194/acp-21-13425-2021.
Zuo, M., W. Man, T. Zhou, and Z. Guo, 2018: Different impacts of northern, tropical, and southern volcanic eruptions on the tropical Pacific SST in the last millennium. J. Climate, 31, 6729–6744, https://doi.org/10.1175/JCLI-D-17-0571.1.
Zuo, M., T. Zhou, and W. Man, 2019: Hydroclimate responses over global monsoon regions following volcanic eruptions at different latitudes. J. Climate, 32, 4367–4385, https://doi.org/10.1175/JCLI-D-18-0707.1.
Zuo, M., W. Man, and T. Zhou, 2021: Dependence of global monsoon response to volcanic eruptions on the background oceanic states. J. Climate, 34, 8273–8289, https://doi.org/10.1175/JCLI-D-20-0891.1.
