1. Introduction
Explosive strong tropical volcanic eruptions (SVEs) such as Mount Pinatubo (1991) significantly affect the climate by injecting sulfur-rich gases into the stratosphere. The resulting increased concentration of stratospheric sulfate aerosols acts to scatter and absorb incoming solar radiation, which leads to a temporary reduction in the amount of solar radiation reaching the surface. The relatively short persistence of volcanic aerosols in the stratosphere (2–3 yr) identifies SVE forcing as a narrow-peak-type perturbation of the climate system (Robock 2000). However, volcanically induced cooling at the surface could penetrate deeper into the ocean where it may persist for several years (e.g., Church et al. 2005; Gleckler et al. 2006). The effects of SVEs are not, therefore, limited to direct changes in the radiative energy balance, but may also alter the atmospheric and oceanic circulation and modulate interannual to decadal climate variations (e.g., Mann et al. 2005; McGregor et al. 2010; Shiogama et al. 2010; Zanchettin et al. 2012).
The El Niño–Southern Oscillation (ENSO), which consists of a quasiperiodic (3–7-yr time scale) warming (El Niño) and cooling (La Niña) of the tropical central and eastern Pacific (CEP) Ocean, forms the main pattern of the earth's interannual climate variability. The prediction of ENSO is of practical and scientific interest as it has large environmental and societal impacts. Volcanically induced cooling in the tropics alters the surface climate, which then rapidly reduces the predictability of ENSO. As ENSO affects the global climate, it is of considerable importance to determine how its phases are altered by the impact of volcanic forcing. At the end of the twentieth century, several articles discussed (using limited observed episodes) whether a SVE can alter ENSO (Handler 1984) or whether it does not affect it (Nicholls 1988; Self et al. 1997; Robock 2000). The largest eruptions of the last 50 years (Agung in 1963, El Chichón in 1982, and Pinatubo in 1991) have occurred in conjunction with the warm phase of ENSO, and it is therefore possible that contributions of a SVE have been hidden. However, recent analysis of much longer-term paleoclimate records derived from multiple proxy data suggests that the radiative effects of a tropical SVE can lead to an El Niño–like state and increase the probability of El Niño occurrences (e.g., Adams et al. 2003; McGregor et al. 2010). These studies document the fact that volcanic forcing exerts a relatively weak, but discernible, influence on ENSO.
The process by which volcanic forcing influences ENSO is not well understood. The relationship between ENSO and explosive volcanism has previously been studied (Mann et al. 2005; Emile-Geay et al. 2008) using an intermediate air–sea coupled model (Zebiak and Cane 1987). Mann et al. and Emile-Geay et al. demonstrate a warming in the CEP in response to a uniform reduction of model surface heat fluxes, which can be explained by the dynamical thermostat hypothesis (Clement et al. 1996). In this hypothesis, the mean oceanic advection makes it harder for radiative forcing to change sea surface temperature (SST) in the eastern equatorial Pacific. Given a uniform reduction of incoming surface solar radiation, the SST therefore cools faster in the west, initially reducing the climatological zonal SST gradient. The resultant positive zonal gradient of SST initiates El Niño and this effect is subsequently amplified by the Bjerknes feedback (Bjerknes 1969). However, some studies using coupled general circulation models (CGCMs) show that feedback from tropical cooling can initially lead to a strong negative phase of ENSO (e.g., McGregor and Timmermann 2011; Zanchettin et al. 2012). Details of the short-term dynamical responses of the climate to SVEs still remain unclear. Because the air–sea coupled system in the Pacific includes a strong nonlinearity (Ohba and Ueda 2009), it is possible that the ENSO response to volcanic forcing may be different depending on its phase.
The aim of this study was to examine the ENSO response to tropical SVEs and its dependency on the ENSO phase using a CGCM. To further the understanding of SVE effects on ENSO, we performed climate simulations with and without volcanic forcing. This paper is organized in the following manner. Section 2 presents a brief description of our CGCM and experimental design. Section 3 examines the volcanic impact on ENSO in the CGCM when initialized at different ENSO phases. Section 4 investigates the physical causes of the El Niño–like response in the CGCM, and section 5 presents the discussion and a summary of the conclusions.
2. Model and experimental design
a. MIROC5i
In the present study, we use an interim version of the Model for Interdisciplinary Research on Climate (MIROC) (Watanabe et al. 2010), named MIROC5i, which contains several minor differences from the official fifth version (MIROC5, see Watanabe et al. 2011 for details). This version will also be employed in the Fifth Assessment Report (AR5) of the Intergovernmental Panel on Climate Change (IPCC). The design of MIROC5 is based on MIROC3 (Hasumi and Emori 2004), which employs a global spectral dynamical core and implements a standard physics package for the atmosphere. The ocean and sea ice models within MIROC5 are taken from the updated Center for Climate System Research ocean component model (Hasumi 2006). In addition, a land model (which includes a river module) is coupled to the system. In MIROC5, many of the schemes have been replaced with recent ones. A preindustrial control experiment showed a remarkable improvement in the ENSO amplitude, spatiotemporal structure, and their asymmetry between El Niño and La Niña in comparison to that recorded by MIROC3 (Watanabe et al. 2010; Ohba and Watanabe 2012).
The standard resolution for MIROC5 is T85L40 for the atmosphere. Because the model's mean state and variability are not seriously altered when the horizontal resolution of the atmosphere is reduced, we decided to use a coarser resolution of T42L40 for MIROC5i in this study. Our decision was based solely on the increased computational burden of the new physics package in MIROC5. To implement volcanic forcing, we simply incorporated a variation in the stratospheric aerosol optical thickness at 0.55 μm in the radiation code. This implementation enables the CGCM to reproduce the observed climate cooling after a significant volcanic eruption (Yokohata et al. 2005; Shiogama et al. 2006).
b. Volcanic forcing experiments
We initially performed a 200-yr control simulation with no-volcanic activity (hereafter referred to as Ctrl). Regarding Ctrl as a “benchmark,” three sets of idealized experiments (or so-called perfect model studies) were conducted with MIROC5i to examine the effects of volcanic forcing on the tropical Pacific and to help understand the mechanisms underlying the MIROC5 response. Each of these experiments used the same model configuration and the only difference between the simulations was the phase of ENSO used. Three different ENSO conditions were extracted from Ctrl (neutral, peak of El Niño, and La Niña). Figure 1 shows a scatter diagram of the simulated December–February (DJF) Niño-3.4 index with that recorded in the following year. The Niño-3.4 index is defined as the average SST anomaly in the 5°S–5°N, 170°–120°W region. We see that the air–sea coupled system over the Pacific remains in a weak La Niña state for up to two years, while El Niño tends to turn rapidly into La Niña after the mature phase (e.g., Ohba and Ueda 2009; Ohba et al. 2010). The well-known nonlinear relationship between ENSO and the SST in the following year was well reproduced by the model, and moderate El Niño events persisted into the following year (Fig. 1). Because of the computational burden, we selected five initial conditions (1 July) during no-ENSO (i.e., neutral) and strong El Niño (La Niña) events that showed peculiar transition (duration) features in Ctrl. The selected cases were plotted using green, red, and blue dots, which represent ENSO-neutral, El Niño, and La Niña years, respectively.
The scatterplot of the simulated Niño-3.4 index (°C) during DJF0/+1 against the following year, DJF+1/2. The selected ENSO-neutral, El Niño, and La Niña cases for the identical-twin experiments are denoted by green, red, and blue dots, respectively.
Citation: Journal of Climate 26, 14; 10.1175/JCLI-D-12-00471.1
The volcanic forcing in each of these ensembles was at a level where volcanic stratospheric aerosol concentrations increase six months after initialization and reach their peak in the boreal winter (Fig. 2a). We make the assumption, for computational simplicity, that the eruption occurs in this season. The spatiotemporal structure of the forcing was derived from the lag regression against the tropical–subtropical mean optical thickness (30°S–30°N) in the stratosphere given by Sato et al. (1993). In this temporal profile, the volcanically induced tropical stratospheric aerosols persisted for approximately 2–3 yr. The idealized forcing provides suitable test beds for examining the potential of climate forecasting. The amplitude of volcanic forcing is about 1.5 times larger than that of the forcing of the 1991 eruption of Mount Pinatubo (Sato et al. 1993), which resulted in a clear-sky surface radiative forcing of approximately −12 W m−2 around the SVE peak that is comparable to the magnitude of the 1258 eruption, undoubtedly the largest in the past millennium and by some accounts, the third largest of the Holocene (Stothers 2000). Months in the SVE developing years are denoted by a superscript “0,” and those in the succeeding years are represented by “+1” and “+2.” Ensemble forecasts using five members over 30 months were performed from 1 July0 until the end of December+2 with (SVE run) and without (noSVE run) SVE forcing, respectively. In addition to the CGCM integration, similar SVE experiments using the same atmospheric general circulation model (AGCM) with prescribed climatological SST were additionally conducted to evaluate the effect of air–sea coupling.
(a) Global-mean optical depth forcing used in the SVE run and (b) its meridional distribution.
Citation: Journal of Climate 26, 14; 10.1175/JCLI-D-12-00471.1
3. Impact of volcanic eruptions on ENSO
a. Neutral year experiment
In this section, we present the results of the normal year experiments to illustrate the impact of a SVE on ENSO. Figure 3a shows the composited ensemble mean of the Niño-3.4 index and the global mean surface shortwave radiation anomaly derived from ENSO-neutral year cases. Corresponding with the SVE forcing (Fig. 1a), we find a reduction of incoming surface shortwave radiation at about −6.5 W m−2 around the SVE peak. In agreement with the observed findings from paleoclimate records (Adams et al. 2003), the ensemble simulations in response to the volcanic shortwave forcing show an El Niño–like warming that lags one year behind the peak SVE forcing. The difference between the ensemble mean in SVE and Ctrl exceeds 1°C around the end of year (+1). The SVE run also exhibited widespread variability among the ensemble members (shaded orange) around the El Niño peak, implying that the trajectories of the simulated ENSO in the CGCM are affected by chaotic or stochastic forcing, irrespective of forcing by SVEs.
(a) Evolution of the composited Niño-3.4 index (°C) for Ctrl (black) and ensemble mean of SVE (red) overlaid from January0 to December+2 for ENSO-neutral cases. Global mean surface shortwave radiation anomaly (W m−2) in the SVE run is denoted by a green line. The orange shading illustrate how each ensemble is distributed (i.e., probability density function). Longitude–time section of (b) SST (°C), (c) zonal wind (m s−1), and (d) precipitation (mm day−1) anomalies along the equator (2°S–2°N) from January0 to December+2 in the SVE run. The contour interval of SST (zonal wind) is 0.5°C (0.6 m s−1). Peak SVE forcing is represented by a brown triangle.
Citation: Journal of Climate 26, 14; 10.1175/JCLI-D-12-00471.1
To describe the time evolution of the ENSO transition in relation to variability in wind forcing, we also plotted a longitude–time section of the simulated SST (Fig. 3b), zonal wind (Fig. 3c), and precipitation (Fig. 3d) anomalies over the equatorial Indo-Pacific. A strong anomalous westerly wind at the surface (i.e., a weakening of the trade winds) is recognized around the peak of SVE (Fig. 3c), corresponding to weak surface warming (cooling) over the CEP (Indo-western Pacific, Fig. 3b). Lagging behind the peak of the westerly wind anomalies by about one-half to one year, the SST anomalies in the CEP reveal remarkably positive values, which expand westward. As a linear response of the ocean to wind forcing, the anomalous surface westerlies act to deepen the equatorial Pacific thermocline and advect warm surface water from the western to the eastern Pacific. The resultant decreased cold water upwelling and zonal advection further amplify the CEP warming. It appears evident that the air–sea coupled feedback can significantly amplify the initial SVE response. The SST anomalies evolve rapidly from the succeeding summer+1 to winter+1/2 to ultimately show a strong warming with westerly wind anomalies. In the following seasons, the SST anomalies gradually decrease to zero, indicating that the termination of El Niño is established in the winter+2/3.
Figure 4 presents the spatial distribution of the simulated surface air temperature, precipitation, and wind anomalies over the Indo-Pacific Ocean. In the early stage (i.e., around the SVE peak), an anomalous cooling of the surface is seen, particularly over continental areas. Previous studies have reported the Indian Ocean basin as being one of the most sensitive regions to radiative forcing (e.g., Guemas et al. 2013), and we note that cooling (and reduced precipitation; Fig. 3d) in the Indo-Maritime Continent (MC) is robust in comparison with that of the Pacific Ocean. The resultant zonal temperature gradient over the equator can, therefore, be expected to contribute to the reduction of the Walker circulation via an increase (decrease) in the local convective instability (as seen in Fig. 3d) and decrease (increase) in the sea level pressure (e.g., Lindzen and Nigam 1987). Actually, we found surface westerly wind anomalies over the MC at around the SVE peak, while the warm anomalies over the CEP are much weaker. The SVE-related change in the surface temperature can also reduce (intensify) the precipitation over the MC (western-central Pacific), which can drive the anomalous equatorial westerly wind. The surface cooling with the reduced evaporation in the tropics tends to increase atmospheric stability and reduce the convective activity over the MC, which could result in enhanced convergence at the east of the region by an atmospheric Kelvin wave–like response. This response can further amplify the equatorial westerly wind anomalies. Therefore both the surface pressure gradient and indirect effect through the precipitation tend to cause the westerly anomalies that can be enhanced by the air–sea feedback. Such a surface response is also found in other studies of SVEs using different models (Robock et al. 2008). In the following summer [June–August+1 (JJA+1)], significant warm anomalies are seen associated with the onset of El Niño, which enters its mature phase in the subsequent winter. The simulated characteristics of the wind and SST closely resemble observations of El Niño. Also of note is an initial cooling in the equatorial eastern Pacific preceding the large-scale warming in the CEP, which could occur in relation to land surface cooling in the vicinity of the South American continent. The initial warming pattern is relatively similar to that of the central Pacific El Niño (e.g., Kug et al. 2009; Newman et al. 2011) when the following development of El Niño switches off the signal.
Simulated surface temperature (°C, color shading), precipitation (mm day−1, green and purple contours) and surface wind (m s−1, vectors) anomalies for the SVE run of ENSO-neutral cases from September0 to February+2. The contour interval is 1 mm day−1.
Citation: Journal of Climate 26, 14; 10.1175/JCLI-D-12-00471.1
In global warming simulations, we find characteristic features of land–sea contrast with a stronger warming over land than over oceans, implying that land surface temperature is more sensitive to radiative forcing. This is largely due to the different surface and atmospheric feedback that occurs over land in comparison to over oceans (e.g., Sutton et al. 2007; Joshi et al. 2007. The land–sea response ratio of the surface air temperature to an increase in CO2 exceeds 1 in the tropics and subtropics, which is largely consistent between models (Sutton et al. 2007). Such a land–sea contrast is also seen in the SVE experiments. The cooling of the land surface over the tropical–subtropical region is about two times faster and stronger than that of the ocean surface (not shown).
b. SVE impact during the warm and cold phase of ENSO
Because of the probabilistic nature of climate forecast, it is not sufficient to perform the simulations only during the ENSO-neutral year. To assess its impact on climate and its potential for forecasting, an ensemble of many simulations initiated from various phases of ENSO is required. Figures 5a and 5b show the composited time evolution of the Niño-3.4 index for the SVE experiments during El Niño and La Niña, respectively. The results presented here are based on ensemble averages over the 25 individual integrations. We found that the SVE forcing contributes to the relative CEP warming during both El Niño and La Niña, but the effect is much stronger in the warm phase. The SVE forcing during El Niño prevents the transition of El Niño to La Niña (Figs. 5a,b), while during La Niña it weakens the duration of cold events.
Evolution of the composited Niño-3.4 index (°C) for the ensemble mean of noSVE (solid) and SVE (dash) overlaid from January0 to December+2 for (a) El Niño and (b) La Niña cases. (c) RMSE using ensemble mean SST in the tropical Pacific region (20°S–20°N, 120°E–90°W) between SVE and noSVE for El Niño (red), La Niña (blue), and ENSO-neutral (black dashed) cases. Peak SVE forcing is represented by a brown triangle.
Citation: Journal of Climate 26, 14; 10.1175/JCLI-D-12-00471.1
To quantitatively measure the difference between the sensitivity of El Niño and La Niña to a SVE, we calculated the root-mean-square error (RMSE) from noSVE using the ensemble mean SST in the tropical Pacific region (20°S–20°N, 120°E–90°W) for the respective warm and cold ENSO phases (Fig. 5c). In the El Niño phase, the RMSE deteriorated rapidly and exceeded 1.0°C near the boreal spring–summer+1, which is approximately triple that of the La Niña phase. Such asymmetry in the sensitivity of ENSO to the SVE forcing of El Niño and La Niña, can be attributed to the difference in subsequent air–sea process.
To better describe the asymmetry in the ENSO responses, we adopted one each El Niño and La Niña case that showed a marked response (Figs. 6a,b). The results presented here are based on ensemble averages over the five individual integrations. We find a significant difference between the SVE and Ctrl in the winter+1/2 in Niño-3.4 (in the ENSO phase), while no significant difference is seen between noSVE and Ctrl. Figures 6c and 6e (Figs. 6d,f) present the longitude–time section, highlighting the differences in the simulated surface zonal wind (precipitation) between SVE and noSVE near the equator. When the model is integrated without SVE forcing (Fig. 6c, black contour), a close examination of the mature phase reveals surface easterly anomalies along the equator within 120°–160°E. As described in previous studies (e.g., Kug and Kang 2006; Ohba and Ueda 2007), the easterly wind anomalies are known to be associated with a warming of the Indian Ocean, which can accelerate the El Niño transition. Lagging behind the easterly anomalies by a few months, the Niño-3.4 index drops rapidly between the succeeding summer+1 and winter+1/2 to ultimately exhibit negative values (Fig. 6a, black line), indicating that the transition from El Niño to La Niña is established when the model is run without volcanic forcing.
As in Fig. 3a but for simulation of one (a) El Niño and (b) La Niña case. Longitude–time section of the difference in zonal wind (m s−1) between SVE and noSVE (shaded, green contours) along the equator (2°S–2°N) from January0 to December+2 for simulation of one (c) El Niño and (e) La Niña case. Zonal wind anomaly in noSVE is overlaid by black contours: contour interval 1 m s−1. (d),(f) As in (c),(e), but for the difference in precipitation (mm day−1): contour interval 2 mm day−1.
Citation: Journal of Climate 26, 14; 10.1175/JCLI-D-12-00471.1
The differences between the results in the presence or absence of SVEs highlight the impact of a SVE during ENSO events (Figs. 6c,e, shaded). In accordance with ENSO-neutral cases (Fig. 3c), the difference in zonal winds in the central Pacific reveals remarkably positive values that start to strengthen around the SVE peak (Fig. 6c). The SVE-enhanced anomalous westerly wind stress can be a significant contributor in preventing the El Niño transition (Fig. 6a) by exciting a downwelling oceanic Kelvin wave. The anomalous westerly wind anomalies are highly collaborated with the simulated precipitation anomalies in both phases (Figs. 6d,f). Figure 7 presents the wind anomalies and spatial distribution of the simulated precipitation and surface temperature during the El Niño phase for noSVE and SVE around the SVE peak. In the noSVE run, the simulated anomalies show a large-scale structure: the Indian Ocean and CEP warming and cooling in the northern western Pacific (WP). The precipitation and SST anomalies are accompanied by an anomalous anticyclonic circulation centered over the northern WP, which induces enhanced equatorial trade winds. It is widely accepted that ENSO variability exerts a significant impact on the Indian Ocean (e.g., Klein et al. 1999). Positive SST anomalies are known to appear over the Indian Ocean around the mature phase of warm ENSO events, which then persist through the following summer. The increase in incoming solar radiation is mainly responsible for the warming of the Indian Ocean (Klein et al. 1999), with ocean dynamics also playing an important role in the southwestern part of the basin (Xie et al. 2002). However, in the SVE run, the warming (cooling) is suppressed (enhanced) over the tropical Indian Ocean (WP) by a reduction in the incoming shortwave radiation. We find an enhanced surface temperature gradient between the equatorial WP and CEP with the eastward shift in convective anomalies in the SVE run that can be a main factor in enhancing CEP westerly wind anomalies. The strengthened Bjerknes feedback can dominate and weaken the transition process and contribute to the regeneration of El Niño. In contrast to El Niño, the SVE forcing in the La Niña phase reduces the easterly wind anomalies around the peak of forcing (Fig. 6e). The reduced easterly results in the termination of the La Niña duration and therefore the CEP SST anomalies at the end of year (+2) are in an approximately neutral condition (Fig. 6b).
As in Fig. 4 but for one El Niño case in the (a) noSVE and (b) SVE run during DJF0/+1. The contour interval of the precipitation is 2 mm day−1.
Citation: Journal of Climate 26, 14; 10.1175/JCLI-D-12-00471.1
Of significant interest is the stronger warming seen in the El Niño case than that in the La Niña case with SVE forcing (Fig. 5.). This feature is not significant in the comparison between the El Niño and neutral cases (Figs. 3a and 5a). This nonlinearity of the response could be related to the direction of the feedback. In neutral-ENSO and El Niño phases, positive feedback of El Niño significantly amplifies the SVE impact. However, during the La Niña phase, the effect of the SVE forcing is regarded as a damping against the positive feedback of La Niña. Much stronger SVE forcing may be needed to cause the breaking of La Niña.
To investigate the sensitivity of ENSO to changes in the amplitude of SVE forcing, similar experiments were conducted by scaling the SVE intensity to between half and double the times (Fig. 1c). Five ensemble members were used for SVE, along with the additional integration of one member each for 0.5, 1.5, and 2.0 times the value (Fig. 8a). This experimental design allowed us to determine how SVEs of various amplitudes contribute to the CEP warming, regardless of the ENSO phase. We adopted the experiments from one of the El Niño and La Niña cases (Figs. 6a,b). Figures 8b and 8c show the RMSE between SVE and noSVE over the Pacific during DJF+1/2 (squares) and the difference in the Niño-3.4 index between DJF0/+1 and the following DJF+1/2 (circles) of the following year, derived from the ensemble mean of the experiments for El Niño and La Niña phases. In this scaling range, the warming response of the CEP to SVE forcing is relatively linear for both phases. The RMSE and the warming of the Niño-3.4 index are found to be larger for a larger SVE. Compared with La Niña, the sensitivity of El Niño to a change in SVE amplitude is relatively high. This indicates that the coupled feedback greatly amplifies the differences during El Niño, while the effect in the opposite phase is weak. It is also worth noting that the CEP warming (in the 1.5 times and 2.0 times runs in the La Niña phase) overcame the La Niña–related cooling anomalies. Clear-sky surface radiative forcing of approximately −18 to −24 W m−2 could therefore be a threshold point of the CGCM that allows warming of the CEP, despite the La Niña cooling.
(a) Global mean optical depth forcing used in the additional SVE run: 2 (red line), 1.5 (orange line), and 0.5 times the amplitude (blue line) are denoted in addition to the normal SVE forcing (black dashed line). Scatterplot between the SVE amplitude (squares) vs the RMSE between SVE and noSVE, using ensemble mean SST in the tropical Pacific region (20°S–20°N, 120°E–90°W), and the difference in simulated Niño-3.4 index (°C, circles) between DJF0/+1 minus DJF+1/2 for (b) El Niño and (c) La Niña. The spread of individual ensemble members for the Niño-3.4 index is denoted by the error bar.
Citation: Journal of Climate 26, 14; 10.1175/JCLI-D-12-00471.1
4. Physical causes of the El Niño–like response
In this section, we examine the physical causes of the El Niño–like response in the CGCM. It is significant that SVE forcing induces the anomalous surface westerly wind stress over the equatorial WP (Fig. 3c). As also seen in SVE experiments using different CGCMs (Robock et al. 2008), the SVE forcing effectively reduces the simulated precipitation around the MC, which could potentially contribute to causing the anomalous surface westerly wind over the equatorial WP. As documented in previous studies (Ohba and Ueda 2006; Xie et al. 2009), a reduced surface temperature around the Indo-MC can contribute to a reduction of in situ precipitation. Such precipitation around the Indo-MC region can be amplified by the intensified precipitation and cyclonic circulations over the western North Pacific. Similar to the CGCM experiment, an anomalous westerly wind is also evident when the SVE experiments are conducted using the AGCM only (Fig. 9), implying that the incipient wind response can be attributed to the response of the atmosphere–land system to the SVE forcing.
Simulated precipitation (mm day−1, color shading) and surface wind (m s−1, vectors) anomalies for September0–November0 (SON0) derived from the SVE run by using the AGCM.
Citation: Journal of Climate 26, 14; 10.1175/JCLI-D-12-00471.1
To examine whether the equatorial wind response can contribute to the CEP warming ahead of the thermostat mechanism of Clement et al. (1996), we show the time–depth section of ocean temperature in the equatorial central Pacific derived from the ENSO-neutral case (Fig. 10a). In the early stage of a SVE, the surface temperature warming is preceded by the subsurface warming, which implies that the surface warming could be mainly due to a deepening of the thermocline in relation to the initial equatorial westerly wind anomaly.
(a) Time–depth section of the ocean temperature anomaly (°C) in the equatorial central Pacific (2°S–2°N, 190°–220°E) derived from the ENSO-neutral case. (b) Time evolution of the mixed layer heat budget terms (°C month−1). Each line denotes the vertical advection of anomalous subsurface temperature by the climatological-mean upwelling (blue) and the zonal advection of mean SST by anomalous current (red) in the Niño-3.4 region that are essential for the equatorial SST anomaly associated with ENSO. The green line represents the effect of the negative radiation anomaly averaged globally over the tropics between 30°S and 30°N. Its thermostat effect (reduced vertical advection of subsurface temperature by the climatological mean upwelling) in the Niño-3.4 region is denoted by a black dashed line.
Citation: Journal of Climate 26, 14; 10.1175/JCLI-D-12-00471.1
Here 
Using the intermediate hybrid coupled model, we made five ensembles, each of which consisted of a 2.5-yr integration. To separately evaluate the effect of a direct (oceanic) and indirect (atmospheric) response on ENSO, the surface heat flux anomaly, which consists of the surface shortwave and longwave radiation, and the latent and sensible heat fluxes derived from the CGCM ensembles, were forced for an initial 1-yr (Jul0–Jun+1) period. The indirect effect of the MC cooling was represented by including the CGCM's MC cooling in Tdiff, which indirectly affects the model ocean via the surface wind response U0. A simulation, imposing both surface heat flux anomalies and MC cooling, was also conducted (denoted as ALL in Fig. 11).
Evolution of the ensemble-mean Niño-3.4 index (°C) for the intermediate coupled model experiment overlaid from January0 to December+2. The effect of surface shortwave radiation (SSW) anomaly and the MC cooling on ENSO are respectively denoted by orange and blue lines. The ALL forcing run (i.e., both the surface heat flux anomalies and MC cooling) and its spread of individual forecast members are denoted by a black line with gray shading. The forced period is represented by yellow shading.
Citation: Journal of Climate 26, 14; 10.1175/JCLI-D-12-00471.1
The results of the experiments, indicating that the simplified model can reproduce the CGCM-simulated El Niño onset, are presented in Fig. 11. Two possible variables of importance are identified; one is the relatively weakened effect of the reduced incoming solar radiation on the surface cooling in the CEP (known as the dynamical thermostat) and the other is the rapid surface cooling around the MC. Consistent with the mechanism in Clement et al. (1996), the surface radiative cooling contributes to reduce the zonal SST contrast, as a direct oceanic response to SVE, which could then result in CEP warming. Since the simulated surface shortwave radiation is not spatiotemporally uniform, the effect is significantly reduced when compared with the previous studies (Mann et al. 2005; Emile-Geay et al. 2008), as discussed in McGregor and Timmermann (2011). In addition to the radiative cooling, our simulations additionally present the important role of the anomalous zonal gradients of surface temperature on the El Niño onset (via modulation of the precipitation), especially after the SVE peak. In the model experiment, the direct effect of the reduced solar radiation explains about 30% of the total effect in the model, while there is an indirect effect of 50%. However, the thermostat mechanism is arguably the only physical mechanism at play in the Zebiak and Cane (1987) model on those time scales, and it is therefore not surprising that it should show up relatively strongly in experiments using its oceanic component.
5. Discussion and summary
The motivation for our study was to systematically examine the ENSO response to a tropical SVE using a CGCM, conditional on ENSO phase. We found that the radiative forcing of volcanic aerosols in the stratosphere initially creates an El Niño–like response that can be significantly amplified by air–sea interactions in seasons following the SVE. The peak of this equatorial response follows the time of the volcanic forcing by about one year. The results obtained from the CGCM experiments are in excellent agreement with those obtained from the recent proxy evidence of Adams et al. (2003) and McGregor et al. (2010). We therefore conclude that the SVE response of the air–sea coupled dynamics in the Pacific can increase the probability of an El Niño event, in particular one year after the forcing peak during neutral, or El Niño, years. This also implies that a rapid change in radiative forcing could create the additional risk of other events, such as widespread drought and reduced freshwater resources, via a modulation of ENSO.
We also investigated the response to SVE forcing during El Niño and La Niña, because ENSO exhibits a significant asymmetry. The SVE forcing during El Niño significantly prevents the transition to a cold state. Because of self-amplification by air–sea coupled dynamics, the response of the CEP during El Niño is larger than that during La Niña. Therefore, the intensity of the dynamical thermostat-like response of ENSO to a SVE is clearly dependent the ENSO phase. We suggest that it could plausibly be the case in nature as well.
In addition to simple model studies (e.g., Mann et al. 2005), the dynamical thermostat-like response documented in Clement et al. (1996) is also seen in the full CGCM. To diagnose which component excites the El Niño–like response, we analyzed the mixed layer heat budget and an intermediate coupled model. Our model experiments revealed a new mechanism: that the effect of a land–sea cooling contrast (and a relatively rapid cooling of the Indian Ocean) is the dominant mechanism, instead of the direct response of oceanic dynamics to the radiative forcing proposed in Clement et al. (1996). We need to verify that this mechanism is also at play in other coupled GCMs.
We note that this study assumes the same temporal evolution of volcanically induced stratospheric aerosols in all experiments, with magnitude as the only variable (Fig. 8a). We are interested in studying the model response for when the SVE begins at a more rapid rate or occurs in the extratropics, which can potentially alter the sensitivity of ENSO to a SVE. In addition, our experiments also fix the volcanic eruption peak in the boreal winter. Because of the seasonal change in the instability of ENSO, the sensitivity of ENSO to the SVE forcing would, therefore, possibly be different in each season.
McGregor and Timmermann (2011) provide a different explanation for the influence of volcanic eruptions on the ENSO. The SVEs in the CCSM3 model induce enhanced trade winds, which then lead to a deepened thermocline and SST warming after a period of several months, via a recharge process (Jin 1997). The contrast in the response between the models may be due to the difference in the initial response of low-level cloud response in the CEP, or to the intensity of the following feedback process (such as the recharge–discharge process). As denoted by the recent Coupled Model Intercomparison Project phase 3 (CMIP3) and phase 5 (CMIP5) multimodel comparison studies, CGCMs represent various feedbacks (e.g., Guilyardi et al. 2009b). The analysis of cloud radiative feedbacks in convection/subsidence dynamical regimes in the CMIP3 models (Bony and Dufresne 2005) shows that the simulation of marine boundary layer clouds is at the heart of tropical cloud feedback uncertainties in current CGCMs. Marine boundary layer clouds occur in the CEP and therefore biases in their representation can contribute to the diversity of the ENSO response. The other possibility is that the difference in the ENSO system itself between the models. Ohba et al. (2010) investigate the simulated transition process of ENSO in the CMIP3 models and find diversity of the simulated ENSO transition system. Some of the models reproduce the features of the observed transition process of El Niño/La Niña, whereas most models fail to concurrently reproduce the process during both phases. Many of the differences between the models can be traced to the representation of deep convection, trade wind strength, and cloud feedbacks (e.g., Guilyardi et al. 2009a; Lloyd et al. 2009; Sun et al. 2009). In addition to the model biases, the starting point used for the volcanic forcing is also different between McGregor and Timmermann (2011) and this study. The SVE forcing in this study starts from summer with the peak in winter while that in McGregor and Timmermann (2011) uses random peaking. This difference could also contribute to the difference in the responses. Further detailed experiments in view of the seasonal dependence should be performed in the future.
Time evolution of SST and zonal wind anomalies in response to the SVE forcing are relatively similar to other CGCM experiments conducted using the Geophysical Fluid Dynamics Laboratory Climate Model version 2.1 (Stenchikov et al. 2007) and ECHAM and the global Hamburg Ocean Primitive Equation (Lim and Yeh 2012). However, an analysis by the IPCC Fourth and Fifth Assessment Report phases 3 and 5 of the Coupled Model Intercomparison Project (e.g., Taylor et al. 2012, http://cmip-pcmdi.llnl.gov/cmip5) twentieth-century simulation response to volcanic forcing does not provide a statistically significant response in the equatorial CEP (not shown). Numerical simulations produce a considerable range of dynamical responses to volcanic forcing (Stenchikov et al. 2006), which likely depend on diverse aspects of model formulation. The sensitivity to volcanic forcing differs considerably between models. Intermodel comparison of millennium CGCM simulation is needed in order to further discuss the threshold level of the SVE–ENSO relationship.
Acknowledgments
We express special thanks to Drs S. Watanabe, M. Sugiyama, and S. Emori for their helpful suggestions and discussions. This work was supported by the “Program for Risk Information on Climate Change (PRICC)” from MEXT Japan and by the Global Environmental Research Fund (S10) of MOE Japan.
REFERENCES
Adams, J., M. Mann, and C. Ammann, 2003: Proxy evidence for an El Niño-like response to volcanic forcing. Nature, 426, 274–278, doi:10.1038/nature02101.
An, S.-I., F.-F. Jin, and I.-S. Kang, 1999: The role of zonal advection feedback in phase transition and growth of ENSO in the Cane–Zebiak model. J. Meteor. Soc. Japan, 77, 1151–1160.
Bjerknes, J., 1969: Atmospheric teleconnections from the equatorial Pacific. Mon. Wea. Rev., 97, 163–172.
Bony, S., and J. L. Dufresne, 2005: Marine boundary layer clouds at the heart of tropical cloud feedback uncertainties in climate models. Geophys. Res. Lett., 32, L20806, doi:10.1029/2005GL023851.
Cane, M. A., and R. J. Patton, 1984: A numerical model for low-frequency equatorial dynamics. J. Phys. Oceanogr., 14, 1853–1863.
Church, J. A., N. J. White, and J. M. Arblaster, 2005: Significant decadal-scale impact of volcanic eruptions on sea level and ocean heat content. Nature, 438, 74–77, doi:10.1038/nature04237.
Clement, A., R. Seager, M. A. Cane, and S. E. Zebiak, 1996: An ocean dynamical thermostat. J. Climate, 9, 2190–2196.
Emile-Geay, J., R. Seager, M. A. Cane, E. Cook, and G. H. Haug, 2008: Volcanoes and ENSO over the past millennium. J. Climate, 21, 3134–3148.
Gleckler, P. J., K. R. Sperber, and K. AchutaRao, 2006: Annual cycle of global ocean heat content: Observed and simulated. J. Geophys. Res., 111, C06008, doi:10.1029/2005JC003223.
Guemas, V., S. Corti, J. Garcìa-Serrano, F. Doblas-Reyes, M. Balmaseda, and L. Magnusson, 2013: The Indian Ocean: The region of highest skill worldwide in decadal climate prediction. J. Climate, 26, 726–739.
Guilyardi, E., P. Braconnot, F. F. Jin, S. T. Kim, M. Kolasinski, T. Li, and I. Musat, 2009a: Atmosphere feedbacks during ENSO in a coupled GCM with a modified atmospheric convection scheme. J. Climate, 22, 5698–5718.
Guilyardi, E., A. Wittenberg, A. Fedorov, M. Collins, C. Wang, A. Capotondi, G. J. van Oldenborgh, and T. Stockdale, 2009b: Understanding El Niño in ocean–atmosphere general circulation models: Progress and challenges. Bull. Amer. Meteor. Soc., 90, 325–340.
Handler, P., 1984: Possible association of stratospheric aerosols and El Niño type events. Geophys. Res. Lett., 11, 1121–1124.
Hasumi, H., 2006: CCSR Ocean Component Model (COCO) version 4.0. CCSR Rep. 25, 103 pp. [Available online at http://www.ccsr.u-tokyo.ac.jp/~hasumi/COCO/index.html.]
Hasumi, H., and S. Emori, Eds., 2004: K-1 coupled GCM (MIROC) description. K-1 Tech. Rep., 34 pp. [Available online at http://www.ccsr.u-tokyo.ac.jp/kyosei/hasumi/MIROC/tech-repo.pdf.]
Jin, F.-F., 1997: An equatorial ocean recharge paradigm for ENSO. Part I: Conceptual model. J. Atmos. Sci., 54, 811–829.
Jin, F.-F., and S.-I. An, 1999: Thermocline and zonal advective feedbacks within the equatorial ocean recharge oscillator model for ENSO. Geophys. Res. Lett., 26, 2989–2992.
Joshi, M. M., J. M. Gregory, M. J. Webb, D. M. H. Sexton, and T. C. Johns, 2007: Mechanisms for the land/sea warming contrast exhibited by simulations of climate change. Climate Dyn., 30, 455–465, doi:10.1007/s00382-007-0306-1.
Klein, S. A., B. J. Soden, and N. C. Lau, 1999: Remote sea surface temperature variations during ENSO: Evidence for a tropical atmospheric bridge. J. Climate, 12, 917–932.
Kug, J.-S., and I.-S. Kang, 2006: Interactive feedback between the Indian Ocean and ENSO. J. Climate, 19, 1784–1801.
Kug, J.-S., F.-F. Jin, and S.-I. An, 2009: Two types of El Niño events: Cold tongue El Niño and warm pool El Niño. J. Climate, 22, 1499–1515.
Lim, H.-G., and S.-W Yeh, 2012: Volcanic effect on the tropical sea surface temperature in a millennium coupled model simulation. Geophysical Research Abstracts, Vol. 14, Abstract EGU2012-4574. [Abstract available online at http://meetingorganizer.copernicus.org/EGU2012/EGU2012-4574.pdf.]
Lindzen, R. S., and S. Nigam, 1987: On the role of sea surface temperature gradients in forcing low-level winds and convergence in the tropics. J. Atmos. Sci., 44, 2418–2436.
Lloyd, J., E. Guilyardi, H. Weller, and J. Slingo, 2009: The role of atmosphere feedbacks during ENSO in the CMIP3 models. Atmos. Sci. Lett., 10, 170–176.
Mann, M., M. A. Cane, S. E. Zebiak, and A. Clement, 2005: Volcanic and solar forcing of the tropical Pacific over the past 1000 years. J. Climate, 18, 447–456.
McGregor, S., and A. Timmermann, 2011: The effect of explosive tropical volcanism on ENSO. J. Climate, 24, 2178–2191.
McGregor, S., A. Timmermann, and O. Timm, 2010: A unified proxy for ENSO and PDO variability since 1650. Climate Past, 5, 1–17.
Newman, M., S.-I. Shin, and M. A. Alexander, 2011: Natural variation in ENSO flavors. Geophys. Res. Lett., 38, L14705, doi:10.1029/2011GL047658.
Nicholls, N., 1988: Low latitude volcanic eruptions and the El Niño-Southern Oscillation. Int. J. Climatol., 9, 91–95.
Ohba, M., and H. Ueda, 2006: A role of zonal gradient of SST between the Indian Ocean and the western Pacific in localized convection around the Philippines. SOLA, 2, 176–179.
Ohba, M., and H. Ueda, 2007: An impact of SST anomalies in the Indian Ocean in acceleration of the El Niño to La Niña transition. J. Meteor. Soc. Japan, 85, 335–348.
Ohba, M., and H. Ueda, 2009: Role of nonlinear atmospheric response to SST on the asymmetric transition process of ENSO. J. Climate, 22, 177–192.
Ohba, M., and M. Watanabe, 2012: Role of the Indo-Pacific interbasin coupling in predicting asymmetric ENSO transition and duration. J. Climate, 25, 3321–3335.
Ohba, M., D. Nohara, and H. Ueda, 2010: Simulation of asymmetric ENSO transition in WCRP CMIP3 multimodel experiments. J. Climate, 23, 6051–6067.
Robock, A., 2000: Volcanic eruptions and climate. Rev. Geophys., 38, 191–219.
Robock, A., L. Oman, and G. Stenchikov, 2008: Regional climate responses to geoengineering with tropical and Arctic SO2 injections. J. Geophys. Res., 113, D16101, doi:10.1029/2008JD010050.
Sato, M., J. Hansen, M. McCormick, and J. Pollack, 1993: Stratospheric aerosol optical depths, 1850–1990. J. Geophys. Res., 98 (D12), 22 987–22 994.
Self, S., M. R. Rampino, J. Zhao, and M. G. Katz, 1997: Volcanic aerosol perturbations and strong El Niño events: No general correlation. Geophys. Res. Lett., 24, 1247–1250.
Shiogama, H., T. Nagashima, T. Yokohata, S. A. Crooks, and T. Nozawa, 2006: Influence of volcanic activity and changes in solar irradiance on surface air temperatures in the early twentieth century. Geophys. Res. Lett., 33, L09702, doi:10.1029/2005GL025622.
Shiogama, H., S. Emori, T. Mochizuki, S. Yasunaka, T. Yokohata, M. Ishii, T. Nozawa, and M. Kimoto, 2010: Possible influence of volcanic activity on the decadal potential predictability of the natural variability in near-term climate predictions. Adv. Meteor., 2010, 657318, doi:10.1155/2010/657318.
Stenchikov, G., K. Hamilton, R. J. Stouffer, A. Robock, V. Ramaswamy, B. Santer, and H.-F. Graf, 2006: Artic oscillation response to volcanic eruptions in the IPCC AR4 climate models. J. Geophys. Res., 111, D07107, doi:10.1029/2005JD006286.
Stenchikov, G., T. Delworth, and A. Wittenberg, 2007: Volcanic climate impacts and ENSO interactions. Eos, Trans. Amer. Geophys. Union, Vol. 88 (Spring Meeting Suppl.), Abstract A43D-09.
Stothers, R., 2000: Climatic and demographic consequences of the massive volcanic eruption of 1258. Climatic Change, 45, 361–374.
Sun, D.-Z., Y. Yu, and T. Zhang, 2009: Tropical water vapor and cloud feedbacks in climate models: A further assessment using coupled simulations. J. Climate, 22, 1287–1304.
Sutton, R. T., B. Dong, and J. M. Gregory, 2007: Land/sea warming ratio in response to climate change: IPCC AR4 model results and comparison with observations. Geophys. Res. Lett., 34, L02701, doi:10.1029/2006GL028164.
Taylor, K. E., R. J. Stouffer, and G. A. Meehl, 2012: An overview of CMIP5 and the experiment design. Bull. Amer. Meteor. Soc., 93, 485–498.
Vialard, J., C. Menkes, J. P. Boulanger, P. Delecluse, E. Guilyardi, M. J. McPhaden, and G. Madec, 2001: A model study of oceanic mechanisms affecting equatorial Pacific sea surface temperature during the 1997–98 El Niño. J. Phys. Oceanogr., 31, 1649–1675.
Watanabe, M., and Coauthors, 2010: Improved climate simulation by MIROC5: Mean states, variability, and climate sensitivity. J. Climate, 23, 6312–6335.
Watanabe, M., M. Chikira, Y. Imada, and M. Kimoto, 2011: Convective control of ENSO simulated in MIROC. J. Climate, 24, 543–562.
Xie, S.-P., H. Annamalai, F. A. Schott, and J. P. McCreary, 2002: Structure and mechanisms of south Indian Ocean climate variability. J. Climate, 15, 864–878.
Xie, S.-P., K. Hu, J. Hafner, H. Tokinaga, Y. Du, G. Huang, and T. Sampe, 2009: Indian Ocean capacitor effect on Indo–western Pacific climate during the summer following El Niño. J. Climate, 22, 730–747.
Yokohata, T., S. Emori, T. Nozawa, Y. Tsushima, T. Ogura, and M. Kimoto, 2005: Climate response to volcanic forcing: Validation of climate sensitivity of a coupled atmosphere–ocean general circulation model. Geophys. Res. Lett., 32, L21710, doi:10.1029/2005GL023542.
Zanchettin, D., C. Timmreck, H.-F. Graf, A. Rubino, S. Lorenz, K. Lohmann, K. Krüger, and J. H. Jungclaus, 2012: Bi-decadal variability excited in the coupled ocean–atmosphere system by strong tropical volcanic eruptions. Climate Dyn., 39, 419–444, doi:10.1007/s00382-011-1167-1.
Zebiak, S. E., and M. A. Cane, 1987: A model El Niño–Southern Oscillation. Mon. Wea. Rev., 115, 2262–2278.
