Simulation of the climate impact of Mt. Pinatubo eruption using ECHAM5 – Part 2: Sensitivity to the phase of the QBO and ENSO

Abstract. The sensitivity of the climate impact of Mt. Pinatubo eruption in the tropics and extratropics to different QBO phases is investigated. Mt. Pinatubo erupted in June 1991 during the easterly phase of the QBO at 30 hPa and the phase change to westerly took place in August 1992. Here, the consequences are analyzed if the QBO phase had been in the opposite phase during the eruption of Mt. Pinatubo. Hence, in this study, simulations are carried out using the middle atmosphere configuration of ECHAM5 general circulation model for two cases – one with the observed QBO phase and the other with the opposite QBO phase. The response of temperature and geopotential height in the lower stratosphere is evaluated for the following cases – (1) when only the effects of the QBO are included and (2) when the effects of aerosols, QBO and SSTs (combined response) are included. The tropical QBO signature in the lower stratospheric temperature is well captured in the pure QBO responses and in the combined (aerosol + ocean + QBO) responses. The response of the extratropical atmosphere to the QBO during the second winter after the eruption is captured realistically in the case of the combined forcing showing a strengthening of the polar vortex when the QBO is in its westerly phase and a warm, weak polar vortex in the easterly QBO phase. The vortex is disturbed during the first winter irrespective of the QBO phases in the combined responses and this may be due to the strong influences of El Nino during the first winters after eruption. However, the pure QBO experiments do not realistically reproduce a strengthening of the polar vortex in the westerly QBO phase, even though below normal temperatures in the high latitudes are seen in October-November-December months when the opposite QBO phase is prescribed instead of the December-January-February winter months used here for averaging.


Introduction
The quasi-biennial oscillation (QBO) in the zonal winds in the equatorial lower stratosphere is a well known mode of interannual variability. The zonally symmetric easterly 20 and westerly wind regimes alternate regularly with a mean period of 28-29 months. The alternating wind regimes develop in the upper stratosphere near 3 hPa and propagate downward at an approximate rate of 1 km/month to the tropopause. The amplitude of the easterly phase is stronger than the westerly phase. The easterly zonal winds can reach as high as 35-40 m/s, whereas the westerly zonal winds reach [15][16][17][18][19][20]  stratosphere by a broad spectrum of vertically propagating waves including Kelvin and Rossby-Gravity waves (refer Baldwin et al. (2001) for details). There is considerable variability of the QBO in period and amplitude. The QBO influences the extratropical northern stratosphere. Studies have shown that the geopotential height at high latitudes is significantly lower during the westerly 5 phase of QBO than during the easterly phase Tan, 1980, 1982). Labitzke (1987) and Labitzke and Van Loon (1988) found a strong relation of the QBO signal to the 11-year solar cycle during January and February and it was shown that during the easterly phase, for solar maxima, there exists an intensified cold polar vortex and vice versa for solar minima. The QBO also affects the winter stratospheric tempera-10 tures depending on the ENSO phase (Garfinkel and Hartmann, 2007), for example, our model simulations with observed SSTs and QBO show that when ENSO is in its warm state, the influence of QBO is reduced (refer to part-I of this paper). Several studies showed that the phase of the QBO influences the weather events in the troposphere. Indian summer monsoon rainfall activity seems to be directly related with the phase 15 of the QBO (Bhalme et al., 1987;Mukherjee at al., 1985). Using General Circulation Model simulations,  showed that the tropical tropospheric circulation is significantly influenced by the QBO wherein less precipitation is observed in the western Pacific, but, more in the Indian subcontinent during the westerly QBO phase. Yasunari (1989) showed that the QBO in the lower stratosphere is coupled with the sea 20 surface temperature anomalies in the equatorial Pacific. Chattopadhyay and Bhatla (2002) showed that the Indian monsoon rainfall is strongly inversely correlated with the SST anomalies over the Niño3 region for all the seasons from the concurrent summer to the following winter during the easterly QBO phase.
The QBO also plays an important role in the distribution of chemical constituents like 25 ozone, water vapor and methane and aerosols (Trepte and Hitchman, 1992;Trepte et al., 1993;Baldwin et al., 2001). Planetary wave activity is much less in the easterly phase of the QBO compared to the westerly phase, which means that the aerosols are trapped in the equatorial belt during the easterly phase of QBO and are dispersed  et al., 1993). Lidar observations of the stratospheric aerosol layer at 11.1 • E) show that for about 3 years, the tropical explosive eruptions such as Mt. Pinatubo (1991) andEl Chichon (1982) eruption show the same decay rate of 12 months when the QBO phases of these two eruptions are synchronized (Jaeger, 2005). 5 The easterly and westerly phases of the QBO have different effects on the stratospheric extratropical circulation. Here, the sensitivity of the effect of large volcanic eruptions on the high latitude circulation to the QBO phase is evaluated. Mt. Pinatubo erupted on 15 June 1991 during the easterly phase of the QBO at 30 hPa and the change to the westerly phase took place in August 1992 at 30 hPa and remained in 10 the same phase till May 1993. It would be interesting to understand the climate impact of Mt Pinatubo eruption if it had erupted during the opposite phase. Here, the main focus is to see whether the radiative and dynamical responses following Mt. Pinatubo eruption are modulated by the phase of the QBO. Most GCMs are not able to simulate a spontaneous QBO. But, in the recent years, attempts have been made to include 15 QBO forcing in GCMs either by assimilating the observed zonal winds at Singapore to the model winds or by considering a sufficient spatial resolution, a realistic simulation of tropical convection and the consideration of the effects of gravity waves (Hamilton, 1998;Bruhwiler and Hamilton, 1999;Giorgetta et al., 2002;Stenchikov et al., 2004;Giorgetta et al., 2006). For this study, the middle atmosphere version of ECHAM5 is 20 modified to include the QBO forcing by nudging the zonal mean zonal winds in the tropics to the prevailing zonal wind observations at Singapore following .
The response for individual or combined forcings, including volcanic aerosols and ozone anomalies, observed SSTs and the QBO in two opposite phases are discussed 25 in detail in the following sections. The responses in the observed phase are already discussed in part-I of this paper, but are shown here again for easy comparison.  (Manzini et al., 2006), topmost level at 0.01 hPa. Both the volcanic aerosol forcing data and the ozone anomaly data (Stenchikov et al., 2002) are compiled by G. Stenchikov and are used in this study for the specific model resolution.
For the runs including the QBO, a spin up of 17 months is carried out with observed 10 SST and with the observed QBO phase. Ten ensemble runs are carried out with different initial conditions. The initial conditions are chosen arbitrarily from the 15 year unperturbed run with climatological SST as boundary conditions. To include the QBO forcing in this study, the zonal winds in the tropics are nudged towards the zonal wind observations at Singapore . The nudging is applied 15 uniformly in a core domain and extends with decreasing nudging rate to the boundary of the domain. The latitudinal core domain specified for the study here is 7 N-7 S and the domain boundary is 10 N-10 S. In the vertical the core domain and the boundary is over the levels extending from 70 hPa to 10 hPa. The nudging rate is (10 days) −1 . The opposite QBO phase is prescribed along with observed SST and sea ice as boundary 20 conditions for both the perturbed and unperturbed runs. As mentioned before, there is significant variability of the QBO in period and amplitude. To extract the QBO-related zonal winds that are opposite of that occurring during the Pinatubo eruption, the correlation co-efficient is calculated between the 50 hPa zonal mean zonal winds at Singapore for the years 1953-2004 and the 50 hPa zonal 25 winds of 1991/1993. The time period of maximum negative correlation co-efficient is chosen as the opposite QBO phase (hereafter referred to as QBO) and in this case, the best anti-correlated years are from June 1975-May 1977 run) and the unperturbed run with observed SST as boundary conditions. The combined aerosol+ocean+QBO responses are denoted by AOQ for the observed QBO phase and by AOQ for the opposite QBO phase. These response are calculated as the difference between the combined AOQ/AOQ experiment and the unperturbed run with climatological SST (C c ) as boundary conditions. 20

Results and discussion
The first part discusses the responses in temperature and geopotential height at 30 hPa to the QBO phases alone. The QBO exhibits a clear signature in stratospheric temperature with pronounced signals in tropics and extratropics (Baldwin et al., 2001). The tropical temperature QBO is in thermal wind balance (Andrews et al., 1987)  namely one for the phase change from from easterly to westerly (QBO) and the other from westerly to easterly (QBO) is investigated. Fig. 2 shows the lower stratospheric temperature response to (a) QBO and to ( Baldwin et al. (2001). However, the temperature response associated with the westerly phase of the QBO in Fig. 2a,b is comparatively weaker. To explain this better, the climatological mean differences in the annual cycle of lower stratospheric temperature at 30 hPa between the experiments including and excluding the QBO is shown in Fig. 2c. It can be clearly seen that the stratospheric temperature clima-Introduction  Giorgetta, personal communication). This explains why the warm temperature anomalies observed during the westerly QBO shear are weaker.

30 hPa geopotential height response in boreal winter
The differences in the 30 hPa geopotential height anomaly in response to the QBO 5 phase is presented in Fig. 3a,b for the two boreal winters following the eruption. The anomalies are weaker in DJF (Dec-Jan-Feb) 1991/1992 in the QBO anomalies where positive anomalies are observed over southern Europe, Russia and Siberia (up to 40 m) and negative anomalies over Scandinavia, parts of Greenland and northern Canada (up to -40 m). The second winter shows a strong and larger area of below normal geopotential height anomalies (as low as -160 m) over northern Eurasia and Greenland and above normal geopotential height anomalies over Canada and North Atlantic. The vortex observed in the winter of 1992/1993 is slightly shifted over northern Eurasia. For comparison, the response with the observed QBO phase is shown in Fig. 3c,d.

15
In the first winter following the eruption, the QBO is in its easterly phase and QBO is in its westerly phase at 30 hPa and the opposite is observed during the second winter. During the first winter, the anomaly patterns do not simulate the strengthening of the polar vortex in either of the cases, though the QBO favors a strong polar vortex (Holton and Tan, 1980). This may be because of the strong influence of El Niño. Compar-20 ing the response to QBO for the second winter with the response to QBO, the model simulates negative polar geopotential height anomalies in both the QBO phases. This means that the model simulates the anomalously cold polar vortex irrespective of the QBO phase. This contradicts the study by Holton and Tan (1980) Figure 4 shows the 30 hPa temperature response when the aerosol forcing, El Niño and QBO effects are included for two years following the eruption. The only difference 5 is that Fig. 4a has the observed QBO phase and the Fig. 4b has the opposite phase as can be seen in the color bars given at the bottom of the Figure. The effects due to the contrasting QBO phases are clearly evident. A cooling of about 1-2 K from June 1991-April 1992 in (a) and from January 1992-May 1993 in (b) is seen in the latitudinal belt 10 N-10 S and warmer temperatures are observed in the subtropics. This dual peak with a relative maximum in the subtropics and minimum at the equator during the easterly phase of QBO is well simulated by both experiments. The response in the latitudinal belt 50N -50S is statistically significant at > 90% significance level. Colder temperature anomalies are observed during the westerly phase of QBO in northern hemisphere (NH) winter in the polar latitudes as in Fig. 4a,b and this may 15 be associated with the strengthening of the polar vortex. This cooling is more prominent during the westerly phases in AOQ where the cooling is persistent over October-November-December, whereas the cooling is confined to Dec in AOQ. Strong warm anomalies can be seen in January-February-March during the westerly QBO phases in both the experiments, but, the warming in the AOQ forcing is much weaker than 20 seen in AOQ forcing. Significant above normal temperature anomalies are also seen during the easterly phases of QBO in December-January-February months. These anomalies are statistically significant at 90% significance level. As mentioned before, there was an ongoing El Niño and when one compares these results with the pure QBO temperature responses, it can be seen that the strong anomalies in the high 25 latitudes are insignificant meaning that these anomalies are a result of the complex in- teractions between aerosols, QBO and SSTs. It is shown that the change of the QBO phase can also bring about changes in the extratropical winter circulation in the lower stratosphere.

30 hPa geopotential height response
The geopotential height anomalies at 30 hPa for the two winters following the eruption 5 are shown in Fig. 5a,b in AOQ and in Fig. 5c,d in AOQ runs. The anomalies in AOQ during the winters of 1991/1992 and 1992/1993 exhibit a wave number one pattern with positive anomalies over northern Pacific, Canada, Alaska and Siberia and negative anomalies over north western Europe and North Atlantic. The geopotential height anomalies reach as low as -100 m and as high as 140-160 m. There are no notable 10 differences between the anomalies of the two winters except that the anomalies in the second winter following the eruption are relatively stronger than in the first winter. As mentioned before, the westerly phase of the QBO favors a strong polar vortex. But, in AOQ, the westerly phase occurs during the El Niño winter, which in turn, disturbs the polar vortex. 15 Comparing the responses for the combined forcings including the observed QBO phase and the opposite QBO phase, it can be seen that the anomaly pattern is more or less similar in the first winter, where positive anomalies positioned over the Arctic circle and the negative anomalies cover a smaller region compared to the anomalies in the AOQ response. This similarity in the responses is due to the fact that the El Niño 20 effects override the effects due to the phase change of QBO. However, the response in the second winter differs considerably with a large center of low geopotential height anomalies over the Arctic circle, Greenland and north eastern Europe and Siberia in AOQ when the QBO is in the westerly phase that favors a strong polar vortex and is not seen in AOQ, when the QBO is in the easterly phase, again, supporting the studies 25 by Holton and Tan (1980) and Holton and Tan (1982 Figure 6a and b show the ensemble mean surface temperature anomalies for the first and second winters respectively in AOQ. For comparison purposes, the ensemble mean surface temperature anomalies in AOQ are also shown in Fig. 6c and d. It can be seen that one of the main features of the volcanic forcing, the so-called "volcanic 5 winter pattern" (Graf et al., 1993;Kirchner and Graf, 1995;Robock and Mao, 1995;Stenchikov et al., 2002) is not simulated by the model in any of the winters. The tropical warming in the Pacific due to the El Niño event of 1991/1992 is clearly seen in (a). During the first winter, the anomalies in AOQ and AOQ are more or less the same, except for some minor differences. The warming over northern North America associated with El Niño is statistically significant in AOQ, though the magnitude of the anomalies are captured irrespective of the phases. The pattern exhibited in Fig. 6a and c is similar to the ocean response (refer Fig. 3g,h of Part-I), thereby clearly signifying the dominance of ENSO effects over the effects due to the change of QBO phase. However, during the second winter when the effects of El Niño are negligible, the combined effects due to 15 aerosols and the change of phase of QBO are seen. The warming over northern parts of Europe, Russia and Greenland and cooling over N. America and parts of Canada are simulated irrespective of the phase of the QBO. But, strong statistically significant cooling in the Middle East, India and China is simulated only during the easterly QBO phase as in Fig. 6b, while these anomalies are weaker and not significant during the 20 westerly phase of QBO as seen in Fig. 6d. Hence, it can be seen that the surface temperature response is independent of the phase of the QBO during the first winter after Mt. Pinatubo eruption due to the presence of El Niño, but differs over Asia during the second winter when El Niño effects are reduced. QBO in NH winter in the polar latitudes associated with the strengthening of the polar vortex. Strong warm anomalies are observed in northern high latitudes in late winter during the easterly phase of the QBO. This warming is statistically significant at 90% confidence level and is also evident during the westerly QBO phase in the AOQ experiment. 15 Our results show that the climate response after explosive tropical eruptions is significantly modulated by the QBO phase. Major differences owing to the QBO phase are observed in the tropics and extratropics in the lower stratosphere temperature response. While significant differences in the dynamical response at the surface are seen in the tropics and the subtropics, this study shows that the modulation by the QBO is 20 minimal beyond 60N. The use of prescribed aerosol and nudged QBO in this study restricts the understanding of the the effects of the different QBO phases on the transport and mixing of the aerosols. However, studies will be carried out to investigate the effect of aerosols on the QBO. Introduction