Multi decadal variability of sudden stratospheric warmings in an AOGCM

Transcription

1 GEOPHYSICAL RESEARCH LETTERS, VOL. 38,, doi: /2010gl045756, 2011 Multi decadal variability of sudden stratospheric warmings in an AOGCM S. Schimanke, 1 J. Körper, 1 T. Spangehl, 1 and U. Cubasch 1 Received 6 October 2010; revised 10 November 2010; accepted 23 November 2010; published 4 January [1] The variability in the number of major sudden stratospheric warmings (SSWs) is analyzed in a multi century simulation under constant forcing using a stratosphere resolving atmosphere ocean general circulation model. A wavelet analysis of the SSW time series identifies significantly enhanced power at a period of 52 years. The coherency of this signal with tropospheric and oceanic parameters is investigated. The strongest coherence is found with the North Atlantic ocean atmosphere heat flux from November to January. Here, an enhanced heat flux from the ocean into the atmosphere is related to an increase in the number of SSWs. Furthermore, a correlation is found with Eurasian snow cover in October and the number of blockings in October/November. These results suggest that the multi decadal variability is generated within the oceantroposphere stratosphere system. A two way interaction of the North Atlantic and the atmosphere buffers and amplifies stratospheric anomalies, leading to a coupled multi decadal mode. Citation: Schimanke, S., J. Körper, T. Spangehl, and U. Cubasch (2011), Multi decadal variability of sudden stratospheric warmings in an AOGCM, Geophys. Res. Lett., 38,, doi: /2010gl Introduction [2] The influence of stratospheric anomalies on tropospheric weather and climate is most intense during major sudden stratospheric warmings (hereafter referred to as SSWs) [e.g., Baldwin and Dunkerton, 2001]. A dramatic reduction of SSWs at the end of the 20th century is reported by Gillett et al. [2002], followed by a strong increase in the last decade [Cohen et al., 2009]. It has been argued that decadal to multi decadal variability in the strength of the polar vortex is driven by external forcing factors, such as solar variability [Kodera and Kuroda, 2002]. In the present study we investigate to what extent such variability can be internally generated by a feedback mechanism in the fully coupled ocean atmosphere system. [3] Potential precursors of SSWs on intra seasonal timescales are positive El Niño Southern Oscillation [Brönnimann, 2007, and references therein] and blocking events [Martius et al., 2009] that are connected to strong tropospheric wave fluxes into the stratosphere. On the seasonal time scale, positive snow cover anomalies over Eurasia during October have been shown to enhance wave activity and weaken the polar vortex [Fletcher 1 Institute for Meteorology, Freie Universität Berlin, Berlin, Germany. Copyright 2011 by the American Geophysical Union /11/2010GL et al., 2009]. A 20 year long ( ) positive trend in observed snow cover extend consistent with the increase in the number of SSWs is reported by Cohen et al. [2009]. North Atlantic decadal variability can be viewed as a coupled ocean atmosphere mode [Wu and Liu, 2005]. Proxy records reveal a 62 year cycle in North Atlantic sea surface temperatures (SSTs) [Fischer and Mieding, 2005]. Positive SST anomalies can lead to the development of blockings [Croci Maspoli and Davies, 2009]. In turn blockings are reported to be precursors of SSWs [Martius et al., 2009]. Moreover, while there is an indication that a coupling between the stratosphere, the troposphere, and the ocean exists on even longer time scales, the mechanisms are still poorly understood. While seasonal predictability may arise from interactions between land and atmosphere [Cohen et al., 2009], we hypothesize that the ocean drives multi decadal fluctuations of SSWs. 2. Model and Data [4] Due to the shortness and sparseness of oceanic and stratospheric observations, climate models are an important tool to analyze coupling processes of these subsystems. However, due to computational limits, comprehensive studies on a fully coupled system including ocean, troposphere, and stratosphere are still at the very beginning. Here, we use the fully coupled Atmosphere Ocean General Circulation Model EGMAM (ECHO G with Middle Atmosphere Model) [Huebener et al., 2007; Körper et al., 2009; Spangehl et al., 2010]. The atmospheric component is ECHAM4, extended for the middle atmosphere (MA ECHAM4). The dynamic part is represented in T30 spectral resolution. There are 39 vertical levels with the top level located at 0.01 hpa ( 80 km). A gravity wave parameterization and horizontal diffusion following Manzini and McFarlane [1998] are implemented. [5] The coupled ocean model (HOPE G) has a horizontal resolution equivalent to T42 with equator refinement and 20 vertical levels. The model includes a dynamic and thermodynamic sea ice model [Legutke and Voss, 1999]. The model does not produce a quasi biannual oscillation (QBO). Thus, the model wind in the tropical lower stratosphere is a permanent weak easterly corresponding to a continuous QBO easterly phase. [6] The model has been run for 410 years under constant pre industrial conditions (e.g., constant CO 2 concentration of ppmv and fixed solar forcing). The investigation focusses on decadal to multi decadal variability in the number of SSWs. SSWs are identified by employing an algorithm based on Charlton and Polvani s [2007], using the zonal mean of zonal wind at 60 N. A temperature gradient criteria 1of6

2 and a climatological threshold to better exclude final warmings were implemented. 3. Results [7] The model develops 2.1 SSWs/decade under preindustrial conditions. This is less than half the number observed (6.0 SSWs/decade), and a common problem of many GCMs [Charlton et al., 2007]. In our model SSWs are mainly underestimated during early winter, while the number of SSWs in February and March is similar to observations. The duration and the strength of SSWs are comparable to observations. [8] Positive snow cover anomalies over Eurasia in October are related to a higher probability of SSWs during January (not shown), consistent with the mechanism proposed by Cohen et al. [2007]. Moreover, these snow anomalies are accompanied by negative temperature anomalies and a strengthening of the northern part of the Siberian high, as has been proposed by Cohen et al. [2007]. After the occurrence of SSWs the simulated stratospheric circulation anomalies propagate down to the surface, which is in accordance with observations [Baldwin and Dunkerton,2001].Overall, seasonal variations connected with SSWs are properly representedinthemodel. [9] The time series of the number of SSWs per winter demonstrates fluctuations on inter annual to multi decadal time scales (Figure 1a). For instance, around model year 50 there is a period with only a few SSWs, and the number of SSWs is particularly high around model year 330. The absolute minimum of the number of SSWs in a 30 year period is 1, while the absolute maximum is 14 in 30 years (4.7 SSWs/decade). This means that during periods with an anomalously high number of SSWs, the value is close to the observed mean. [10] A continuous wavelet analysis following Torrence and Compo [1998] is employed to analyze variability of SSWs in the time frequency domain. The most pronounced spectral power in the time series of SSWs is found for periods between 40 and 60 years, with a maximum at 52 years (Figure 1b). Enhanced power on this time scale extends over the whole simulation outside the cone of influence. Still, using a Monte Carlo permutation the significance level of 95% is reached in the second half of the simulation only. Enhanced power on other time scales exists only temporarily. For this reason the following investigations focus on the multi decadal variability, since it is here that the strongest signal is found. [11] As hypothesized, the multi decadal variability of SSWs is not expected to be self generated inside the stratosphere but is assumed to be connected to other climate subsystems. Therefore, further lower atmospheric and oceanic parameters are investigated. As in the case of the SSW wavelet analysis, significantly enhanced power is found for the 100 hpa meridional heat flux (a proxy for tropospheric wave forcing entering the stratosphere [Newman et al., 2001]) for winter months (DJF), and at periods between 40 and 60 years over the whole experiment (Figure 1c). In particular, between model year 200 and 350 enhanced wavelet power is found at periods of 50 to 60 years, as was seen in the SSW time series. [12] In order to investigate the oceanic influence on the number of SSWs, ocean atmosphere heat fluxes are computed. The heat fluxes of all tropical ocean basins (Pacific, Atlantic and Indian Ocean) reveal only short term and no multi decadal variability. In the North Pacific, however, variability is most pronounced on time scales even longer than for SSWs (e.g., up to 100 years) (not shown). We conclude that the tropical ocean basins and the North Pacific can therefore only marginally influence the multidecadal variability of SSWs. The wavelet analysis of the North Atlantic (40 80 N, 70W 30E, November to January) ocean atmosphere heat flux (OAF) displays power on time scales similar to SSWs (Figure 1d). Especially enhanced variance on periods between 40 and 60 years is present throughout the whole simulation. Here, the 95% significance threshold is past for two periods, both lasting approximately 100 years (Figure 1d). [13] We perform wavelet analyses for Eurasian snow cover and blocking events for late autumn/early winter, following the mechanism proposed by Cohen et al. [2007] to explain a feedback on the seasonal time scale. The analysis of Eurasian snow cover (0 188 E and north of 24 N)/blocking events (following Tibaldi and Molteni [1990]) is based on October/October and November fields. While observed snow cover exhibits a quasi decadal oscillation [Saito and Cohen, 2003], the modeled frequency response of Eurasian snow cover is similar to white noise. However, enhanced power is found at periods around 20 years and in the second half of the run for periods of 40 to 60 years, though not statistically significant (Figure 1e). The blocking time series has temporarily significantly enhanced power between 10 and 20 years, while less power is displayed for longer periods (Figure 1f). [14] To investigate interactions of these parameters, wavelet coherencies are calculated following Grinsted et al. [2004] (Figure 2). The 100 hpa meridional heat flux and SSWs reveal a non stationary coherency on all time scales (Figure 2a). Since strong tropospheric wave forcing triggers SSWs, the phase relationship between the series indicated by the phase arrows shows an in phase variation. The most persistent coherence of both time series is found for the multi decadal variability (Figure 2a). In the first half of the experiment there are coherent periods in the range of 20 to 40 years. During the second half a narrow but very consistent band of coherency is found for periods around 60 years. [15] A connection between OAF and SSWs time series is found on periods between 40 and 65 years (Figure 2b). The maximum coherency is highly significant, and is found for periods of 55 years, which is close to the most pronounced period of SSWs (52 years). The direction of phase arrows indicates that a strong heat flux from the North Atlantic into the atmosphere is connected with an enhanced number of SSWs. Variations in SSWs and OAF are slightly lead by OAF (1 year) during the first 150 years of the simulation, while in the second half the time series of SSWs leads by approximately 3 4 years. [16] The wavelets of snow cover extent and blockings have only marginal power on multi decadal time scales. Nevertheless, they still reveal temporarily significant coherence with SSWs for periods close to 55 years. Both parameters vary in phase with SSWs on all time scales (Figures 2c 2of6

3 Figure 1. (a) The time series of SSWs as single winter events. Continuous wavelet power spectrums following Torrence and Compo [1998] conducted for the time series of (b) SSWs, (c) 100 hpa heat flux (40 to 80 N, DJF), (d) oceanatmosphere heat flux in the North Atlantic (40 to 80 N, November to January), (e)eurasian snow cover in October and (f) blockings (October and November). Black lines indicate significant power on the 95% level compared to red noise based on AR1 coefficient. The cone of influence (COI) is shown as a lighter shade. and 2d). This indicates that the seasonal mechanism is part of the multi decadal variability of SSWs. 4. Conclusions and Discussion [17] In this study, based on an AOGCM simulation driven by constant forcing, we identified significantly enhanced multi decadal variability in the number of SSWs with a period of 52 years. Wavelet analysis shows similar behaviour for 100 hpa meridional heat flux and the ocean atmosphere heat flux in the North Atlantic (OAF). The wavelet coherences illustrate that the parameters analyzed (100 hpa meridional heat flux, OAF, blockings and Eurasian snow cover) vary in phase at multi decadal periods, indicating a close relationship. Note that since OAF, snow cover extent, and blockings are examined for late autumn/early winter, an inherent phase shift to SSWs of several months is already taken into account in our analysis. Thus, by definition the oceanic and tropospheric parameters precede the number of SSWs. The strongest influence is found for OAF. [18] We suggest a mechanism for the multi decadal variability in the number of SSWs which is illustrated in Figure 3. 3of6

4 Figure 2. Squared wavelet coherence following Grinsted et al. [2004] between the time series of SSWs and (a) 100 hpa meridional heat flux, (b) ocean atmosphere heat flux in the North Atlantic, (c) Eurasian snow cover and (d) blockings. Arrows indicate the relative phase relationship between the series (e.g., pointing right: in phase; left: anti phase; down: SSWs leading other parameter by 90). It is an extension of the conceptual models of Reichler et al. [2005] and Cohen et al. [2007]. Positive heat flux anomalies from the North Atlantic into the atmosphere in conjunction with enhanced snow cover over Eurasia and more blocking events strengthen the 100 hpa meridional heat flux, which is proportional to wave flux entering the stratosphere. Consequently, the polar vortex weakens and the number of SSWs increases. Subsequently, surface weather is influenced by downward propagating stratospheric anomalies in the following weeks. These anomalies may then act as a stochastic forcing on the North Atlantic, and trigger an oscillation with the eigenfrequency of the ocean consistent with the suggestion by Hasselmann [1976]. This eigenmode stimulates an oscillation of the entire coupled system with all subsystems varying nearly in phase. In this manner anomalies persist over an extended period and result in a multi decadal variability of SSWs. [19] Our investigations indicate that the tropical ocean basins contribute to variability on higher frequencies (not shown). In our model at these frequencies the strongest coherence with the occurrence of SSWs is found for the heat flux of the Indian Ocean. A potential contribution from the northern extratropical Pacific to multi decadal variability is not observed in this study since the wavelet coherency between OAF and SSWs is low (not shown). Pinto et al. [2010] find multi decadal periods with significantly enhanced or 4of6

5 Figure 3. Schematic mechanism for how decadal variability of SSWs is generated. For more detail see text. Based on the works by Reichler et al. [2005] and Cohen et al. [2007]. weakened coupling of the Pacific North American pattern and the North Atlantic oscillation, and they speculate that the stratosphere could be part of a mechanism explaining the variable link between those two modes. Therefore, additional analyses are needed to determine the role of tropospherestratosphere interactions in the Pacific region for long term variability of SSWs. [20] Many studies examine projected changes in the mean number of SSWs by comparing simulated data for and [e.g., Huebener et al., 2007;SPARC CCMVal, 2010]. Variability on the multi decadal time scale can possibly explain the contradictory results. A large multi decadal variability can mask the trends. Thus, the interpretation of the projections can differ, especially if periods with a length in the range of the multi decadal variability are compared. [21] Finally, the observed bidecadal upward trend in the number of SSWs [Cohen et al., 2009] hints at multi decadal variability in observations. The observational time series is, however, still too short to analyse multi decadal characteristics. Here, only multi century simulations using oceantroposphere stratosphere models provide sufficient data to support our findings. Further model studies are needed to confirm the existence of enhanced multi decadal variability in the number of SSWs. Finally, understanding of the underlying mechanisms would provide the potential to improve decadal predictions for the Northern Hemisphere, at least for the winter season. [22] Acknowledgments. Parts of this work were funded by the EU project ENSEMBLES (contract GOCE CT ), and by the DFG project ProSECCO/CAWSES (contract LA 1025/5 1). The authors thank the Deutsches Klimarechenzentrum (DKRZ) for providing computing time, and Irina Fast and Falk Niehörster for running the experiment. References Baldwin, M. P., and T. J. Dunkerton (2001), Stratospheric harbingers of anomalous weather regimes, Science, 294, Brönnimann, S. (2007), Impact of El Niño Southern Oscillation on European climate, Rev. Geophys., 45, RG3003, doi: /2006rg Charlton, A. J., and L. M. Polvani (2007), A new look at stratosphric sudden warmings. Part I: Climatology and modeling benchmarks, J. Clim., 20, Charlton, A. J., L. M. Polvani, J. Perlwitz, F. Sassi, E. Manzini, K. Shibata, S. Pawson, J. E. Nielsen, and D. Rind (2007), A new look at stratospheric sudden warmings. Part II: Evaluation of numerical model simulations, J. Clim., 20, Cohen, J., M. Barlow, P. J. Kushner, and K. Saito (2007), Stratospheretroposphere coupling and links with Eurasian land surface variability, J. Clim., 20, Cohen, J., M. Barlow, and K. Saito (2009), Decadal fluctuations in planetary wave forcing modulate global waming in late boreal winter, J. Clim., 22, , doi: /2009jcli Croci Maspoli, M., and H. C. Davies (2009), Key dynamical features of the 2005/06 European winter, Mon. Weather Rev., 137, Fischer, H., and B. Mieding (2005), A 1000 year ice core record of interannual to multidecadal variations in atmospheric circulation over the North Atlantic, Clim. Dyn., 25, Fletcher, C., S. Hardiman, P. Kushner, and J. Cohen (2009), The dynamical response to snow cover pertubations in a large ensemble of atmospheric GCM integrations, J. Clim., 22, , doi: / 2008JCLI Gillett, N. P., M. R. Allen, R. E. McDonald, C. A. Senior, D. T. Shindell, and G. A. Schmidt (2002), How linear is the Arctic Oscillation response to greenhouse gases?, J. Geophys. Res., 107(D3), 4022, doi: / 2001JD Grinsted, A., J. C. Moore, and S. Jevrejeva (2004), Application of the cross wavelet transform and wavelet coherence to geophysical time series, Nonlinear Processes Geophys., 11, Hasselmann, K. (1976), Stochastic climate models. Part I. Theory, Tellus, 28, , doi: /j tb00696.x. Huebener, H., U. Cubasch, U. Langematz, T. Spangehl, F. Niehörster, I. Fast, and M. Kunze (2007), Ensemble climate simulations using a fully coupled 5of6

6 ocean troposphere stratosphere general circulation model, Philos. Trans. R. Soc. A, 365, Kodera, K., and Y. Kuroda (2002), Dynamical response to the solar cycle, J. Geophys. Res., 107(D24), 4749, doi: /2002jd Körper, J., T. Spangehl, U. Cubasch, and H. Huebener (2009), Decomposition of projected regional sea level rise in the North Atlantic and its relation to the AMOC, Geophys. Res. Lett., 36, L19714, doi: / 2009GL Legutke, S., and R. Voss (1999), The Hamburg atmosphere ocean coupled circulation model ECHO G, Tech. Rep. 18, German Clim. Comput. Cent., Hamburg, Germany. Manzini, E., and N. A. McFarlane (1998), The effect of varying the source spectrum of a gravity wave parameterization in a middle atmosphere general circulation model, J. Geophys. Res., 103, 31,523 31,539. Martius, O., L. M. Polvani, and H. C. Davies (2009), Blocking precursors to stratospheric sudden warming events, Geophys. Res. Lett., 36, L14806, doi: /2009gl Newman, P. A., E. R. Nash, and J. E. Rosenfield (2001), What controls the temperature of the Arctic stratosphere during the spring?, J. Geophys. Res., 106, 19,999 20,010. Pinto, J. G., M. Reyers, and U. Ulbrich (2010), The variable link between the PNA and NAO in observations and in multi century CGCM simulations, Clim. Dyn., doi: /s x. Reichler, T., P. J. Kushner, and L. M. Polvani (2005), The coupled stratosphere troposphere response to impulsive forcing from the troposphere, J. Atmos. Sci., 62, Saito, K., and J. Cohen (2003), The potential role of snow cover in forcing interannual variability of the major Northern Hemisphere mode, Geophys. Res. Lett., 30(6), 1302, doi: /2002gl Spangehl, T., U. Cubasch, C. C. Raible, S. Schimanke, J. Körper, and D. Hofer (2010), Transient climate simulations from the Maunder Minimum to present day: Role of the stratosphere, J. Geophys. Res., 115, D00I10, doi: /2009jd SPARC CCMVal (2010), SPARC Report on the Evaluation of Chemistry Climate Models, SPARC Rep. 5, WCRP 132, WMO/TD 1526, edited by V. Eyring, T. G. Shepherd, and D. W. Waugh, SPARC, Toronto, Ont., Canada. Tibaldi, S., and F. Molteni (1990), On the operational predictability of blocking, Tellus, Ser. A, 42, Torrence, C., and G. P. Compo (1998), A practical guide to wavelet analysis, Bull. Am. Meteorol. Soc., 79, Wu, L., and Z. Liu (2005), North atlantic decadal variability: Air sea coupling, oceanic memory, and potential Northern Hemisphere resonance, J. Clim., 18, U. Cubasch, J. Körper, S. Schimanke, and T. Spangehl, Institute for Meteorology, Freie Universität Berlin, Carl Heinrich Becker Weg 6 10, D Berlin, Germany. (semjon.schimanke@met.fu berlin.de) 6of6