Multiple time scales of the southern annular mode

Abstract

Using the ensemble empirical mode decomposition method, this study systematically investigates the multiple timescales of the southern annular mode (SAM) and identifies their relative contributions to the persistence of the SAM. Analyses show that the persistence of SAM mainly depends on the contribution of longer-timescale variabilities, especially the cross-seasonal variability of SAM. When subtracting the cross-seasonal variability from the SAM, the long-term positive covariance between the eddy forcing and zonal flow disappears. Composite analysis shows that the meridional shift of zonal wind, eddy momentum forcing and baroclinicity anomalies can be maintained for more than 40 days only under the circumstance of strong cross-seasonal variability, indicating the dominant role played by the cross-seasonal variability for the low-frequency persistence of the SAM. Analysis further shows that the cross-seasonal variability of the SAM, in addition to the internal dynamics, is associated with the extratropical air–sea interaction. About half of the strong cross-seasonal SAM events are accompanied by evident extratropical dipolar SST anomalies, which mostly occur in austral summer. The cross-seasonal dependence of the low-frequency change in SAM suggests that the contribution of longer-timescale variabilities, especially the cross-seasonal variability, cannot be neglected in subseasonal prediction of the SAM.

Appendices

Appendix 1: Spectra for SAM Variabilities with Different Timescales

Using EEMD, the SAM index can be decomposed into different timescales. Figure 11 shows the spectral distributions of the synoptic, subseasonal, cross-seasonal and interannual components of SAM index displayed in Fig. 1b–e. As shown in Fig. 11a, the spectrum of synoptic component of SAM index features significant power between time scales shorter than 10 days. The spectrum of subseasonal component displays significant power between 10 and 70 days (Fig. 11b), and the cross-seasonal component exhibits significant power between 80 and 350 days, with two peaks at around 150 and 240 days (Fig. 11c). The spectrum of interannual component is also shown in Fig. 11d, with an evident peak at around 550 days (1.5 years).

Fig. 11

Spectral distribution for the a synoptic, b subseasonal, c cross-seasonal, and d interannual components of SAM index. The dashed red lines denote values at 95% confidence level based on red noise test

Appendix 2: Cross-seasonal air–sea interactions

Figures 12a–d display the covariance between s(t) and the oceanic Ekman heat transport, surface heat flux, surface air temperature anomalies and lower tropospheric meridional temperature gradient, respectively. At negative lags, the oceanic Ekman heat transport in Fig. 12a exhibits significant dipolar structure in meridional direction, suggesting a major role of Ekman transport to the formation of the dipolar SST anomalies. As the Ekman transport is primarily surface wind driven, the surface wind anomaly in the SAM is likely one of the main drivers of the extratropical SST anomalies as in Xiao et al. (2016). At positive lags, in which s(t) leads the Ekman transport, the dipolar pattern lasts for more than one month, contributing to the SST anomalies as well.

Fig. 12

Lagged covariance between the normalized s(t) and a Ekman heat transport (shading interval: 0.3 Wm⁻²), b surface sensible and latent heat flux (shading interval: 0.12 Wm⁻²), c 2 m air temperature (shading interval: 0.02 K), d zonal mean baroclinicity at 850 hPa (shading interval: 0.3 × 10^–7 km⁻¹), e eddy momentum forcing (shading interval: 0.5 × 10^–6 ms⁻²) and f zonal mean zonal winds (shading interval: 0.2 ms⁻¹). Positive lags denote that s(t) leads. In Fig. 7a, positive Ekman heat transport denotes heating the sea surface. In Fig. 7b, positive surface flux denotes warming the sea surface and cooling the lower atmosphere. The dotted regions in those panels denote values above the 95% confidence level

The surface sensible and latent heat flux in Fig. 12b, however, exhibits different patterns at negative and positive lags. At negative lags, the surface heat flux exhibits roughly similar patterns to the SST anomalies in Fig. 6a though with a poleward shift, suggesting that the surface flux also contributes to the SST anomalies. When s(t) leads, the pattern of surface flux anomalies changes. Around 40°S, the anomalous surface heat flux acts to heat the atmosphere and cool the ocean surface. Combined with the warm SST anomaly in the region, such change in surface flux suggests a heating from the ocean surface to the atmosphere. In the lower and higher latitudes, the anomalous surface flux acts to cool the atmosphere. Combined with the cold SST anomalies in those regions, the pattern again suggests the influence of the ocean to the atmosphere. With such surface heat flux, the surface air temperature in Fig. 12c exhibits consistent change with the SST anomaly at positive lags, contributing to the persistent change in meridional temperature gradient in lower troposphere (Fig. 12d). As suggested by previous observational and modeling studies of SAM (e.g. Nie et al. 2013, 2016; Xiao et al. 2016), such persistent latitudinal shift in lower level temperature gradient can maintain a persistent SAM by resulting in a latitudinal shift of eddy activity thus eddy forcing exerting on the zonal wind. The above suggested processes are observed in our diagnostics as well. As shown in Figs. 12e, f, when s(t) leads, the eddy momentum forcing and zonal wind all exhibit consistent poleward shift, lasting for more than one month. All the above results suggest that, the dipolar SST anomalies in the extratropics in turn affect the atmosphere above, acting to extend the SAM.