Climate Dynamics

, Volume 46, Issue 11, pp 3689–3707

A decomposition of ENSO’s impacts on the northern winter stratosphere: competing effect of SST forcing in the tropical Indian Ocean

Article

DOI: 10.1007/s00382-015-2797-5

Cite this article as:
Rao, J. & Ren, R. Clim Dyn (2016) 46: 3689. doi:10.1007/s00382-015-2797-5

Abstract

This study applies WACCM, a stratosphere-resolving model to dissect the stratospheric responses in the northern winter extratropics to the imposed ENSO-related SST anomalies in the tropics. It is found that the anomalously warmer and weaker stratospheric polar vortex during warm ENSO is basically a balance of the opposite effects between the SST anomalies in the tropical Pacific (TPO) and that over the tropical Indian Ocean basin (TIO). Specifically, the ENSO-related SST anomalies over the TIO are to induce an anomalously colder and stronger stratospheric polar vortex during warm ENSO, which acts to partially cancel out the much stronger warmer and weaker polar vortex response to the SST anomalies over the TPO. Further analysis indicates that, while the SST forcing from the TPO contributes to the anomalously positive Pacific North America (PNA) pattern in the troposphere and the enhancement of the stationary wavenumber (WN)-1 in the stratosphere during warm ENSO, the TIO SST forcing is to induce an anomalously negative PNA and a reduction of both WN-1 and WN-2 in the stratosphere. Diagnosis of E–P flux confirms that, the anomalously upward propagation of stationary waves in the extratropics mainly lies over the western coast of North America during warm ENSO, which is mainly associated with the TPO-induced positive PNA response and is partially suppressed by the effect of the accompanying TIO SST forcing.

Keywords

El Niño-Southern Oscillation (ENSO) Indian Ocean basin (IOB) Stratospheric polar vortex WACCM Pacific–North America pattern (PNA) 

1 Introduction

As one of the major sources of interannual variability in the ocean–atmosphere system, the El Niño-Southern Oscillation (ENSO) events occurring in the tropical eastern Pacific can influence not only the oceanic and atmospheric conditions in the entire tropics but also those across the global extratropics in both hemispheres. Particularly, the significant effects of ENSO on the northern extratropical stratosphere have been identified in observational data (Van Loon et al. 1982; Labitzke and Van Loon 1989; Camp and Tung 2007; Free and Seidel 2009; Xie et al. 2014a, b) and model simulations (Hamilton 1995; Sassi et al. 2004; Manzini et al. 2006; García-Herrera et al. 2006; Cagnazzo and Manzini 2009; Cagnazzo et al. 2009; Ineson and Scaife 2008). It was found that the northern stratospheric polar vortex tends to be anomalously warmer/colder and weaker/stronger during El Niño/La Niña winters (Van Loon et al. 1982; Labitzke and Van Loon 1989; Camp and Tung 2007; Free and Seidel 2009). Numerical experiments with the stratosphere-resolved model WACCM under perpetual January conditions by Taguchi and Hartmann (2006) indicated that the stratospheric sudden warming (SSW) events are twice as likely to occur in El Niño winters than in La Niña winters. It was also reported by Bell et al. (2009) that SSW events occur more frequently in the Reading Intermediate General Circulation Model under El Niño conditions than that under the climatological SST conditions. It was indicated that ENSO regulates the intensity of the stratospheric polar vortex by influencing the upward propagation of the planetary-waves from the troposphere to the stratosphere (Manzini et al. 2006; Camp and Tung 2007; Ineson and Scaife 2008; Ren et al. 2012). Relative to that by La Niña events, the teleconnection pattern in the troposphere induced by El Niño events is to increase the planetary wavenumber (WN)-1 but reduce the planetary WN-2, and the effect of WN-1 overwhelms that of the WN-2 to result in an anomalously warmer and weaker stratospheric polar vortex (Garfinkel and Hartmann 2007, 2008). Nevertheless, there are earlier studies indicating that the relationship between ENSO and the polar stratospheric variability may not be robust or statistically significant (Hamilton 1993, 1995; Baldwin and O’sullivan 1995). A fundamental difficulty in determining ENSO’s effects on the stratosphere is how to separate them from the entangled Quasi Biannual Oscillation (QBO) signals in the extratropical stratosphere (Garfinkel and Hartmann 2007, 2008; Wei et al. 2007; Calvo et al. 2009) due to the phase coincidence of QBO with ENSO in early records (Wallace and Chang 1982; Van Loon and Labitzke 1987; Baldwin and O’sullivan 1995). Later studies by Camp and Tung (2007) showed that the spatial patterns of the ENSO perturbations to the polar stratosphere are nearly orthogonal to that of QBO, and the magnitudes of ENSO perturbations are comparable with QBO perturbations. By analyzing the model results with sufficient data length and without QBO variability, Manzini et al. (2006) and García-Herrera et al. (2006) identified a significant relationship between ENSO and the stratospheric polar vortex variability. They also showed that warm ENSO tends to induce a PNA-like pattern and enhance WN-1 in the stratosphere, thus resulting in the anomalous polar warming during the late ENSO winter and the early subsequent spring.

It is known that the PNA pattern that links the ENSO forcing in the tropics to the atmospheric variability in the extratropics is intimately related to the diabatic heating induced by the ENSO SST anomalies over the tropical eastern Pacific (Wallace and Chang 1982; Simmons et al. 1983; Jin and Hoskins 1995; Newman and Sardeshmukh 1998; Annamalai et al. 2007). However, recent studies have found that, the anomalous diabatic heating over other tropical oceans particularly over the Indian Ocean may also play a role in modulating the PNA pattern (Barsugli and Sardeshmukh 2002; Annamalai et al. 2007). Linear model experiments with idealized tropical forcings showed that the 500-hPa PNA pattern forced by the imposed SST anomalies over the Indian Ocean, tends to destructively interfere with that forced by the idealized SST anomalies over the tropical eastern Pacific (e.g., Simmons et al. 1983; Branstator 1985; Ting and Sardeshmukh 1993; Jin and Hoskins 1995; Newman and Sardeshmukh 1998; Annamalai et al. 2007). Model studies by Kumar and Hoerling (1998), Farrara et al. (2000), and Spencer et al. (2004) also confirmed the possible different effects of the Indian Ocean SST forcing on the PNA pattern from that of Pacific SST forcing. Analysis of the model results from an atmospheric general circulation model (AGCM) in the National Centers for Environmental Prediction (NCEP) by Barsugli and Sardeshmukh (2002) indicated that, there exists a nodal line near 100°E in terms of the sensitivity of the PNA-like responses to the tropical SST forcing, across which the polarity of the PNA response seems to reverse.

Because of the “atmospheric bridge” processes induced by ENSO in the tropics, interannual variations of the SST over the tropical Indian Ocean as well as the SST over the tropical Atlantic, often follow that of the ENSO SST over the tropical eastern Pacific (Klein et al. 1999; Alexander et al. 2002; Kumar and Hoerling 2003). Following the occurrence of warm/cold ENSO (El Niño/La Niña) events in the tropical eastern Pacific, the tropical Indian Ocean and the tropical Atlantic also becomes warmer/colder. Many previous studies have proved the significant lead/lag correlations between the SST anomalies in the tropical eastern Pacific in winter and those in the tropical Indian Ocean (e.g., Cadet 1985; Tourre and White 1995; Lanzante 1996; Nicholson 1997; Klein et al. 1999; Murtugudde and Busalacchi 1999; Yu and Rienecker 1999; Ding and Li 2012) and the tropical Atlantic (e.g., Weare et al. 1976; Covey and Hastenrath 1978; Curtis and Hastenrath 1995; Lanzante 1996; Enfield and Mayer 1997; Nicholson 1997; Klein et al. 1999; Alexander et al. 2002) throughout the winter and subsequent spring. Some studies also related these SST responses in the tropical Indian Ocean and Atlantic Ocean to the zonally homogeneous atmospheric warming/cooling responses in the tropical troposphere to warm/cold ENSO (e.g., Newell and Weare 1976; Angell and Korshover 1978; Angell 1981; Pan and Oort 1983; Reid et al. 1989).

The different role played by the SST forcing in the Indian Ocean from that in the tropical eastern Pacific, in modulating the PNA pattern, seems to suggest that the SST forcing in the tropical Indian Ocean, though it is ENSO-induced, may act to counter with the original ENSO SST forcing in the tropical eastern Pacific and to weaken the otherwise stronger effects of ENSO on the extratropical stratosphere, regarding the intimate associations of the PNA anomalies in the troposphere with the planetary wave activity in the extratropical stratosphere in northern winter. However, because of the difficulties in distinguishing the effects of the SST forcings from different tropical oceans in observations, specific evidence on the roles of different tropical oceans in modulating the ENSO’s effects on the extratropical stratosphere is still lack. Most modeling studies usually consider the canonical ENSO SST anomalies over the eastern Pacific as the tropical ENSO forcing when they examine the responses of the extratropical atmosphere to ENSO, though the extratropical responses to ENSO in their simulations are usually much stronger than that in the observation [e.g., Fig. 5 in Taguchi and Hartmann (2006); Figs. 2, 3 in Bell et al. (2009); Fig. 4 in Ineson and Scaife (2008); Fig. 9 in Lan et al. (2012)].

The objective of this study is to use the WACCM and conduct a series of numerical experiments with the ENSO-related tropical SST forcing confined to different ocean basins for a dissection of the ENSO’s effects on the extratropical stratosphere. Particularly, we will provide modeling evidence to demonstrate, how the anomalous tropical SST forcings from different oceans, coordinate with one another to modulate the PNA pattern in the troposphere and contribute to the consequent ENSO’s effects on the extratropical stratosphere. The results will further clarify the dynamical linkage between the ENSO-related tropical forcings and the atmospheric responses in the extratropical stratosphere, and advance our knowledge of the dynamical roles of different tropical oceans in contributing to the ENSO’s effect on the extratropical atmosphere.

The remainder of this paper is organized as follows. Section 2 describes the data, model, experiment design, and analysis procedure. Section 3 demonstrates the different responses of the extratropical stratosphere to the ENSO-related tropical SST forcings over different ocean basins. Section 4 analyzes the dynamical connections between the tropical forcings over different ocean basins and the circulation responses from the troposphere to the stratosphere. Summary and conclusions are presented in Sect. 5.

2 Model, data and analysis procedure

2.1 Model

We employed the version 4 of the stratosphere-resolved AGCM model WACCM (WACCM4, Garcia et al. 2007) which is developed by the National Center for Atmospheric Research (NCAR). The WACCM4 is one of the atmosphere components of the NCAR Community Earth System Model CESM (version 1.0.4). It is also a superset of the Community Atmospheric Model version 4 (CAM4), and includes all of the physical parameterizations in CAM4 (Neale et al. 2013). WACCM is a “high top” chemistry–climate model with 66 vertical levels extending from the surface to 5.1 × 10−6 hPa (approximately 150 km), and with a horizontal resolution of 1.9° (latitude) × 2.5° (longitude). The main improvements in version 4 include the updated parameterization schemes for non-orographic gravity waves generated by frontal systems and convection, and for surface stress due to unresolved topography (Garcia et al. 2007; Richter et al. 2010). It was found that the improved parameterization for surface stress or the turbulent mountain stress has led to a dramatic improvement on frequency of SSW events in northern winter (Richter et al. 2010; Marsh et al. 2013). The relatively high reproducibility of WACCM for the stratospheric variability as well as the ENSO signals in stratosphere has been confirmed in many previous studies (e.g., Sassi et al. 2004; Taguchi and Hartmann 2006; Garcia et al. 2007; Garfinkel and Hartmann 2008; Taguchi 2010; Calvo and Marsh 2011; Xie et al. 2012; Hegyi et al. 2014).

2.2 Data and experiment design

The monthly SST fields covering the period 1950–2010 are extracted from the SST and sea–ice datasets provided by the Meterological Office, Hadley Centre for Climate Prediction and Research (Rayner et al. 2006). The area mean SST anomalies in the Niño3 region (5°S–5°N, 150°–90°W) are used to define the monthly Niño3 index as a representation of ENSO signal. The climatological mean air temperature, geopotential height, and wind fields, which are used for validation of the WACCM simulations, are from the European Center for Medium Range Weather Forecasting interim reanalysis data (ERA-Interim). The ERA-Interim data has a horizontal resolution of 1.5° (latitude) × 1.5° (longitude) and 37 vertical levels from 1000 hPa to 1 hPa (Dee et al. 2011).

Table 1 lists all our WACCM experiments and their SST configurations. The WACCM4 cannot reproduce the QBO spontaneously, but QBO can be artificially added in the model as an external forcing. In our experiments, QBO forcing is turned off to avoid entanglement of QBO and ENSO signals. In specific, the control experiment is forced by the annual cycle of monthly SST obtained for the period 1950–2010 (i.e., SST is a function of month but not of year). We run the control experiment for 32 continuous years and use the last 30 years to obtain a reference state of the seasonal evolution of the atmospheric circulation without ENSO forcing. We referred to it as the “ClmSST” experiment.
Table 1

Experimental SST forcing descriptions

Experiment

Forcing field descriptions

ClmSST

Climatological SST fields of 1950–2010

ENSO

Same as ClmSST, but the ENSO-related SST anomalies in Fig. 1a added in the tropical Oceans (30°S–30°N, 0°–360°)

TPO

Same as ClmSST, but the ENSO-related SST anomalies in Fig. 1a added in the tropical Pacific Ocean (30°S–30°N, 135°E–70°W)

TIO

Same as ClmSST, but the ENSO-related SST anomalies in Fig. 1a added in the tropical Indian Ocean (30°S–30°N, 30°E–135°E)

TAO

Same as ClmSST, but the ENSO-related SST anomalies in Fig. 1a added in the tropical Atlantic Ocean (30°S–30°N, 70°W–10°E)

See context for details

Note that both El Niño and La Niña events generally mature in winter season and the dominant variabilities of ENSO lie in the timescale of 3–5 years (~87 % of the total variance of Niño3 index, see Ren et al. 2012), we first applied a 3–5 year band-pass digital filter on the monthly Niño3 index to effectively capture only those winter-matured ENSO events and their effects on the winter stratosphere. Then we performed a linear lead-lag (−24 to +24 months) regression of the monthly SST anomalies in the tropics (30°S–30°N) against the filtered Niño3 index and obtained a typical spatial–temporal evolution of the tropical SST anomalies following the development of canonical ENSO events (Fig. 1). Based on the Niño3-regressed tropical SST anomalies in the tropics, a complete ENSO cycle in an approximate 4-year period was constructed, which evolves from neutral to warm ENSO winter and then back to neutral state, and continues to progress to cold ENSO winter and then back to neutral ENSO winter again (solid in Fig. 1b). We apply this quadrennial SST anomaly evolution to represent the tropical SST forcing during canonical ENSO. All the following sensitivity experiments listed in Table 1 were carried out by imposing this quadrennial SST anomaly field on the climatological SST evolution. The only differences among these sensitivity experiments are the different regions where the quadrennial SST anomalies were imposed. The SST anomalies throughout the entire tropics were imposed in the “ENSO” experiment to simulate the overall effects of ENSO on the atmospheric circulation. The quadrennial SST anomalies only in the tropical eastern Pacific were imposed in the “TPO” experiment to simulate the effects of the ENSO-related SST forcing purely from the tropical eastern Pacific; and it is similar for the “TIO” and the “TAO” experiment to represent the SST forcing that is respectively from the tropical Indian ocean (TIO) and the tropical Atlantic Ocean (TAO) during ENSO. In addition, the initial condition fields for all the sensitivity experiments are from the ClmSST results, and we restarted the integrations every 4 years and run them in total for 30 ENSO cycles (or 120 model years) that are initiated with 30 different initial conditions. In other words, each ensemble (cycle) member was run for 4 years with the quadrennial ENSO evolution for every sensitivity experiment and 30 members were performed in total. Then, the ensemble mean is calculated over the 30 ensemble members for every sensitivity experiment. The ensemble mean represents the climatology in each sensitivity experiment. Note that the time length of the ensemble mean in each sensitivity experiment is 48 months, and the difference between the ensemble mean in a sensitivity experiment and the climatology in ClmSST represent the corresponding response in different stages of the quadrennial ENSO cycle. The composite anomaly fields across these 30 sample cycles are used to get the spatial–temporal patterns of the atmospheric response to the ENSO-related SST forcings, which are displayed in the form of warm-minus-cold ENSO composites.
Fig. 1

a Canonical El Niño SSTA pattern obtained by regressing the tropical sea surface temperature anomalies (units: K) against the DJF-mean Niño3 (5°S–5°N, 150°W–90°W) index. b Lead-lag regressions of the area-averaged SSTA over the Niño3 region (solid line, left ordinate), over the tropical Indian Ocean [(30°S–30°N, 40°E–110°E), dashed line, right ordinate], and the tropical Atlantic Ocean [(30°S–30°N, 70°W–10°E), dotted line, right ordinate] against the DJF-mean Niño3 index

2.3 Analysis methods

To demonstrate the anomalous planetary wave activity due to tropical SST forcing, the two dimensional (2D) quasi-geostrophic E–P flux and its divergence in spherical coordinates (Andrews et al. 1987) were diagnosed for all the experiments. The meridional (Fy) and vertical (Fz) components of E–P flux are expressed as:
$$\begin{aligned} &F_{y} = - \rho_{0} a\cos \varphi \overline{{u^{'} v^{'} }} \hfill \\ F_{z} = \rho_{0} a\cos \varphi \frac{Rf}{{HN^{2} }}\overline{{v^{'} T^{'} }} , \hfill \\ \end{aligned}$$
(1)
where ρ0 is air density, a is the radius of the earth, φ is the latitude, R is the gas constant, f is the Coriolis parameter, H is the constant scale-height (7 km), N is the buoyancy frequency, u and v are the zonal and meridional wind, and T is the air temperature. The overbar denotes the zonal mean, and prime denotes the departure from the zonal mean. The divergence of E–P flux is then expressed as:
$$D_{F} = \frac{{{\nabla } \cdot {\mathbf{F}}}}{{\rho_{0} a\cos \varphi }} = \frac{{{{\partial \left( {F_{y} \cos \varphi } \right)} \mathord{\left/ {\vphantom {{\partial \left( {F_{y} \cos \varphi } \right)} {a\cos \varphi \partial \varphi + {{\partial F_{z} } \mathord{\left/ {\vphantom {{\partial F_{z} } {\partial z}}} \right. \kern-0pt} {\partial z}}}}} \right. \kern-0pt} {a\cos \varphi \partial \varphi + {{\partial F_{z} } \mathord{\left/ {\vphantom {{\partial F_{z} } {\partial z}}} \right. \kern-0pt} {\partial z}}}}}}{{\rho_{0} a\cos \varphi }}.$$
(2)
We also diagnosed the three-dimensional (3-D) E–P flux defined by Plumb (1985) to display the spatial patterns of stationary wave activity. The quasi-geostrophic 3-D E–P flux in a spherical and a vertical log-pressure coordinate is defined as:
$$F_{s} = \left( \begin{aligned} F_{x} \hfill \\ F_{y} \hfill \\ F_{z} \hfill \\ \end{aligned} \right) = \frac{p\cos \varphi }{1000hPa}\left( \begin{aligned} \frac{1}{{2a^{2} \cos^{2} \varphi }}\left[ {\left( {\frac{\partial \psi '}{\partial \lambda }} \right)^{2} - \psi '\frac{{\partial^{2} \psi '}}{{\partial \lambda^{2} }}} \right] \hfill \\ \frac{1}{{2a^{2} \cos^{2} \varphi }}\left( {\frac{\partial \psi '}{\partial \lambda }\frac{\partial \psi '}{\partial \varphi } - \psi '\frac{{\partial^{2} \psi '}}{\partial \lambda \partial \varphi }} \right) \hfill \\ \frac{{f^{2} }}{{2N^{2} a\cos \varphi }}\left( {\frac{\partial \psi '}{\partial \lambda }\frac{\partial \psi '}{\partial z} - \psi '\frac{{\partial^{2} \psi '}}{\partial \lambda \partial z}} \right) \hfill \\ \end{aligned} \right),$$
(3)
where Fx, Fy, and Fz denote the longitudinal, latitudinal, and vertical components of the 3-D E–P flux, respectively; p and \(\psi^{'}\) denotes the pressure and the perturbed quasi-geostrophic stream function; and other variables are the same as for the 2-D E–P flux.

3 Stratospheric responses to the tropical ENSO-related SST forcings from different ocean basins

3.1 Winter stratospheric climatology in WACCM

Before showing the stratospheric responses to ENSO forcings, we first display in Fig. 2 the winter (DJF) climatology (contours) and the standard deviation (shadings) of the zonal-mean temperature and zonal wind from the ERA-Interim reanalysis and from our control experiment (ClmSST) by WACCM respectively, to validate the reproducibility of the WACCM in simulating the northern winter stratosphere. Comparing Fig. 2a with c and Fig. 2b with d, it is seen that the winter climatology as well as the variability of the zonal-mean temperature and zonal wind in northern hemisphere are generally well reproduced by the model. Specifically, ERA-Interim shows a cold action center (~205 K) at about 50 hPa over the pole region, which is well reproduced by WACCM. The strength and location of the stratospheric polar jet (10 hPa, ~25 m s−1, 65–70°N) as well as that of the upper tropospheric subtropical jet (~200 hPa, ~30°N, ~40 m s−1) in Fig. 2b are both fairly well reproduced in Fig. 2d. In addition, the model can successfully reproduce the large variability of the polar stratospheric temperature (shadings in Fig. 2a, c) and the accompanied dominant variability of the intensity of the stratospheric polar jet (shadings in Fig. 2b, d). As indicated in Rao et al. (2015), WACCM is currently one of the best GCM models in reproducing the circulation variabilities in winter stratosphere.
Fig. 2

Climatology (contours) and standard deviation (shadings) of the DJF zonal-mean a, c temperature (units: K) and b, d zonal wind (units: m s−1) respectively from a, b ERA-Interim and c, d the control experiment ClmSST. Contour interval is 5 K for temperature and 5 m s−1 for zonal wind

3.2 Stratospheric responses in boreal winter

Shown in Fig. 3 are the warm-minus-cold ENSO composites of the zonal-mean temperature and zonal-mean zonal wind anomalies in northern winter from our sensitivity experiments with different tropical SST configurations for ENSO cycle. It is seen that relative to cold ENSO winters, the tropical troposphere is anomalously warmer during warm winters (Fig. 3a), which is accompanied by a stronger subtropical westerly jet (Fig. 3b). Related to the responses in the tropical troposphere, the tropical stratosphere is anomalously colder which is coupled with an anomalously warmer extratropics from the midlatitude to the polar stratosphere (Fig. 3a). This corresponds to an anomalously warmer and weaker stratospheric polar vortex during winters of warm ENSO, also as indicated by the easterly anomalies of the stratospheric polar jet in Fig. 3b. The warmer and weaker stratospheric polar vortex in response to warm ENSO is quite similar with that found in the observations (Manzini et al. 2006; Lan et al. 2012; Ren et al. 2012).
Fig. 3

Composite responses of the DJF zonal-mean temperature (left panels, units: K) and zonal-mean zonal wind (right panels, units: m s−1) in Northern Hemisphere in a, b ENSO; c, d TPO; e, f TIO and g, h TAO with respect to the control experiment ClmSST; and i, j the differences of that between ENSO and TPO. Contour interval is 0.5 K for temperature and 1 m s−1 for zonal wind. Light and dark shadings indicate the 90 and 95 % statistical confidence levels for a two-tailed Student’s t test

When the ENSO SST forcing is confined to the tropical eastern Pacific in the TPO experiment, or the coordination of the SST forcings in the tropical Indian Ocean and the tropical Atlantic Ocean are excluded, it is seen that the responses in the tropical troposphere do become slightly weaker than that in the “ENSO” experiment. In spite of this, the stratospheric polar vortex responses to warm ENSO do not become relatively weaker, but instead become much stronger than that in the “ENSO” experiment (Fig. 3c, d vs. Fig. 3a, b). This seems to suggest that, the ENSO-associated SST forcings in the tropical Indian Ocean or in the tropical Atlantic do contribute to the ENSO effects in the tropical troposphere, but may act to partially cancel out that in the polar stratosphere. To further confirm this, we show in Fig. 3e, f and 3g, h the atmospheric responses to the ENSO-associated SST forcing imposed in the tropical Indian Ocean only (TIO experiment) and in the tropical Atlantic only (TAO experiment), respectively. It is seen that, though the TIO SST forcing also induces a tropical tropospheric warming response as the TPO forcing, it does not yield a weakening stratospheric polar jet response or a stratospheric polar warming response, but oppositely, a significant response of stratospheric polar cooling and strengthening of the stratospheric polar jet. Comparing with the responses to TIO, the responses to TAO in Fig. 3g, h are much weaker and insignificant. The trivial atmospheric responses to TAO suggest that the cancelling effect may mainly lie in the tropical Indian Ocean. By comparing Fig. 3e, f with 3i, j, it is further clear that the difference in responses to tropical SST forcings between “ENSO” and TPO can be largely accounted for by the effects of TIO.

3.3 Seasonal evolutions of the stratospheric responses

Seasonal evolutions of the zonal-mean temperature and zonal-mean zonal wind responses in the northern stratosphere are shown in Fig. 4 for each sensitivity experiment. Relative to the cold ENSO, the tropical stratosphere becomes anomalously colder since the developing summer of the warm ENSO, and meanwhile the extratropical stratosphere is anomalously warmer, from the midlatitudes to the pole in midwinter (Fig. 4a). The cold anomalies in the tropical lower stratosphere are mainly caused by the strengthened overshooting of tropical convective activities and are always coupled with the warmer anomalies in the tropical troposphere (see Fig. 3a), as also indicated by Ren et al. (2012) and Rao et al. (2014). By using a model without chemistry modules, Rao et al. (2014) found that the convective anomalies cause the out-of-phase relationship of the temperature anomalies between the troposphere and the lower stratosphere by overshooting. At the same time, consistent with the results of Bekki et al. (2013) and Hu et al. (2014), our simulations with coupled chemistry suggest that ozone decrease in the tropical lower stratosphere (not shown) may partially contribute to the cold anomalies. Also, an enhanced Brewer-Dobson circulation associated with an weakened westerly flow in the midlatitude stratosphere causes ascent and cooling in the tropics and descent and warming in the extra-tropics (e.g. Bekki et al. 2013; Hu et al. 2014). Scaife et al. (2003) indicated that more water vapor enters the stratosphere during warm ENSO, which may also contribute to the cooling in the tropical lower stratosphere by emitting more long wave radiation as a greenhouse gas. Accompanied with the colder response in the tropical stratosphere and the warmer response in the polar stratosphere shown in Fig. 4a are the easterly anomalies of the stratospheric polar jet throughout the winter in Fig. 4b. Both the temperature and zonal wind responses exhibit an out-of-phase relationship between the lower and the higher latitudes.
Fig. 4

Seasonal evolutions of the zonal-mean temperature (left panels, units: K) and zonal-mean zonal wind (right panels, units: m s−1) responses in the northern stratosphere in a, b ENSO, c, d TPO, e, f TIO and g, h the differences between ENSO and TPO. Contour interval is 0.5 K for temperature and 1.0 m s−1 for zonal wind. Light and dark shadings indicate the 90 and 95 % statistical confidence levels for a two-tailed Student’s t test

When the ENSO SST forcing is only considered in the tropical eastern Pacific and that in the tropical Indian Ocean and Atlantic is turned off (the TPO experiment), it is seen that the seasonal evolutions of the stratospheric responses are quite similar to that in the “ENSO” experiment, but the stratospheric polar vortex responses to warm ENSO are much stronger than that in the “ENSO” experiment (Fig. 4c, d vs. Fig. 4a, b). This proves once again that the ENSO-associated SST forcings in the tropical Indian Ocean partially cancel out the polar stratospheric responses to the tropical eastern Pacific ENSO SST forcing. The TIO experiment in Fig. 4e, f indeed induces a significant response of stratospheric polar cooling and polar jet strengthening throughout the warm ENSO winter. By comparing Fig. 4e, f with Fig. 4g, h, it is seen that the difference in responses to the tropical SST forcings between “ENSO” and TPO can be mainly attributed to the effect of the TIO SST forcing.

3.4 Downward propagations of the extratropical stratospheric signals

The pressure–time cross sections of the zonal-mean temperature and zonal wind anomalies were widely used (e.g., Manzini et al. 2006; Ineson and Scaife 2008; Zubiaurre and Calvo 2012; Rao et al. 2014) to illustrate the downward propagation of stratospheric anomaly signals (e.g., Kodera et al. 1990; Baldwin and Dunkerton 1999; Cai and Ren 2006, 2007; Ren and Cai 2007). Figure 5 shows seasonal evolutions of the polar cap (60°–90°N) temperature and the circumpolar (65°–75°N) westerly responses in all sensitivity experiments. The most conspicuous features in Fig. 5 are that the stratospheric temperature and zonal wind responses always lead that in the lower levels. Following the minor cooling and a strengthened stratospheric polar vortex in autumn, the stratospheric warming and easterly anomaly signals initiate since the early winter of warm ENSO years and gradually propagate downwards (Fig. 5a).
Fig. 5

Seasonal evolutions of the polar temperature (units: K, 60°–90°N, left panels) and the polar jet (units: m s−1, 65°–75°N, right panels) responses in the Northern Hemisphere in a, b ENSO, c, d TPO, e, f TIO and g, h the differences between ENSO and TPO. Contour interval is 0.5 K for temperature and 1.0 m s−1 for zonal wind. Light and dark shadings indicate the 90 and 95 % statistical confidence levels

Once the ENSO SST forcing is only confined to the tropical eastern Pacific (the TPO experiment), the seasonal evolutions of the extratropical responses are quite similar to that in the “ENSO” experiment, but the downward propagating signals are much stronger and more significant than that in the “ENSO” experiment (Fig. 5c, d vs. Fig. 5a, b). This confirms that the ENSO-associated SST forcings in the tropical Indian Ocean partially cancel out the polar stratospheric responses to the tropical eastern Pacific ENSO SST forcing. The TIO experiment in Fig. 5e, f indeed induces a downward propagating response, out-of-phase with that in the TPO experiment, throughout the ENSO winter. By comparing Fig. 5e, f with 5g, h, it once again assures that the difference between the “ENSO” and the TPO in response to the tropical SST forcings can be largely explained by the opposite effect of the TIO SST forcing.

4 Understanding the cancelling effect of the Indian Ocean on the stratospheric responses to the ENSO SST forcing from the tropical eastern Pacific

4.1 Tropical latent heating perturbations and tropical geopotential height anomalies

The diabatic heating causes divergence or convergence in the tropics, which is traditionally viewed as the Rossby wave source or sink (e.g., Hoskins and Karoly 1981; Hoskins et al. 1983; Sardeshmukh and Hoskins 1988; Kosaka and Nakamura 2006; Li et al. 2015; Zhao et al. 2015). The latent heating release affects the generation and subsequent poleward propagation of Rossby waves (Jin and Hoskins 1995). Based on this, we firstly analyzed the latent heating perturbation in the tropics and the related geopotential height response, which is not caused by the SST anomalies directly, but by the diabatic heating in the atmosphere. The DJF-mean latent heating disturbances averaged in the layer 1000–100 hPa (shadings) associated with different SST configurations are shown in Fig. 6. Relative to cold ENSO, the convective activities over the central tropical Pacific are significantly intensified during the warm ENSO winter, as indicated in Fig. 6c, though the strongest SST anomalies lie in the eastern equatorial Pacific. In pace with the intensification of the convective activities over the central tropical Pacific, convections over the tropical Indian Ocean are effectively suppressed. The east–west dipole structure of convection anomalies from the equatorial Indian Ocean to the central tropical Pacific suggests a large-scale circulation adjustment during the warm ENSO winter. In spite of considerations of the TIO warming, the local latent heating anomalies over the Indian Ocean are still mostly negative in the “ENSO” experiment, suggesting that ENSO SST forcings from the eastern tropical Pacific have a remote and dominant influence on tropical convections. The tropical latent heating perturbations and the 200-hPa height responses (contours) match quite well in the tropics. The strong positive height responses straddle the equator in the central–eastern tropical Pacific, just downstream of the areas covered with the positive latent heating anomalies.
Fig. 6

The DJF-mean climatology of the latent heating rate (shadings, units: K d−1) averaged in layer 1000–100 hPa, the zonal deviation of geopotential height (units: m, contours) at 200-hPa, in a the control experiment ClmSST; and height responses in c ENSO, e TPO, and g TIO; and i the difference of that between ENSO and TPO. Only the heating responses that are statistically significant at a 90 % confidence level for a two-tailed Student’s t test are shown. Displayed in b, d, f, h and j are the zonal-means of the geopotential height fields in a, c, e, g, i. Purple lines in the left panels mark the height anomalies above a 90 % confidence level

Once the tropical SST forcings are merely confined to the tropical Pacific, the positive height response over the tropical Indian Ocean in the TPO experiment becomes relatively weaker than that in the “ENSO” experiment. The relatively weaker height response in the TPO experiment may be caused by the absence of the TIO SST forcings. When the SST forcings are only considered in the tropical Indian Ocean, as shown in Fig. 6g, the maximum latent heating perturbations lie over the tropical Indian Ocean. However, the latent heating perturbations over the tropical Pacific and Atlantic oceans are quite weak and insignificant and therefore ignorable, suggesting that the ENSO-related TIO SST forcing has a relatively local effect on the tropical convections. The contributions of the TIO SST forcing to the local positive height response can been clearly seen from Fig. 6g and further verified by the response difference between the “ENSO” and TPO experiment, with the positive height responses at 200 hPa over the tropical Indian Ocean (Fig. 6i). Therefore, the positive TIO SST anomalies following the maturing warm ENSO favor the zonally symmetric air warming and thus the positive height response at 200 hPa in the tropics.

From the above analysis, it can also be inferred that the positive height responses always straddle the nearby equatorial latent heating regions. A pair of positive height centers are located off the equator and distributed in both hemispheres, which is a typical forced Rossby wave response. There also exists a belt of positive height responses eastward of the heating sources along the equator, as a typical forced equator-trapped Kelvin wave response (Gill 1980), leading to the zonally symmetric warming in the tropical troposphere.

4.2 Extratropical circulation responses to tropical heating anomalies

Accompanied with the positive tropical height responses in the upper troposphere, anomalous cyclones are excited poleward. The extratropical circulation responses to the tropical latent heating over the TIO and TPO are quite different, especially in the Pacific–North America pattern. Specifically, a canonical PNA-like spatial pattern was reproduced in the “ENSO” and the TPO experiments with an anomalous low located over the Aleutian Islands (a deepened Aleutian low) and an anomalous high centered over Canada–Greenland (an intensified Canadian Arctic high), as indicated in Fig. 6c, e. The most obvious difference between the “ENSO” and the TPO experiments is that the positive PNA-like response in Fig. 6e seems to be stronger than that in Fig. 6c. Nevertheless, when the ENSO-related forcing is simply confined to the tropical Indian Ocean, an opposite spatial pattern is reproduced in the TIO experiment, with an anomalous high over the North Pacific (a weakened Aleutian low) and an anomalous low over Canada–Greenland (a weakened Canadian Arctic high), which together project onto the negative phase of PNA, as shown in Fig. 6g. This agrees with the results of Hoerling et al. (2004) and Annamalai et al. (2007), and can be further verified by the difference between the “ENSO” and the TPO experiments in Fig. 6i.

Apart from the waveform responses, annular mode responses can also be found in the sensitivity experiments. In the “ENSO” and especially the TPO experiments, from the southern United States and Mexico to the North Atlantic and well into Western Europe, a belt of highly significant low-pressure anomalies is a very prominent feature of the upper tropospheric geopotential height anomaly pattern. The positive upper tropospheric geopotential height anomaly center appears over South Greenland and the Davis Strait, also covers large parts of Canada and the North Pole, and it extends to the east towards Iceland, Spitsbergen and the Russian Arctic (Fig. 6c, e). The high-latitude positive and mid-latitude negative geopotential height anomalies clearly project onto a strong negative phase of the northern annular mode (NAM) in Fig. 6d, f, with the latter’s amplitude greater than the former’s. This is consistent with the results of Fletcher and Kushner (2011) and Graf et al. (2014). By contrast, the TIO experiment yields a negative geopotential height response from the Canadian Arctic and the Davis Strait to Greenland and Iceland and well into Svalbard and the Russian Arctic, and a positive geopotential response from the southern United States and Mexico to the North Atlantic and well into Southwestern Europe as indicated in Fig. 6g, which together corresponds to a positive NAM response in Fig. 6h. This is also proved by the difference between the “ENSO” and TPO experiments in Fig. 6j. The response to the TIO and that to the TPO forcing exhibit similar patterns but are opposite signed. This is mainly attributed to the fact that the thermodynamic forcing is shifted by 180° longitude. In other words, since the diabatic heating is shifted by 180° longitude, the extratropical wave is also shifted by 180° in the zonal direction (The SST anomaly center in the TIO is just ~180 degrees longitude from that in the TPO). A PNA is mainly projected to the planetary wavenumber-1 in the upper troposphere, which can propagate into the stratosphere. If the zonal wavenumber-1 is shifted by 180 degrees in the zonal direction, the phase is thus changed to the opposite polarity. Since positive/negative NAM corresponds to the westerly/easterly anomalies in the circumpolar regions, the NAM response can be connected with the downward propagating signals from upper levels (see Fig. 5), while the anomalous PNA-like response is obviously a manifestation of the stationary Rossby wave responses.

4.3 Extratropical planetary wave activities

Next, the extratropical planetary wave activities are analyzed. The zonal deviation of the geopotential height in the extratropics is mainly dominated by the WN 1–2, especially in the upper levels. The zonal deviation and the WN-1 and WN-2 components of the 200-hPa DJF-mean geopotential height responses to SST forcings in different topical oceans are displayed in Fig. 7. To further investigate the extratropical waves’ behaviors in the vertical direction, longitude–height cross sections of the zonal deviation and the WN-1 and WN-2 components of the DJF geopotential height responses in the northern extratropics (45–75ºN) to different tropical oceans’ forcings are displayed in Fig. 8. Shadings in Figs. 7 and 8 show the corresponding DJF-mean climatology from the control experiment ClmSST.
Fig. 7

Composite responses (contours, units: m) of the zonal deviation (left panels) and the wavenumber-1 (middle panels) and wavenumber-2 (right panels) components of 200-hPa geopotential height (units: m) in DJF in ac ENSO and df TPO, gi TIO and jl the difference of that between ENSO and TPO. Shadings show the DJF-mean climatology (units: m) from the control experiment ClmSST

Fig. 8

Longitude–height cross sections of the composite responses (contours, units: m) of the zonal deviation (left panels) and the wavenumber-1 (middle panels) and wavenumber-2 (right panels) components of the DJF geopotential height (units: m) in the northern extratropics (45–75°N), in ac ENSO and df TPO, gi TIO and jl the differences of that between ENSO and TPO. Shadings show the DJF-mean climatology (units: m) from the control experiment ClmSST

4.3.1 Different behaviors of WN-1 and WN-2

In the “ENSO” and the TPO experiments, an anomalous low response over North Pacific and anomalous high response over Canada were reproduced, as indicated in Fig. 7a, d, which is consistent with Fig. 6c, e. The extratropical anomalous response centers amplify the climatological stationary waves (i.e., an anomalous low response superposed on the Aleutian low, and an anomalous high response superposed on the Canadian continental high). From Fig. 7b, e, it can be seen that the strengthening climatological waves in the upper troposphere in the “ENSO” and TPO experiments are dominanted by the contributions of WN-1. On the contrary, the WN-2 response centers are nearly out of phase with the climatological WN-2 in Fig. 7c, f.

Conversely, in the TIO experiment, an anomalous high response over North Pacific and an anomalous low response over Canada were produced (see Fig. 7g), which is consistent with Fig. 6g. The extratropical anomalous response centers weaken the climatological stationary wave (i.e., an anomalous high response superposed on the Aleutian low, and an anomalous low response superposed on the Canadian continental high). The response in TIO is obviously opposite to the TPO and therefore weakens the ENSO signal in the extratropics. The depressed climatological waves in the extratropics in TIO are partially from the contributions of WN-1 in Fig. 7h associated with the negative PNA pattern in Fig. 6g, and partially from WN-2 in Fig. 7i, which will be further proved below by analyzing the E–P flux. Different from the competitive relationships between the zonal WN-1 and WN-2 responses in the “ENSO” and “TPO” experiments, the WN-1 and WN-2 responses to the TIO forcing coordinate with each other to weaken the climatological stationary waves. Such characteristic in the TIO experiment can also be found in the difference between the “ENSO” and the TPO experiment shown in Fig. 7j, l.

Figure 8 shows that the zonal deviation and wave responses generally exhibit a westward tilt and their amplitudes are generally amplified with height increasing. Such a structure corresponds to an upward propagation of Rossby waves, and will contribute to a positive eddy heat flux response in the “ENSO” and the TPO experiments in Fig. 8a, f. It is the opposite in the TIO experiment in Fig. 8g, i, and therefore in the difference between the “ENSO” and the TPO experiments in Fig. 8j, l.

It can be inferred that the contrasting wave responses between the TIO and TPO forcings are caused not only by WN-1, but also by WN-2. While the anomalous WN-1 effects oppositely to the projected positive (for the “ENSO” and the TPO experiments) and negative (for the TIO experiment) PNA response in the upper troposphere, the WN-2 response is always out of phase with the climatological WN-2.

4.3.2 E–P flux and its divergence

The 3-D E–P flux is a convenient and useful tool for stationary wave diagnostics because it is parallel to the group velocity (Plumb 1985). Figure 9 shows the 3-D E–P flux responses to the different tropical SST forcings at 200 hPa. The most dominant feature is the anomalous northeastward propagation of stationary waves around the Pacific–North America in the “ENSO” and the TPO experiments in Fig. 9a, b, and the opposite in the TIO and the “ENSO” minus TPO in Fig. 9c, d. The anomalous upward propagation of stationary waves is observed mainly over the western coast of North America in the “ENSO” and the TPO experiments (Fig. 9a, b), whereas the upward propagation of stationary waves is depressed there in the TIO and the “ENSO” minus TPO experiments (Fig. 9c, d).
Fig. 9

Composite responses of the DJF-mean wave flux at 200-hPa in a ENSO, b TPO, c TIO and d the differences of that between ENSO and TPO. Vectors and shadings respectively indicate the horizontal (units: m2 s−2) and the vertical (units: 10−2 m2 s−2) component

Since the E–P flux is parallel to the group speed of Rossby wave, an increase/decrease in the upward wave flux in the stratosphere means more/less planetary waves from the troposphere. Rossby wave propagation may be further examined by using the 2-D E–P flux cross sections. Figure 10 shows latitude–height cross sections of the E–P flux response and its divergence in the sensitivity experiments. The E–P flux vectors are scaled by the inverse of the air density at the given level so that the wave propagation in the stratosphere can be shown clearly. For the “ENSO” and the TPO experiments, there is a surplus of upward wave fluxes and strong convergences of the E–P flux in the upper stratosphere (Fig. 10a, b). Conversely, associated with the stronger polar vortex observed in the TIO experiment and the “ENSO” minus TPO, the downward E–P flux anomaly and strong divergence of the E–P fluxes imply suppressed wave activity (Fig. 10c, d).
Fig. 10

Latitude–height cross sections of the composite responses of E–P flux (vectors; normalized by air density; units: m2 s−2) and its divergence (shading, units: m s−1 d−1) in a ENSO, b TPO, c TIO, and d the differences of that between ENSO and TPO. The E–P vector has been normalized by density and the vertical component has been multiplied by 100 to facilitate a clear plotting. Contours are the corresponding zonal-mean zonal wind responses (units: m s−1)

The stratospheric polar vortex intensity has been closely linked to the upward wave activity entering the stratosphere. It has been shown that eddy heat flux is anomalously positive/negative for weaker/stronger polar vortex (e.g., Christiansen 2001; Polvani and Waugh 2004; Li and Lau 2013). To measure the upward propagation of planetary waves entering the stratosphere, the anomalous vertical component of the E–P flux, Fz at 50 hPa and in 45°–75°N, is calculated. Table 2 displays the values of Fz for WN-1 and WN-2 in each case. For the “ENSO” and TPO experiments, the magnitude of the Fz associated with WN-1/WN-2 is anomalously positive/negative. Note that in the TPO experiment, the Fz associated with WN-2 (−4.02 × 103 kg s−2) is nearly half of that associated with WN-1 (8.91 × 103 kg s−2), which is also true for the “ENSO” run above 50 hPa (see Fig. 10a for net effect). It can be inferred that the net weakening effect on the stratospheric polar vortex during El Niño winters is dominantly caused by the extratropical WN-1 response, which overwhelms the decrease in WN-2. The results in the “ENSO” and the TPO experiments agree with the result of Garfinkel and Hartmann (2007, 2008). In contrast, for the TIO experiment and the “ENSO”minus TPO difference, it can be easily found that the Fz associated with both WN-1 and WN-2 is always anomalously negative. Though WN-2 component of Fz at 50 hPa is larger, the WN-1 component of Fz is never negligible, since the WN-1 Fz response is about half of that of WN-2 to the TIO forcing.
Table 2

Vertical component of E–P fluxes (Fz; units: 103 kg s−2) at 50 hPa averaged in 45°–75°N for total zonal waves and WN 1–2 for each experiment

 

WN-total

WN-1

WN-2

ENSO

−1.17

7.93

−10.62

TPO

5.47

8.91

−4.02

ENSO–TPO

−6.30

−0.98

−6.60

TIO

−8.56

−3.09

−5.68

5 Summary and discussion

5.1 Summary

Since a developing tropical IOB warming event initiating from winter generally follows a maturing El Niño event, many previous studies based on reanalysis and model outputs have also included the TIO signal when they checked the stratospheric polar vortex response to ENSO events. Different from them, this study applies the WACCM, a stratosphere-resolved GCM model to dissect the stratospheric responses in the northern winter extratropics to the imposed ENSO-related SSTA in different tropical ocean basins. The teleconnections between the TPO SST and the stratospheric polar vortex intensity and between the TIO SST and the stratospheric polar vortex intensity are examined in different sensitivity experiments.

WACCM not only reproduced quite well the climatology of the stratospheric circulations, including the stratospheric polar vortex intensity, but also the interannual variability of the northern winter stratospheric polar vortex. When forced with the ENSO-related SST in the TPO, the stratospheric polar vortex is anomalously warmer and weaker than normal, consistent with the results in many previous studies cited in this article; conversely, when forced with the ENSO-related SST merely in the TIO, the stratospheric polar vortex is anomalously colder and stronger than climatology. Specifically, warm forcing in TPO can enhance the climatological WN-1 pattern and decline the climatological stationary WN-2 pattern in the extratropics, with the increase of the former overwhelming the decrease of the latter. The increase in WN-1 is mainly the result of the projected positive PNA response in the extratropics with the Aleutian low deepened and the Canadian continental high intensified. By contrast, the warm forcing in the TIO weakens both the climatological WN-1 and WN-2 patterns and induces a negative PNA response with the Aleutian low weakened and the Canadian continental high undermined. Diagnosis on E–P flux confirms that, the anomalously upward propagation of stationary wave in the extratropics mainly lies over the western coast of North America in the TPO experiment, while the TIO warming suppresses the upward propagation of stationary waves there.

This paper proves that SSTA in the TIO modulate the influence of ENSO on the extratropical circulation. Previous studies have also noted the opposite-signed zonal-mean responses to SSTA over the TPO and the TIO (e.g., Branstator and Haupt 1998; Barsugli and Sardeshmukh 2002; Annamalai et al. 2007; Fletcher and Kushner 2011). Specifically, a positive-PNA and negative-NAM response corresponds to an imposed warming in the tropical Pacific, whereas a negative-PNA and positive-NAM signal is forced by an imposed warming in the tropical Indian Ocean. Our study further indicated a strong in-phase/out-of-phase relationship between WN-1/WN-2 response and the background stationary WN-1/WN-2 counterpart in the TPO, which is also consistent with the result of Fletcher and Kushner (2011). We also found an out-of-phase relationship between WN-1 (or WN-2) response and their background stationary counterpart for imposed warming in the TIO.

5.2 Discussion

A weaker stratospheric polar vortex is expected in the El Niño winter. Due to the atmospheric chaos, a stronger polar vortex and an El Niño event may also coexist in observations. It needs to be further studied whether the decrease in WN-2 for each case (e.g., an imposed warming in the TPO, in the TIO, or in both) and the decrease in WN-1 for imposed warming in the TIO can partially explain the concurrence of an El Niño event and a stronger stratospheric polar vortex event in the observations. Few previous studies have given reasonable explanations for this. Much attention has been paid to the general influence of the tropical Pacific ENSO SST on the northern winter stratospheric polar vortex (i.e., El Niño/La Niña and weaker/stronger stratospheric polar vortex), with less attention on the warming in other ocean basins. It would be worth exploring the relative importance of the warming over the two ocean basins to the relationship between ENSO and the stratospheric polar vortex, as also suggested by Li and Lau (2013).

Toniazzo and Scaife (2006) confirmed the statistically significant influence of ENSO on winter North Atlantic climate by the stratosphere pathway, which can also be seen in our “ENSO” and TPO experiments (negative NAM/NAO response in Fig. 6). However, the northern stratospheric response to the ENSO-related SST forcing in the TAO is rather weak in the stratosphere. The weak modulation of the ENSO-related SST forcing in the TAO on the northern winter stratospheric polar vortex may be attributed to the weak SST anomalies in our TAO experiment. The role of the TAO SST forcing on the northern winter circulation has been reported in literature. For example, results from linear models (Li et al. 2007) and AGCMs (Mechoso and Lyons 1988; Watanabe and Kimoto 1999; Robertson et al. 2000; Peng et al. 2005; Sutton and Hodson 2007; Wu et al. 2007; Wang et al. 2008) suggested that a positive TAO SST anomaly tend to produce a south–north dipole in geopotential heights, also much like the NAO. Since the tropical Atlantic warming develops even later than the tropical IOB warming, and SSTA in the tropical Atlantic are still very weak in northern winter when El Niño has matured (Fig. 1), signals are seemingly insignificant and feeble for the imposed weak warming in the tropical Atlantic (Fig. 3g, h). It can also be verified by the similar features between the “ENSO” minus TPO and the TIO experiment (i.e., ENSO–TPO ≈ TIO), though the SSTA in the tropical Atlantic were included in the “ENSO” experiment. Due to the limited scope of the article, the influence of the tropical Atlantic conditions needs to be further studied.

Since the TIO experiment uses specified SST anomalies, there is possibility that ocean forcing in the atmosphere may be misleaded to some degrees. The TIO experiment may deviate from the realistic world where the TIO SST anomalies are forced remotely by ENSO and are passive responses. For example, Copsey et al. (2006) used the HadAM3 model and found that the negative sea level pressure (or positive rainfall) trend overlies the warming SST trend in the Indian Ocean, which is inconsistent with observations. It necessitates further investigations with fully coupled models in the future. As is known to the scientific community, on the longer timescale, the tropical Indian Ocean is getting warmer with a growth rate different from that in other ocean basins in recent decades, whereas the eastern equatorial Pacific is cooling. At the same time, the northern winter stratospheric polar vortex shows a cooling trend and the stratospheric NAM is also strengthened toward its positive phase (Ren and Yang 2012; Rao et al. 2015). It is worthwhile to study whether the mechanism found in this article can provide an explanation for the concurrence of the two phenomena. Also, coupled models tend to reproduce a colder eastern tropical Pacific SST pattern (Wei et al. 2007) and therefore a stronger northern winter stratospheric polar vortex than that in observations. Given the fact that the TIO SST seems to be more efficient (1-K warming in the TIO induced stronger response in the stratosphere than that in the TPO) in modulation of the northern winter stratospheric polar vortex intensity, our findings may be used to understand model biases in reproducing the stratospheric polar vortex by attributing them to the TPO or the TIO SST biases, which obviously can help model developers to improve model’s reproducibility of SST and the stratospheric polar vortex.

Acknowledgments

This work are jointly supported by research grant from the National Natural Science Foundation of China (41575041, 41430533 and 91437105), a Chinese Academy of Sciences project (Grant No. XDA11010402) and a China Meteorological Administration Special Public Welfare Research Fund (Grant No. GYHY201406001). The authors thank the reviewers and editors for their helpful comments and kind suggestions. We acknowledge the UK Met Office providing HadISST dataset. We also thank the NCAR providing the WACCM model.

Copyright information

© Springer-Verlag Berlin Heidelberg 2015

Authors and Affiliations

  1. 1.State Key Laboratory of Numerical Modeling for Atmospheric Sciences and Geophysical Fluid Dynamics, Institute of Atmospheric PhysicsChinese Academy of SciencesBeijingChina
  2. 2.Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters and KLMENanjing University of Information Science and TechnologyNanjingChina
  3. 3.University of Chinese Academy of SciencesBeijingChina

Personalised recommendations