Characterizing Atlantic interhemispheric teleconnection established by South American monsoon in austral summer

This study aims to characterize the interhemispheric teleconnection pattern, which is established by the South America (SA) summer monsoon over the Atlantic Ocean during January and February, and referred it as the Atlantic symmetric pattern (ASP). The ASP is characterized using the leading mode of interannual sea surface temperature (SST) variability of the Southwest Atlantic Ocean, where strong convection-SST coupling occurs. The pattern is manifested as two anomalous cyclonic-anticyclonic-cyclonic circulation trains aligned meridionally over the Atlantic Ocean, with a distinct SST dipole of the Southwest Atlantic Ocean and a tripole of the North Atlantic Ocean. The interhemispheric wave trains of the ASP are excited as a Gill-type response to convective activity in the SA summer monsoon, as confirmed in linear baroclinic model. Complementing previous studies on observed interhemispheric connection in the Atlantic, our findings highlight the importance of characterizing the ASP and its role in linking the South Atlantic SST, the SA summer monsoon, and North Atlantic climate. Further research is warranted to explore the impacts of the ASP on the Northern Hemisphere and its interactions with other climatic modes.


Introduction
Variations in sea surface temperature (SST) over the Atlantic Ocean are known to be actively coupled with interannual to decadal large-scale atmospheric patterns and have notable influence on synoptic and seasonal variability (Czaja and Frankignoul 1999, Haarsma et al 2003, Robertson et al 2003, Barreiro et al 2005, Nnamchi et al 2011, Kayano et al 2013, Chen et al 2015, Tirabassi et al 2015. In particular, SST variation over the South Atlantic Ocean is closely coupled with the seasonal convective systems, especially the South America (SA) monsoon system and the South Atlantic convergence zone (SACZ) (Chaves andNobre 2004, Jorgetti et al 2014). Studies have found that the mutual interactions between SA monsoon circulations, convection activities, and air-sea exchange can greatly control rainfall and SST patterns over the South Atlantic (Nobre et al 2009).
Locally, the SA monsoon system and SACZ have the effects on SST over the Southwest Atlantic region through compensating subsidence and shading radiative effect of SACZ (Gandu and Silva Dias 1998) and a local cyclonic eddy of Rossby waves triggered by enhanced SACZ Mechoso 2000, Carvalho et al 2004). Remotely, the SA monsoon rainfall anomaly is found to influence South Hemispheric regions including southern Indian Ocean (DeBlander and Shaman 2017), and southeast Australia (Xue et al 2018), and South Africa (Grimm and Reason 2011) through Rossby waves traveling through southern midlatitudes. Furthermore, observed correlation and model experiments both suggest an interhemispheric connection between the North Atlantic circulations and variability of SST over the South Atlantic Ocean (Robertson et al , 2003. Convective heating over the SA monsoon region is found to cause interhemispheric energy transport into the North Atlantic and affects Eurasia and further into East Asia through eastward propagating Rossby waves (Grimm and Silva Dias 1995). While the interhemispheric teleconnection between the South Atlantic and the North Atlantic is well recognized, however, its structure, SST coupling, and remote impacts remain to be quantitatively investigated. In addition to the major climate variations and cross-hemisphere connections discussed earlier, it is essential to consider the global inter-hemispheric dipole mode of SST. This mode has been suggested to significantly impact teleconnections in both the Northern and Southern Hemisphere atmospheric circulation, and contribute to the global ocean northward cross-equatorial heat transport (Sun et al 2013).
This study focuses on characterizing the interhemispheric pattern established by the SA monsoon system with a leading SST mode over the Southwest Atlantic Ocean, given the strong coupling between southwestern Atlantic SST and convective heating. Consistent with previous studies, this teleconnection pattern, referred to as the Atlantic symmetric pattern (ASP), is composed of a wave train-like circulation structure that extends over the Atlantic Ocean in both hemispheres. The near-surface cyclonic and anticyclonic circulation of the Rossbywave-type responses modulates the SST in the both South and North Atlantic regions. We also conducted Linear Baroclinic Model (LBM) experiments, in addition to the observed regressed structure, to confirm that the ASP pattern is likely a Rossby-wave-type response to the convective activity of the SA monsoon and SACZ (Tseng 2022). The present study comprehensively investigated the ASP, including its interhemispheric atmospheric structure, its SST signature, and its effects in the Northern Hemisphere. Section 2 provides a description of the model and methodology followed in the study. Section 3 presents the results in both observations and LBM experiments, and section 4 presents a discussion of the findings. This paper concludes with section 5, emphasizing the significance of characterizing the ASP and its role in interhemispheric teleconnection.

Materials and methods
We used the monthly Extended Reconstructed SST (ERSSTv5) data on a 2 • × 2 • grid from 1870 to 2018 provided by the Physical Sciences Laboratory of the National Oceanic (Huang et al 2017). We also used global precipitation climatology project (GPCP) rainfall data over the period ranging from 1980 to 2018 (GPCP-SG_L3_v2.3;Huffman et al 1997, Adler et al 2003. The atmospheric variables and surface fluxes were obtained from the fifth generation of the European Centre for Medium-Range Weather Forecasts (ERA5, Hersbach et al 2020), which was on a 0.25 • × 0.25 • grid from the period between 1980 and 2018. Our analysis used average daily values calculated according to the original hourly data obtained from the archive of the Copernicus climate change service.
To determine the forcing that induces the ASP, we adopted an idealized LBM developed at the University of Tokyo. It is a linearized version of a primitive equation representing a specified base state with a T21 resolution (∼5 • ) horizontal and 15 Sigma-pressure model vertical levels (Watanabe and Kimoto 2000, Watanabe and Jin 2003. Compared with the full climate models, the LBM provides a more accessible framework to interpret the dynamic responses to convective forcing and underpins several studies regarding the dynamic linkage between tropical climate systems and remote teleconnection systems (Annamalai 2010, Yasui and Watanabe 2010, Hayashi and Watanabe 2017. For the base state of the LBM, we used the climatological state of ERA5 for January and February during 1980-2019. The convective forces applied in the LBM are derived from the regressed diabatic cooling calculated with ERA5 reanalysis. The diabatic cooling is distributed across all 15 layers in the vertical profile. For example, the forcing at SA monsoon region is characterized by a profile with maximum of −0.7 K day −1 at 500 hPa level, representing a cooling anomaly associated with suppressed deep convection (shown in figure 2).

Definition and characteristics of ASP
EOF analysis is extensively used to identify and extract spatiotemporal modes in climate analysis; however, it tends to highlight the global variability because of orthogonality constraints (Preisendorfer et al 1988, Mestas-Nuñez 2000. To address this, we used the rotated empirical orthogonal function analysis (REOF), which enables the identification of robust patterns describing local variability by relaxing the temporal orthogonality constraint through rotation transformation to group local variability (Mestas-Nuñez 2000). REOF analysis has been shown to be highly accurate and effective, particularly in identifying localized patterns (Lian and Chen 2012). The varimax rotation method was applied to emphasize regional SST variability. The first rotated EOF has a lower explained variance the first (unrotated) EOF, but it better represents local extremes.
We applied the REOF analysis to derive the dominant mode of SST over the South Atlantic Ocean (80 • S-EQ, 70 • W-30 • E; figure 1(a)) for January and February (JF) and identify the principal component of first mode (RPC1) as the index of the ASP. We observed a meridional dipolar pattern with a basinwide warm anomaly at 38 • S-20 • S that extended from Argentina to South Africa and a cold SST anomaly within 70 • S-38 • S from Patagonian Coast to central South Atlantic Ocean. The two regions are aligned from northeast to southwest, with larger amplitudes over the western boundary. Spectrum analysis of the RPC1 time series revealed spectral peaks near 3 years (figures 1(b) and (c)). Repeated the analysis with the SST from 1960 to 2019, a similar pattern with 16% explained variance, and a weak but marginally significant 3 year peak (figure 1(c)). Therefore, we conducted the following analysis from 1980 to 2019 using reliable satellite data obtained after 1979.

Atmospheric and SST structure of the ASP
The atmospheric circulation and sea surface characteristics of the ASP, represented by regressed fields onto RPC1, are illustrated in figure 2. The regressed geopotential height at 200 hPa and 850 hPa displayed in figures 2(a) and (b) show a symmetric structure of the ASP relative to the Equatorial Atlantic. The ASP was named so because of this interhemispheric pattern of the symmetric circulation structure. The cyclonic-anticyclonic-cyclonic wave train structure was observed to be meridionally aligned from the Equator toward both poles along the west coast of the Atlantic basin (figure 2(a)). Unlike the two prominent cyclonic circulations off-Equator in the tropics at 200 hPa, a minimal signal was obtained below 300 hPa (figures 2(b) and S1(a)). In the extratropical Atlantic Ocean outside 30 • N and 30 • S, wavelike perturbations with a barotropic structure extended throughout the troposphere to near-surface from the tropics to the poles (figures 2(b) and S1(b)). In the zonal direction, the Atlantic perturbations were associated with wave-like structures in extra tropics of both Southern and Northern Hemispheres embedded in the mid-latitude jet streams (indicated by pink lines in figure 3(a)).
Compared with the positive-negative-positive SST pattern in the North Atlantic Ocean, the SST tripolar pattern was negative-positive-negative in the South Atlantic Ocean. The SST anomaly associated with the ASP was considerably more apparent in the mid-latitude than in the tropics. Outside the Atlantic region, a warm SST anomaly was observed in the northeastern Pacific along the North American west coast. By contrast, signals in the Indian Ocean and the west Pacific Ocean were considerably weak. The regressed net heat flux pattern associated with the ASP indicated the active role of the atmosphere in the ocean-atmosphere coupling (figures 2(c) and (d)). Notably, the downward (positive) net heat flux was well-collocated with a warm SST anomaly and vice versa in the mid-latitudinal North and South Atlantic regions. This in-phase relationship between the surface heat flux and SST was also observed from December to January, suggesting that atmospheric processes modulate the SST variation in the midlatitudinal Atlantic Ocean.
The wave activity flux vectors (Takaya and Nakamura 2001) in figure 3(a) indicated the propagation of wave energy from the Southern Hemisphere, crossing the Equatorial Atlantic, to the North Atlantic region. The divergence of wave activity over the northeastern Brazilian region suggested the presence of a local energy source in the region. After spreading to the Northern Hemisphere, wave activity weakened  Lin 1992, Ambrizzi et al 1995). The wave progresses towards higher latitudes, it gradually changes direction due to the influence of the background flow. Eventually, when the wave reaches near the jet stream region, it undergoes a transition and propagates predominantly in the zonal (east-west) direction due to the waveguide effect of the jet stream (Karoly 1983, Hsu and Lin 1992, Ambrizzi et al 1995. This change in direction is clearly depicted in the wave activity fluxes presented in figure 3(a), which demonstrates the energy propagation associated with the Rossby wave-like perturbations. This is in line with the equatorial westerly duct effect Holton 1982, Webster andChang 1998). The extratropical perturbations in the North Hemisphere were more active than those in the Southern Hemisphere because of the boreal winter's stronger meridional gradient of potential vorticity, which was favorable for the energy propagation of stationary Rossby waves. Such a gradient contrast was noted in the JF climatology of zonal mean streamfunction wherein the gradient was stronger in the winter hemisphere ( figure 3(b); shading).

Remote influence of ASP on near-surface temperature and rainfall
The regressed 2 m air temperature (T2m) and precipitation illustrated in figures 4(a) and (b) indicated the influence of the ASP on interannual variability in remote regions through teleconnection. Both T2m and precipitation were detrended before conducting regression analysis. The primary T2m anomalies in the Southern Hemisphere were observed over the southwest Atlantic Ocean, indicating the atmospheric counterpart of the SST dipole signature of the ASP (figure 4(a)). The T2m was more anomalous in the Northern Hemisphere than in the Southern Hemisphere. The positive temperature anomaly was noted over western North America, northern Eurasia, Siberia, and the Arctic near the Barents and Kara Seas. Negative but weaker temperature anomalies were noted in northeastern North America, North Africa, and Central Asia ( figure 4(a)). The positive-negative-positive T2m anomalies from North America to Siberia resembled Rossby-wavelike perturbations ( figure 4(a)).
Globally, the most prominent precipitation feature is the east-west dipole over northern SA, indicating a dry anomaly extending southeastward from the The latitude-height cross-section of the streamfunction regression coefficient (shading: m 2 s −1 ) averaged within 60 • W-10 • W. The contours represent the JF climatology in the globally averaged mass streamfunction (10 −8 kg s −1 ), and the contour labels are-5,-3,-1, 1, 3, 5, and 8. east coast of Brazil to the Atlantic Ocean and a wet anomaly in the west parallel to the dry anomaly over the SA monsoon system and SACZ ( figure 4(b)). The precipitation anomaly in the Northern Hemisphere exhibits a dry-wet-dry pattern aligned from west to east over southeastern North America, western Atlantic Ocean, and eastern Atlantic Ocean ( figure 4(b)). The SA monsoon system and SACZ are characterized by intense convective activity during JF and are thus effective in triggering the ASP. Moreover, the correlation between Amazon rainfall (30 • S-0 • , 20 • W-55 • W; figure 4(b), black box) and the RPC1 is-0.51, which is statistically significant at 5% (figure 4(e)). This strong correlation indicates a relationship between the convective activity of the SA monsoon systems and the ASP.
Furthermore, to evaluate the robustness of the teleconnections, we conducted regression maps of T2m and precipitation anomalies from ERA5 over the long period of 1960-2020 (figures 4(c) and (d)). The T2m was notable over the southern and northern Atlantic Ocean; the zonal dipole over North America was also notable (figure 4(c)). The correlation over Eurasia, Siberia, and the Arctic was weak, indicating more uncertainty of the teleconnection. For rainfall, analysis of data for a longer period revealed the influence of the ASP on most regions, including the SA monsoon system, SACZ, and North Atlantic Ocean, identified using data obtained for shorter periods ( figure 4(d)). Notably, the wet anomaly over Southern Europe and the dry anomaly over southwest China were strongly correlated with the sea level pressure anomaly in figure S1(b).

SA monsoon system as a forcing triggering the ASP
To examine the robustness of ASP to represent the SA monsoon induced pattern, we regressed the daily 200 hPa streamfunction in JF to the SA summer monsoon rainfall time series, including east Amazonia and SACZ (black box in figure 4(b)). Compared with the ASP-regressed streamfunction ( figure 3(a)),  figure 4(b)). The correlation between the two for 1960-2019 is-0.51. figure 5 shows a very similar structure of wave-train anomalous flow aligned meridionally. However, little signal is found over the west coast of North America compared with figure 3(a), suggesting the anomaly there may be a downstream effect of the teleconnection, not directly related to variability of the SA monsoon rainfall.
To confirm the SA monsoon as a forcing that triggers the ASP, we used the LBM for experiments with forcing assigned to different regions based on the regressed rainfall anomaly map ( figure 4(b)). The simulated circulation patterns were compared with the observed patterns of the ASP, especially focusing on the two Rossby-wave-like patterns that extend poleward over the South and North Atlantic Ocean ( figure 2(a)). We selected the forcing locations based on the precipitation regression of the RPC1 ( figure 4(b)). In the regression maps, the forcing triggering the ASP was most likely the pair of strong dry and wet patches over northern SA and further extending to the South Atlantic Ocean ( figure 4(b)). Regarding the rainfall climatology in austral summer, such wet and dry patches indicated a substantial rain surplus-deficit anomaly pair of the SA monsoon and the SACZ often associated with the ASP. Figure 6(a) illustrated the steady state when the diabatic cooling force is applied to emulate the cooling forcing representing deficit convective anomalies over east Brazil and the oceanic region of SACZ. At the 200 hPa level, the symmetrical negativepositive-negative wave train pattern occurred over the North Atlantic Ocean and was aligned in the meridional direction (figure 6(a)), similar to the observed pattern ( figure 4(b)). The LBM also captured the higher magnitude of the wave train in the Northern Hemisphere due to the stronger vorticity gradient in the winter hemisphere ( figure 6(a)). The results of this study were consistent with the teleconnection patterns reported in previous studies Silva Dias 1995, Grimm andReason 2015) wherein two Rossby-wave branches moved toward both hemispheres.
Three additional experiments were conducted to investigate the remote atmospheric responses to  the tropical heating over the eastern Pacific (EP), western Pacific (WP), and tropical Atlantic (TA) regions, where anomalous tropical convection was observed. In the EP experiment, the tropical cooling over the Pacific triggered two Rossby-wave-like responses emanating poleward in the Northern and Southern Hemisphere ( figure 6(b)). The southern branch that propagated to the South Atlantic Ocean demonstrated its association with ENSO (Nnamchi et al 2011, Rodrigues et al 2015. However, the responses over the North Atlantic and the Northern Hemisphere were different from those associated with the ASP. Similarly, in the WP experiment, the major Rossby wave responses in the Northern Hemisphere did not reproduce the observed pattern ( figure 6(c)). In the TA experiment, the major response was a weak wave-like pattern spread over the North Atlantic region ( figure 6(d)). In addition, the cooling failed to induce wave-like perturbations over the South Atlantic. Therefore, cooling over the tropical South Atlantic region had a minor influence compared with the cooling in the SA monsoon and SACZ.
To further elucidate the generation and propagation of the simulated wave-like perturbations with the SA monsoon forcing, the responses on days 1, 4, 7, and 11 were recorded (figure S3). On day 4, two low-pressure responses centered at 30 • S and 5 • N were formed in the Atlantic region on day 4 ( figure  S3(b)). The two negative anomalies along the Equator and an elongated negative anomaly spreading eastward along the Equator was similar to a Gill-type response to convective cooling (Gill 1980). A cyclonic anomaly (near 30 • S) coinciding with cooling was accompanied by a positive anomaly spreading in the zone between the Equator and 30 • S. While the perturbations in the Southern Hemisphere continued to intensify but with a limited increase over the following days, wave-like perturbations similar to the observed pattern developed over the North Atlantic and started spreading eastward near the jet stream (figures S3(c) and (d)). This simulation demonstrated how the cooling in the SA monsoon and SACZ induced perturbation in the Southern Hemisphere and a northward-propagating wave-like perturbation crossing the westerly duct over Equatorial Atlantic into the Northern Hemisphere (Webster and Holton 1982).

Discussion
Based on the regressed structure and numerical experiments, the structure of ASP and its SST signature is depicted in figure 7. The rainfall deficit (i.e. diabatic cooling) over the SA monsoon system and SACZ induces a Gill-type response in the tropics, which causes two cyclonic circulations on each side of the Equator. Two Rossby wave trains with an equivalent barotropic structure propagate poleward in both Northern and Southern Atlantic regions, resulting in nearly symmetric atmospheric anomalies relative to the Equator with a larger amplitude in the upper troposphere of the Northern Hemisphere. The northward-propagating Rossby wave over the North Atlantic is re-enhanced and emanates downstream along the jet stream waveguide in the Northern Hemisphere. These downstream perturbations influence the temperature and rainfall over North America and Eurasia. The southward-propagating Rossby wave travels to the southwest Atlantic Ocean, emanating from convective cooling in the SACZ. Both branches of the ASP exhibit a baroclinic structure in the tropics and an equivalent barotropic structure in the mid-latitude (figure 7). The anomalous circulation associated with the ASP further modulates the SST in the mid-latitudes of both hemispheres through surface heat fluxes and induces an SST dipole in the South Atlantic and a tripolar pattern in the North Atlantic region. We note that the leadlag correlation between the ASP and other Atlantic climate variability indices, including the Atlantic Meridional Mode and the North Atlantic oscillation, was examined. However, no significant correlations were observed between the ASP and these indices (not shown).
The characteristics of the ASP found in this study are consistent with previous research on teleconnection patterns triggered by the SA monsoon system.  reported a similar Rossby-wave structure related to interannual SACZ variability, with an anomalous circulation at 30 • S that exhibited characteristics of the negative phase of the ASP. They also found that the pattern could induce a strong dipole over the midlatitudinal southwest Atlantic Ocean and was weakly correlated with SST over the tropical Pacific, leading them to conclude that ENSO is not the main cause of this structure. In another study, Grimm and Silva Dias (1995) investigated the global impacts of the SACZ convective heating using a barotropic model and found interhemispheric impacts through Rossby waves toward Eurasia and East Asia. They found that the Rossby waves triggered by convective heating over the SACZ can first propagate through the equator and can further become global circumcircular waves in both hemispheres. Similar features are also found in our LBM experiments, as well as our regression analysis, which circumcircular zonal wave travels through jet streams in both hemispheres, even from east Pacific. More model experiments will be needed to figure out whether such circumcircular feature is the downstream response of the interhemispheric energy transport by the SA monsoon or triggered by convective heating at other regions.
Other regions in the world have also been found to exhibit teleconnection patterns triggered by convective activity. For example, the Pacific-Japan pattern is excited in the tropical west Pacific and has remote influences on East Asia. Additionally, a recent study by Sekizawa et al (2021) has reported remote influences of the interannual variability of the Australian summer monsoon on climate over East Asia in austral summer. Their findings indicate that the energy of Rossby waves crossing the equator is due to the anomalous northward divergent winds from convective activity in the Australian summer monsoon, which reach the jet stream in the Northern Hemisphere. This study provides another example of the monsoon system in the Southern Hemisphere causing cross-equatorial influence, highlighting the importance of investigating teleconnections in various regions around the world. . The surface shading indicates SST anomalies: warm (red) and cold (blue). The dashed cumulus clouds represent the cooling convective forcing due to rainfall deficiency in the SA monsoon system and SACZ. The dashed contours on the surface indicate the Gill-type response and high-pressure anomalies (denoted by H) triggered by cooling. The grey arrows indicate the wave propagation direction induced by convective forcing.

Conclusion
This study comprehensively investigated the interhemispheric teleconnection pattern over the Atlantic Ocean, the ASP, which is established by the SA summer monsoon during JF. Our results also show the ASP has a structure of Rossby-wave-type response with two anomalous cyclonic-anticyclonic-cyclonic circulation trains aligned meridionally over the Atlantic Ocean. With strong atmosphere-convectionocean coupling, the ASP effectively regulates the interannual SST in the South Atlantic Ocean, especially over the western coastal regions with abundant ecological systems, as well as a strong teleconnection counterpart in the North Atlantic. The SST pattern over the North Atlantic Ocean is similar to the tripolar SST in the Atlantic Ocean (Han et al 2016). The perturbations induced by the ASP in 30 • N-60 • N exert influence on T2m over Eurasia, North America, Siberia, and the Arctic and on the precipitation over the North Atlantic Ocean. Complement to previous studies, this study provides a more quantit-ative assessment of the teleconnection pattern and its remote impacts, contributing to a better understanding of the climate variability over the Atlantic Ocean and its linkage to the global climate system.
Lastly, while the Amazon is a tipping element of the climate system (Lenton et al 2008, Lovejoy andNobre 2018), the ASP can further complicate the interacting network that includes the Amazon rainforest and the cascading effects of climate change. Many studies have suggested that the Amazon rainforest may turn into a tropical savanna in the near future if the dry conditions worsen because of the increased temperature and more frequent fires under deforestation (Cox et al 2004, Nobre et al 2016. The dieback of the Amazon rainforest can weaken the convection in the SA monsoon and eventually the effectiveness of Amazonia in modulating the atmospheric water cycle. Considering the wide-range impact of the ASP, a study of the effects of global warming on the ASP should be conducted to further understand the cascading effects on other parts of the climate system, especially in the extratropical Northern Hemisphere.