A change in the relationship between ENSO and the South Atlantic Subtropical Dipole in the past four decades

Yu, Lejiang; Zhong, Shiyuan; Vihma, Timo; Sui, Cuijuan; Sun, Bo

doi:https://doi.org/10.5194/egusphere-2023-2436

Preprints

https://doi.org/10.5194/egusphere-2023-2436

Preprints

11 Dec 2023

| 11 Dec 2023

A change in the relationship between ENSO and the South Atlantic Subtropical Dipole in the past four decades

Lejiang Yu, Shiyuan Zhong, Timo Vihma, Cuijuan Sui, and Bo Sun

Abstract. This study investigates the relationship between sea surface temperature (SST) in the subtropical Atlantic Ocean, as represented by the Southern Atlantic Subtropical Dipole (SASD), and SST in the tropical Pacific Ocean, identified by the El Niño-Southern Oscillation (ENSO). Our analysis reveals a significant inverse correlation between the SASD and Niño indices over a century, with multi-decadal variability that contradicts weak simultaneous correlations previously reported in the literature. The study also highlights a strengthening of their inverse correlations in the most recent two decades compared to the preceding two decades, which can be attributed to the shift in ENSO regime from more frequent eastern Pacific El Niño to central Pacific El Niño around the turn of the century. This shift helps set the stage for changes in convective activity in the critical region (20° S–40° S, 180°–140° W) of the central South Pacific Ocean, triggering wavetrains that propagate along different paths and ultimately contributing to different southern Atlantic subtropical high (SASH) and changes in anomalous SST patterns in the subtropical Atlantic Ocean. These findings advance our understanding of the interactions between South Atlantic and Pacific SST variations, which strongly influence rainfall patterns particularly in South America and southern Africa and may improve sub-seasonal to seasonal precipitation predictions in these regions.

Received: 20 Oct 2023 – Discussion started: 11 Dec 2023

Download & links

Preprint (PDF, 2003 KB)

Supplement (918 KB)

Download & links

Lejiang Yu, Shiyuan Zhong, Timo Vihma, Cuijuan Sui, and Bo Sun

Status: final response (author comments only)

RC1:
'Comment on egusphere-2023-2436', Anonymous Referee #1, 22 Dec 2023

General comments:
This paper examines a multidecadal change in the relationship between ENSO and the South Atlantic Subtropical Dipole (SASD). The authors identify that the SASD tends to have a stronger relationship with ENSO after 2000. They also claim that the shift of ENSO types from the eastern Pacific to the central Pacific one facilicates atmospheric convection variability in the South Pacific, triggers atmospheric Rossby waves propagating eastward along the westerly wave guide, and generates the subtropical high variations in the South Atlantic responsible for the SASD. However, the atmospheric teleconnection between different ENSOs and the SASD is already reported in a previous study using a coupled general circulation model (Rodrigues et al. 2015), so it is unclear what exactly the new finding is in this study.
            First, the authors show multidecadal changes in the regression maps before and after 2000 (Fig. 2c, 3c, 5c, 6c), but most of the signals in the South Pacific are not so robust and significant to support a possible influence of the central Pacific ENSO on the generation of eastward-propagating Rossby waves. This is probably due to the nature of regression analysis that takes into account all years including weak and non-SASD events. So, I would recommend that the authors should make a composite analysis only for strong positive and negative SASD events (defined by above one and below minus one standard deviation of the SASD index), then examine the robustness of the anomaly differences before and after 2000.
            Second, the connection between the convection variability and eastward propagation of Rossby waves in the South Pacific is not clearly addressed. Differences in the regression of OLR anomalies (Fig. 3c) show enhanced convection northeast of New Zealand (160-120ºW, 20-40ºS), but the Rossby waves emanate from east of New Zealand after 2000 (180ºE-140ºE, 40-60ºS; Fig. 4c). There is a clear difference in the location of the convection variability and Rossby wave sources, so what exactly triggers these Rossby waves and why they propagate eastward, not southeastward as observed before 2000 (Fig. 4a)? Does the location and/or intensity of convection variability in the South Pacific matter for the pathways of Rossby waves? If this is the case, I would recommend that the authors should conduct more detailed sensitivity experiments by prescribing SST variability in different regions of the South Pacific (e.g., western, central, and eastern South Pacific).
            Furthermore, the authors prescribed unrealistic SST anomalies (2 ºC) in their sensitivity experiments to see their impact on the atmospheric teleconnection. This is not appropriate setting and should be corrected using more realistic SST anomalies (e.g., one standard deviation of SST anomalies related to ENSO and the SASD). Also, the SASD is a coupled phenomenon, so why did not the authors use a coupled general circulation model rather than a single atmospheric model, as performed by Rodrigues et al. (2015)? The convection variability in the South Pacific may generate the subtropical high variations in the South Atlantic via the atmospheric teleconnection, but the subtropical high variations do not necessarily cause a dipole pattern of SST anomalies associated with the SASD.
            Finally, the authors discuss the relationship between ENSO and the SASD dating back to 1871 (Fig. S1), but there are no reliable observations in the South Atlantic before the satellite period. I understand the need of prolonged data to test the robustness of the multidecadal relationship identified in this study. However, it is hard to believe the relationship between ENSO and the SASD before the 1980s due to low data quality. So, I would recommend to remove the discussion from the main text. Considering these major flaws, I could not recommend this paper for possible publication in this journal unless the substantial revision including further model experiments is made. More specific comments on this paper are given below.

Specific comments:
L6: Rephrase as “ENSO indices from 1979 to the present”.
L12: Please correct the lon-lat information of critical region following the comments below.
L36: Remove “in contrary to the current understanding”.
L38-39: How can you conclude the future ENSO-SASD relationship from the historical analysis presented here?
L47: Rephrase as “represents an opposite sign of sea surface temperature (SST) anomalies”
L51-56: The authors should review the physical processes on the South Atlantic subtropical dipole carefully. Following the previous literature, the SASH variations in austral spring change the surface evaporation thereby the mixed-layer depth, then the warming of mixed-layer by shortwave radiation is modulated by the mixed-layer variations so that the dipole SST anomalies develop in austral summer
L61: Rephrase as “Understanding factors behind the SASD interannual variability”
L84: As mentioned earlier, we do not have any reliable observations in the South Atlantic before the satellite period. How can you show robustness of the relationship between ENSO and the SASD dating back to 1871?
L71-74: Rodrigues et al. (2015) has already pointed out the different teleconnection patterns between the central and eastern Pacific El Nino events over the South Atlantic, but what is the difference between their study and the present work?
L83-84: You cannot extend the analysis on the relationship back to 1871 because of no reliable observations available in the tropical Pacific and South Atlantic.
L97: How did you select 18 years for the sliding correlation?
L97: When you applied the 18-year sliding correlation, did you use the SASD and ENSO indices during austral summer (December-February) or slightly lagged months between them? I would recommend the latter because of the lagged relationship, for example, the SASD index during austral summer and ENSO index during austral spring.
L115: How did you make this approximation in Eq. (1)? Is there any assumption you made for deriving this equation? If so, it would be better to briefly mention in the main text.
L119: 200 hPa would be suitable for identifying Rossby wave sources in the tropics, while 250 hPa would be better to describe the Rossby wave propagation in the extratropics. Would the results improve when you calculate the wave activity flux at 250 hPa?
L133: The SASD is an air-sea coupled phenomenon, but why did the authors use the AGCM to conduct the sensitivity experiments, unlike Rodrigues et al. (2015) using a coupled model? The AGCM experiments are not able to generate the SASD, although they can describe the SASH variations through the atmospheric teleconnections.
L139: As stated earlier, +2 ºC SST anomaly in the South Pacific or tropical Pacific is too large and unrealistic. How large is one standard deviation of the SST anomaly in those regions?
L142-144: Why did the authors prescribe the SST anomaly from January to March during the peak of the SASD? According to the previous literature, the atmospheric teleconnection triggering the SASD starts from October to December, then it takes one or two months for the SASD to reach its maximum.
L175-179: As mentioned earlier, there are no reliable observations available in the South Atlantic before the satellite period. I would recommend this paragraph to be removed from the main text.
L183: Rephrase as “explore reasons (or factors) behind”.
L192-193: The positive SST anomalies in the difference map over the western tropical Pacific and South Atlantic (Fig. 2c) are not significant, so you cannot conclude a stronger influence of ENSO on the SASD in the recent decades. How about the regression of SST anomalies on the SASD index to show significant differences in the tropical Pacific and South Pacific?
L197: The OLR anomalies in the difference map (Fig. 3) are not so significant as the SST anomalies (Fig. 2), so I am wondering how robust the impact of the SST anomalies is onto the OLR anomalies.
L199: Again, why did you calculate the regression of OLR anomalies on the NINO3.4 index rather than the SASD index?
L215: Rephrase as “the southwestern Pacific Ocean during 1979-1999”.
L218: Rephrase as “The wavetrain during 2000-2020”.
L221-233: The authors discuss the underlying processes on the SST anomalies qualitatively, but this is based on a speculation. Why do not the authors explore the detailed mechanisms quantitatively using atmosphere or ocean reanalysis products?
L241: The difference map shows slightly significant and negative OLR anomalies “northeast” of New Zealand indicated by a green box (Fig. 3c), but how do the negative OLR anomalies there contribute to increase of eastward propagation of Rossby waves “east” of New Zealand (Fig. 4c)? The locations of OLR anomalies and Rossby waves look different.
L260: The green box in the difference map (Fig. 3c) does not cover the center of negative OLR anomalies (160-120ºW, 20-40ºS). Why did the authors select the region off the negative OLR anomalies?
L268-271: The zonal wind in the difference map (Fig. 5c) shows stronger westerlies in the South Pacific, but the difference is not significant. How does the small difference in the zonal wind lead to a significant difference in the Rossby wave propagations (Fig. 4a, c)? Is it related to the location and/or intensity of the convection anomalies?
L286: Rephrase as “the regressions of the SST anomalies onto the SASD index”.
L290-291: The SST anomalies in the difference map (Fig. 6c) show significant La Nina-like pattern, but the SST anomalies in the South Atlantic are not significant. Does it mean that ENSO-SASD relationship gets stronger, but does not influence the amplitude and pattern of the SASD?
L297-298: Again, the SST increase by 2 ºC is not realistic, compared to the regressions of the SST anomalies (Fig. 6a,b). Also, the key region for the OLR anomalies does not show significant increase in the SST anomalies (Fig. 6c). The authors should reconsider the experimental designs.
L301-302: I am wondering if the negative geopotential height anomalies in the South Atlantic are related to a negative phase of the SAM rather than the Rossby waves emanating from the South Pacific. Could you describe the WAF over the geopotential height in Fig. S4?
L320-325: The SASD is generated by the meridional shift and strengthening/weakening of the SASH according to the previous literature, but how does the westward shift of the SASH induce the SASD? It is expected to cause a dipole pattern of SST anomalies near the coast of Brazil, but would it be different from the canonical SST anomalies associated with the SASD?
L341: Rephrase as “we propose that”
L374-378: As mentioned earlier, a few-month lagged relationship with ENSO is important for generation of the SASD. Why did not the authors investigate this lagged relationship?
L379-383: The global warming may also affect the SASH variations and hence SASD directly, but this should also be explored in the future study.
L387: Remove “possibilities for”

Citation: https://doi.org/10.5194/egusphere-2023-2436-RC1
- AC1: 'I have addressed the comments of the reviewer', Lejiang Yu, 04 Mar 2024
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2023/egusphere-2023-2436/egusphere-2023-2436-AC1-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2023-2436-AC1
- AC2: 'I have addressed the comments of the reviewer', Lejiang Yu, 04 Mar 2024
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2023/egusphere-2023-2436/egusphere-2023-2436-AC2-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2023-2436-AC2
- AC3: 'The response to review report', Lejiang Yu, 04 Mar 2024
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2023/egusphere-2023-2436/egusphere-2023-2436-AC3-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2023-2436-AC3
RC2:
'Comment on egusphere-2023-2436', Anonymous Referee #2, 24 Jan 2024

General comments:
This manuscript investigates the causes behind a change in the relationship between the El Nino-Southern Oscillation (ENSO) and the South Atlantic Subtropical Dipole (SASD). The authors find that this relationship has strengthened in the most recent two decades and attribute this strengthening to an increase in central Pacific El Nino events. They argue that a shift in convective activity leads to different wave trains and responses over the subtropical Atlantic Ocean in 2000-2020 compared to 1979-1999. Although the multi-decadal variability in the ENSO/SASD relationship is intriguing, the key results in this study are generally not statistically significant, and the authors use an atmosphere-only model to support their argument about a change in a coupled atmosphere-ocean relationship. Therefore, I recommend major revisions before this manuscript can be considered for publication in this journal.
A change in the relationship between ENSO and the SASD is assessed through regression of the Nino 3.4 index onto various fields across 1979-1999 and 2000-2020. However, the differences between these two periods are generally not statistically significant or clear. For example, the SST anomalies over the central Pacific and Atlantic in Figure 2c and the associated changes in OLR in Figure 3c are not significant. Moreover, the key region of convective activity is not particularly strong compared to the changes in OLR across the tropical Pacific and the chosen region does not fully encompass the peak anomalies. Therefore, this analysis does not adequately support the authors’ argument that a change in this relationship is related to more central Pacific El Nino events. This lack of significance could be related to only 21 years being used for the analysis. Can the authors use a different approach to reveal a more robust change?
ENSO influences the SASD through coupled atmosphere-ocean interactions. Therefore, it is unclear why an atmosphere-only model was used to demonstrate a change in their relationship. The current analysis can only show that ENSO teleconnections influence the atmospheric circulation over the Atlantic, not that they can cause the SASD pattern. I suggest that the authors consider using a partially coupled experiment to better support their key results, i.e. impose an SST anomaly in the Pacific in a fully coupled model, so that the Atlantic Ocean can evolve in response to the Pacific anomaly. A quantitative assessment of ocean and SST changes over the Atlantic in response to El Nino, as described in the text, would also be helpful.
Specific comments:
Line 14: Delete “SASH” as the acronym is not used again in the abstract.
Lines 38-39: Suggest deleting this key point as projections have not been examined in this study.
Lines 46-48: Suggest referring to Figure 1a here.
Lines 62-69: It would be helpful to briefly elaborate on how the Antarctic Oscillation and Subtropical Indian Ocean Dipole influence the SASD. The link between ENSO and the SASD should also be introduced here. Extending Figure 1a to include the Pacific Ocean would help to show that the two ocean basins co-vary.
Lines 91-92: The study period could be extended slightly further back to use the full ERA5 data availability and also extended through to 2023.
Line 97: Why was an 18-year sliding correlation chosen?
Lines 129-130: Are the results sensitive to the chosen season (January to March)? Unlike the SASD, ENSO peaks towards the end of the calendar year, rather than in February. I think it might make more sense to look at ENSO in November to January to capture the lagged relationship.
Lines 131-136: Given the aim of these experiments is to demonstrate a link between SSTs in the tropical Pacific Ocean and subtropical Atlantic Ocean, why was an atmosphere-only model used? These results will only be able to show that SSTs in Pacific Ocean lead to changes in the atmospheric circulation over the Atlantic Ocean, but will miss the additional and important step of driving changes in the Atlantic SSTs.
Lines 137-138: Do the climatological SSTs represent present-day or preindustrial conditions?
Lines 142-144: The imposed SST anomalies are designed to mimic different types of El Nino events, but ENSO events generally start to decay during January to March. Why was this period chosen? And what happens to the link between the SASD and ENSO in the rest of the year? Imposing an anomaly with a seasonal cycle would be more realistic.
Lines 161-166: Suggest deleting “which is unsurprising given the known correlations among the three Nino indices” because the next sentence shows that the correlation between the SASD and Nino 3 and 4 indices is quite different in the earlier period.
Lines 185-187: 21 years is a short period for the regression analysis.
Lines 191-195: For the difference between the two periods, the high SST anomalies over the central Pacific are not statistically significant. The anomalies over the Atlantic basin are also not significant. So this figure does not really support a shift to more central Pacific El Nino events with impacts on the SASD.
Line 196: What significant differences are being referred to here? The only significant differences in Figure 2c are the low SST anomalies in the eastern Pacific.
Lines 199-218: The key region of negative OLR for El Nino is over the central tropical Pacific Ocean. The anomalous tropical heating induces upper-level divergence that advects mean vorticity out of the tropics and thus produces a Rossby wave source in the subtropical westerlies.
Lines 221-233: The link between the atmospheric teleconnections forced by El Nino and the SST changes in the Atlantic described in this paragraph is a key part of the ENSO/SASD relationship but it is not sufficiently explored in this manuscript. A quantitative analysis of the interbasin interactions, rather than a description from earlier studies, would help to strengthen the authors’ argument. Using a coupled model for the idealised experimentation would also help to show this link.
Lines 228-229: Sensible heat flux does not appear to be shown in Figures 4b and d.
Lines 238-239: The OLR anomalies for the difference between the two periods are not statistically significant.
Line 248: It is not clear where the weaker northeasterlies are in Figure 4d. Consider showing the difference plot here.
Lines 249-250: Do the authors mean westerly? Does one of the figures show the climatological southeasterlies in the eastern South Atlantic?
Lines 249-253: The current analysis does not show that changes in the winds lead to changes in the SSTs.
Lines 259-250: Why was this particular region chosen as the key region? The negative OLR anomalies extend further eastward.
Lines 267-268: I think it would be helpful to also look at the total stationary Rossby wavenumber (Ks) to get a better indication of changes in the waveguide.
Lines 268-271: The differences in the zonal wind between the two periods are quite subtle and are not statistically significant, apart from a small region over the South Atlantic. Can these small changes really drive the different propagation paths shown in Figures 4a and c?
Lines 286-289: An examination of composites of SST during positive and negative SASD index and eastern and central Pacific El Nino events might make these similarities clearer.
Line 306: The wave train induced by the SST anomaly does not appear to induce a dipole over the Atlantic (Figure S5) like in Figure 1a. It appears to the shifted far southward.
Lines 306-309: Do the OLR anomalies in this experiment resemble the anomalies in Figure 3?
Lines 321-323: It should be noted here that the anomalies in the central tropical Pacific and eastern tropical Pacific experiments are placed in the tropics but the anomaly in the main experiment is located at 20-40S.
Lines 323-325: Can the authors elaborate on how a westward shift of the geopotential height anomalies leads to the development of the negative phase of the SASD?

Citation: https://doi.org/10.5194/egusphere-2023-2436-RC2
- AC2: 'I have addressed the comments of the reviewer', Lejiang Yu, 04 Mar 2024
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2023/egusphere-2023-2436/egusphere-2023-2436-AC2-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2023-2436-AC2
- AC1: 'I have addressed the comments of the reviewer', Lejiang Yu, 04 Mar 2024
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2023/egusphere-2023-2436/egusphere-2023-2436-AC1-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2023-2436-AC1
AC3: 'The response to review report', Lejiang Yu, 04 Mar 2024

The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2023/egusphere-2023-2436/egusphere-2023-2436-AC3-supplement.pdf

Citation: https://doi.org/10.5194/egusphere-2023-2436-AC3

Lejiang Yu, Shiyuan Zhong, Timo Vihma, Cuijuan Sui, and Bo Sun

Supplement

https://doi.org/10.5194/egusphere-2023-2436-supplement

Lejiang Yu, Shiyuan Zhong, Timo Vihma, Cuijuan Sui, and Bo Sun

Viewed

Total article views: 329 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
234	72	23	329	32	14	16

HTML: 234
PDF: 72
XML: 23
Total: 329
Supplement: 32
BibTeX: 14
EndNote: 16

Views and downloads (calculated since 11 Dec 2023)

Month	HTML	PDF	XML	Total
Dec 2023	100	27	10	137
Jan 2024	44	10	4	58
Feb 2024	12	9	2	23
Mar 2024	57	20	6	83
Apr 2024	21	6	1	28

Cumulative views and downloads (calculated since 11 Dec 2023)

Month	HTML	PDF	XML	Total
Dec 2023	100	27	10	137
Jan 2024	44	10	4	58
Feb 2024	12	9	2	23
Mar 2024	57	20	6	83
Apr 2024	21	6	1	28

Viewed (geographical distribution)

Total article views: 316 (including HTML, PDF, and XML) Thereof 316 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 26 Apr 2024

Short summary

In contrary to the current understanding, there can be a strong connection between ENSO and the South Atlantic Subtropical Dipole (SASD). It is highly probable that the robust inverse correlation between ENSO and SASD will persist in the future. The ENSO-SASD correlation exhibits substantial multi-decadal variability over the course of a century. The change in the ENSO-SASD relation can be linked to changes in ENSO regime and convective activities over the central South Pacific Ocean.


Total:	0
HTML:	0
PDF:	0
XML:	0