the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Westward extension of ENSO teleconnections links spring rainfall between northern Southeast Asia and the Arabian Peninsula
Abstract. ENSO teleconnections are not stationary, but how this nonstationarity reorganizes spring hydroclimate across Asia remains unclear. Spring rainfall is particularly important for the Arabian Peninsula, one of the world’s most arid regions, where even modest rainfall changes can affect water availability and drought risk. Here we show that a March–April rainfall seesaw between the Arabian Peninsula and northern Southeast Asia strengthened markedly after the late 1990s. During 1979–1998, ENSO-related convection and circulation anomalies remained mainly confined to the tropical Pacific–Maritime Continent sector and projected only weakly toward the Arabian Peninsula, with limited evidence of a coherent pathway linking the two regions. During 1999–2023, the ENSO-related response extended farther westward and northwestward, linking El Niño-related suppressed convection and reduced rainfall over northern Southeast Asia with enhanced rainfall over the Arabian Peninsula. This shift was associated with a reorganization of Indo-Pacific overturning, including a more distinct double-cell Walker response, a strengthened Arabian Peninsula vertical branch extending farther northward, and enhanced upper-level Rossby-type and thermal adjustment. Idealized model experiments show that western equatorial Indian Ocean heating can generate an Arabian Peninsula-sector response and that this response becomes more effective under the later-period background state. Observed changes in convection, low-level pressure and winds, and the upper-level jet and divergence further indicate that the later-period ENSO anomalies developed within a modified mean-state environment. These results identify a nonstationary ENSO teleconnection pathway in which the westward extension of tropical heating anomalies, Walker–Hadley coupling, and off-equatorial upper-level adjustment connect spring rainfall variability across Southeast and West Asia.
- Preprint
(6786 KB) - Metadata XML
-
Supplement
(7849 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-2784', Anonymous Referee #1, 06 Jul 2026
-
AC1: 'Reply on RC1', Dzung Nguyen-Le, 09 Nov 2026
Dear Anonymous Referee #1,
Thank you very much for your careful reading of the manuscript and for your constructive comments, which will help improve the paper considerably.
I will address all of the comments in the revised manuscript. In particular, the revision will:
- add the ENSO-related SST and rainfall patterns for both periods;
- clarify the interpretation of the Arabian Peninsula vertical-motion response;
- make the significance testing consistent across figures;
- extend the idealized model experiments to separate the roles of the ENSO heating pattern, the background state and the western Indian Ocean heating.A detailed point-by-point response will be provided with the revised manuscript.
Best regards,
Dzung Nguyen-LeCitation: https://doi.org/10.5194/egusphere-2026-2784-AC1
-
AC1: 'Reply on RC1', Dzung Nguyen-Le, 09 Nov 2026
-
RC2: 'Comment on egusphere-2026-2784', Anonymous Referee #2, 11 Nov 2026
The manuscript reports a strengthening of the anti-phase relationship between March–April rainfall over the Arabian Peninsula (AP) and northern Southeast Asia (SEA) after the late 1990s. It attributes this to a westward and northwestward extension of ENSO-related circulation anomalies, supported by Walker–Hadley diagnostics and idealized DREAM experiments with western Indian Ocean (WIO) heating. The question is interesting, and combining observational diagnostics with idealized experiments is a useful approach.
However, the central claim that ENSO links the two regions through a changed teleconnection pathway is stronger than the evidence supports. The statistical evidence for the transition needs closer examination, the individual AP and SEA responses to ENSO are not shown, the contribution of Indian Ocean variability (particularly the IOD) is not assessed, and some circulation interpretations and DREAM-based conclusions go beyond what the diagnostics demonstrate. These concerns affect the main conclusions but appear addressable through additional analyses and more cautious interpretation.
Major comments
1. Does the AP–SEA relationship demonstrate an ENSO-driven dynamical link? An anti-phase correlation does not establish a direct dynamical link, as ENSO could independently force opposite rainfall responses over the two regions. Please show AP–Niño3.4 and SEA–Niño3.4 regressions, with confidence intervals and spatial patterns, for P1 (1979–1998), P2 (1999–2023), and the full period. Fig. S7a suggests that the SEA response remains relatively stable while the AP response changes. If confirmed, the findings may be better interpreted as a change in the ENSO–AP relationship, and the title and “seesaw” terminology should be reconsidered. Furthermore, the manuscript attributes the post-1990s shift partly to a Central Pacific-type or Modoki-like ENSO pattern, yet the analysis relies primarily on the canonical Niño3.4 index. Including an explicit Modoki index, such as the ENSO Modoki Index (EMI), alongside Niño3.4 would help determine whether changes in ENSO spatial characteristics contribute to the regional rainfall responses. The authors should also clarify whether the MA relationship represents a distinct springtime response or is connected to broader changes in ENSO-related circulation documented in other seasons (discuss Hu et al. (2020) and Dasari et al. (2022)). Distinguishing changes in ENSO characteristics from changes in regional atmospheric responses is important for establishing the physical basis and novelty of the proposed mechanism.
2. Robustness of the moving-window correlations and breakpoint: In Fig. 1a, the raw P1 correlation (approximately −0.4 to −0.45 based on visual estimation) is already close to the 95% significance threshold. The IOBM-removed correlation is more negative than the raw correlation over much of the record, while the Niño3.4-removed correlation remains around −0.45 to −0.5 for windows centred between approximately 1989 and 2006. This suggests that an anti-phase component may have existed before 1999, with the ENSO contribution potentially shifting from offsetting to reinforcing it. Please report the raw and partial correlations (controlling for Niño3.4, IOBM, and both jointly), together with the underlying index–rainfall correlations. The abrupt changes around 2006–2008 coincide with the 1997/98 and 2015/16 ENSO events entering or leaving the moving windows. Leave one event out tests, alternative window lengths (17 and 21 years), and the GPCP results mentioned in Line 241 would help establish robustness. Significance estimates should account for effective sample size and serial dependence. Please also justify the 1999 breakpoint against alternative years, particularly since the Niño3.4-removed relationship appears to change mainly after approximately 2007 (i.e, rationale for choosing the P1 and P2 phases as citations if any (example, Hu et al., 2020)). Reporting regression coefficients per °C would help distinguish changes in ENSO variance from changes in the rainfall response. Where appropriate, consider providing quantitative values throughout the manuscript rather than relying solely on qualitative descriptions.
3. The role of western Indian Ocean variability and the IOD: The IOBM does not capture the IOD's zonal SST gradient or localized WIO warming, relevant because the DREAM heating (60°E, 0°N) is near the western IOD pole. Although the IOD peaks in boreal autumn, preceding variability and residual SST anomalies may influence MA rainfall. Zheng and Hoteit (2025), cited in Line 61, show that positive IOD conditions during boreal summer induce lower-level anticyclonic anomalies, suppressing ascent and AP rainfall. This contrasts with the spring DREAM response to WIO heating, which produces cyclonic anomalies and enhanced ascent. The authors should discuss this seasonal contrast and the nonlinear, asymmetric AP rainfall response associated with the mean-state precipitation floor effect highlighted by Zheng and Hoteit, particularly when interpreting linear partial correlations (Lines 237–242). The vertical velocity response should also be assessed in relation to climatological subsidence, as enhanced rainfall may reflect reduced descent rather than deep convective ascent. The DREAM experiments demonstrate the response to imposed heating, not its observed origin. Please examine the DMI, including preceding SON/DJF values, alongside Niño3.4 and IOBM, accounting for their covariance. Regressions of WIO SST and OLR against ENSO and DMI for P1 and P2 would help distinguish their contributions. The conclusions in Lines 475–477 should be qualified until these influences are better established.
4. Circulation diagnostics and DREAM interpretation: Positive −ω anomalies over a region of climatological subsidence may indicate weakened subsidence rather than anomalous ascent. Similarly, the P1 χ200 anomalies over West Asia that disappear in P2 may indicate a loss of an earlier convergence signal rather than newly established divergence. Please show climatological ω and reconcile these features with the interpretation of enhanced AP rainfall. The double-cell circulation change should be quantified, significance conventions across Figs. 3b, S5, and S6 reconciled, and below-ground pressure levels masked.
WAF calculated from the P2–P1 difference is not equivalent to the difference between period-specific WAF fields. Please consider showing the WAF for each period. Additional moisture-flux convergence diagnostics could help connect the circulation anomalies to rainfall, if needed.
Finally, the DREAM experiments impose WIO heating without the concurrent Pacific and Maritime Continent heating components. They therefore demonstrate AP sensitivity to WIO heating under two background states, rather than the full proposed teleconnection. Please provide observational support for the imposed heating, address the limitation of having only one 20-day run per background state, quantify the AP circulation differences, and acknowledge the limitations of the dry, uncoupled configuration.
Minor and Technical Comments:
- Lines 98–101, 85–87 – Introduction: The final paragraph presents results and should instead end with the objectives and hypotheses. The conditional statement at Lines 85–87 is later treated as established. Discuss previous work on decadal changes in the ENSO–AP relationship (Dasari et al., 2022, e.g., Kang et al., 2015).
- Lines 30–44 – AP rainfall region: The 50% spring fraction applies to a small southwestern sector (Fig. S1b), whereas the AP index covers a much larger area. Specify which subregions dominate the index.
- Lines 104–123 – Data and domains: Define CMAP at first use (Climate Prediction Center Merged Analysis of Precipitation).
- Lines 128–140 – Moving correlations and WCI: Describe the moving-window method (length and centring) before Fig. 1 and justify P1/P2 (e.g., Hu et al., 2020). The WCI western box (80–160°E) overlaps the SEA domain, and WCI correlates strongly with Niño3.4 (Fig. S7a). Please discuss the independent information it provides.
- Equations 3, 5 and 6; Lines 167–186: Clarify the treatment of below-ground levels in Eq. 3. Eq. 5 uses (r_x) and (r_y), whereas the text defines (r_1) and (r_2). Eq. 6 uses (F) while the text refers to WAF. Lines 177–179 repeat Lines 167–169. Clarify how Eq. 8 is applied to regressed anomalies.
- Lines 196–203, Fig. S2 – Prescribed heating: Cite Fig. S2a at first introduction and describe the heating structure and amplitude in the experimental set-up. Provide observational evidence or a reference for the WIO heating or convection represented by the forcing.
- Lines 222–229, Fig. 1 – Identify the red and blue curves in Fig. 1 as partial correlations, define or remove “MA[0]” in panel titles.
- Lines 275–306, Fig. 2: State the sign convention for χ200. The westward/northwestward extension is not clear in Fig. 2c, where the most coherent P2–P1 OLR differences are over the equatorial central Pacific. Please quantify the WIO/AP response or moderate the description.
- Lines 339–377, Fig. 3: Identify the panels explicitly and cite Fig. S4 at first comparison. Fig. 3a is averaged over 10°S–10°N, so AP statements should rely on Fig. 3b. Introduce the Indo-Pacific double-cell structure earlier with references (e.g., Ye and Hsieh, 2008; Schwendike et al., 2014), specify the latitude range of the clearer P2 AP signal, and replace “provides a pathway” with “is consistent with” unless a direct relationship and temporal ordering are demonstrated.
- Fig. 4 – Period-specific fields: Include P1 and P2 fields for quantities shown as differences, at least in the Supplement.
- Lines 430–531, Fig. 5 – DREAM and Discussion: The P2–P1 −ω maximum in Fig. 5f appears toward the Arabian Sea/Iran and SEA rather than clearly within the AP box; please quantify the regional responses. Reconcile Lines 502–512 (background state alone does not explain the transition) with the emphasis in the Abstract. Acknowledge data inhomogeneity, breakpoint selection, and individual-event sensitivity. Frame seasonal prediction implications (Lines 523–531) as potential relevance. Does Almazroui et al. (2022) establish nonstationarity in the specific relationship examined here?
- Technical corrections: Use consistent terminology for significance markings in captions (stippling versus hatching). Captions for Figs. 3, S4 and S5 should state the vertical-motion sign convention and units (per standard deviation or per °C). Check the inconsistent year and volume in the Hall and Derome reference (JAS 58, 2001 versus 57, 2000). Correct “Storch, H. V.” to “von Storch, H.” The Leroux and Hall (2025) reference lacks publication details. Clarify the relevance of Wang et al. (2013) to Lines 74–77. Remove the double period at Line 572 and resolve overlapping labels in Fig. 6b.
- Additional references to relevant previous studies would strengthen the manuscript throughout, particularly in the Introduction and Discussion, and provide better support for the proposed mechanisms and interpretations.
References
Dasari, H., Desamsetti, S., Langodan, S., Attada, R., Ashok, K., and Hoteit, I. (2022). Long-term changes in the Arabian Peninsula rainfall and their relationship with the ENSO signals in the tropical Indo-Pacific. Climate Dynamics, 59. https://doi.org/10.1007/s00382-021-06062-7.
Hu, Z., Kumar, A., Huang, B., Zhu, J., L’Heureux, M., McPhaden, M. J., and Yu, J. (2020). The interdecadal shift of ENSO properties in 1999/2000: A review. Journal of Climate, 33, 4441–4462. https://doi.org/10.1175/JCLI-D-19-0316.1.
Schwendike, J., Govekar, P., Reeder, M. J., Wardle, R., Berry, G. J., and Jakob, C. (2014). Local partitioning of the overturning circulation in the tropics and the connection to the Hadley and Walker circulations. Journal of Geophysical Research: Atmospheres, 119, 1322–1339. https://doi.org/10.1002/2013JD020742.
Ye, Z., and Hsieh, W. W. (2008). Changes in ENSO and associated overturning circulations from enhanced greenhouse gases by the end of the twentieth century. Journal of Climate, 21, 5745–5763.
Zheng, Y., and Hoteit, I. (2025). Asymmetric impacts of Indian Ocean Dipole on summer climate over Arabian Peninsula. Geophysical Research Letters, 52, e2025GL118195. https://doi.org/10.1029/2025GL118195.
Citation: https://doi.org/10.5194/egusphere-2026-2784-RC2 -
AC2: 'Reply on RC2', Dzung Nguyen-Le, 11 Nov 2026
Dear Anonymous Referee #2,
Thank you very much for your careful reading of the manuscript and for your detailed and constructive comments, which will help improve the paper considerably.
I will address all of the comments in the revised manuscript. In particular, the revision will:
- show the ENSO relationships of Arabian Peninsula and northern Southeast Asian rainfall separately for both periods, and reconsider the title and the “seesaw” terminology accordingly;
- examine the robustness of the moving-window correlations and the choice of breakpoint, including leave-one-event-out tests;
- assess the roles of the Indian Ocean Dipole and the ENSO Modoki index;
- interpret the vertical-motion and velocity-potential changes relative to the climatological circulation;
- extend the idealized model experiments to include the observed ENSO-related heating, and state their limitations more clearly.A detailed point-by-point response will be provided with the revised manuscript.
Best regards,
Dzung Nguyen-LeCitation: https://doi.org/10.5194/egusphere-2026-2784-AC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 220 | 151 | 37 | 408 | 95 | 24 | 23 |
- HTML: 220
- PDF: 151
- XML: 37
- Total: 408
- Supplement: 95
- BibTeX: 24
- EndNote: 23
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General Comments.
The manuscript examines ENSO teleconnections for spring rainfall over Southeast (SE) Asia and the Arabian Peninsular (AP) in two time periods, from 1979-1998 and 1999-2023. The authors propose a shifting dynamical pathway which enhances the relationship between suppressed SE Asia rainfall and increased AP rainfall during El Nino. The authors carry out idealised experiments prescribing warming in the Indian Ocean to test the sensitivity of the AP response to a changing background climate. I found most of the manuscript to be clear and well-written with good descriptive figure captions. However, when I got to the idealised model discussion and conclusions I found it difficult to piece together those components. I think the manuscript makes some interesting points and would provide additional insight into changing ENSO teleconnection if published. However, the current manuscript could benefit from additional introductory figures, some additional attention to some of the statements supported by the supplemental figures, and a bit more clarity about how the idealised simulations support the conclusions. Many of my comments are probably easily addressed, but because of the quantity of my suggestions and because some relate to the interpretation and statistical significance of the results, I’d recommend major revisions. Find my comments below.
Specific comments
- Can the authors be clearer in this section about which latitudinal portions you are referring to? For example, the text refers to the AP-sector response being ‘clearer’ in P2 (Line 366). From Fig S5, the P2 responses seem relatively weak. Is this referring to the small area of ascent on the northern boundary of AP? It’s not clear to me from the text.
- Overall, the changes discussed seem to indicate a reduced descent response, as opposed to a particularly enhanced ascent response which is implied in the text.
- I note that the hatching is different from Fig S5 bottom panels and Fig 3b;c when it seems the same quantities are plotted. Why are these different?
Minor/Technical Comments.