the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 01 Oct 2026
Explaining the Modulation of Forced Precipitation Changes with Internal Variability Storylines: The case of decadal Sahel Precipitation Changes
Abstract. Decadal Sahel rainfall is strongly influenced by internal climate variability, which can substantially modulate anthropogenically forced changes, particularly in the near- and mid-term. Yet, the complexity and inherent randomness of internal variability make it difficult to interpret, communicate and integrate into decision-relevant climate information. Here, we develop internal variability storylines of Sahel summertime precipitation using a large model ensemble. We identify the Atlantic Multidecadal Variability (AMV) and Pacific Decadal Variability (PDV) as two key oceanic drivers of decadal Sahel rainfall variability that act through distinct atmospheric mechanisms: AMV-related sea surface temperature changes are consistent with changes in the Atlantic inter-hemispheric temperature gradient, shifting the meridional position of the Intertropical Convergence Zone (ITCZ) and associated monsoon circulation, whereas positive PDV increases tropospheric stability and subsidence over the Sahel, suppressing convection. By conditioning on the combined phases of AMV and PDV, we construct physically coherent climate storylines that illuminate how decadal internal variability can organise tropical circulation and Sahel precipitation anomalies. Their effects reinforce most strongly in out-of-phase configurations, with AMV+/PDV− producing a wet Sahel, and AMV−/PDV+ a dry Sahel, while in-phase states show weak responses. These combined influences are captured by the North Atlantic Relative Index (NARI), which combines thermodynamic and dynamic impacts of oceanic forcings on the Sahel. A positive NARI corresponds to AMV+/PDV−, whereas a negative NARI is associated with AMV−/PDV+, with NARI explaining 37% of Total Sahel precipitation variability. Understanding the pathways of oceanic drivers reveals how they modulate the forced response: around mid-century, opposing storylines can more than double or cancel forced regional precipitation changes. Overall, our results demonstrate how internal variability storylines can extract physically grounded decadal climate information from large ensembles, providing a framework complementary to initalised predictions for regional climate information.
- Preprint
(11946 KB) - Metadata XML
-
Supplement
(7893 KB) - BibTeX
- EndNote
Status: open (until 16 Dec 2026)
- RC1: 'Comment on egusphere-2026-5597', Elsa Mohino, 10 Nov 2026 reply
-
RC2: 'Comment on egusphere-2026-5597', Paul-Arthur Monerie, 11 Nov 2026
reply
Review of “Explaining the Modulation of Forced Precipitation Changes with Internal Variability Storylines: The case of decadal Sahel Precipitation Changes” by Julius Mex and co-authors.
The authors have applied a Storyline approach to understand how internal climate variability could shape the future changes in Sahel precipitation, with a focus on decadal variability. They show that both Atlantic and Pacific SSTs can modulate precipitation over the Sahel in CESM2 and constructed a storyline in which out-phase Atlantic and Pacific SST can obscure and amplify the effects of the externally forced response. The NARI is used as a single index to link AMV/PDV with Sahel precipitation and to explain mechanisms. The study is a useful addition to the existing literature. I have a few comments, to help improve the paper.
Specific comments
1) The Storyline is applied to the ensemble members of CESM2 to focus on internal climate variability. The authors have used the piControl simulations to quantify the effect of internal climate variability on Sahel precipitation, but the storyline is based on the CESM2 Large Ensemble, a different set of simulations. The choice of using the piControl and not the CESM2 Large Ensemble to quantify internal climate variability and its effects on Sahel precipitation must be motivated. The authors can, for example, not quantify how the effects of internal variability could change in the future – that could be addressed using the CESM2 Large Ensemble (SSP3-7.0). Line 86 is confusing because it states that internal climate variability was defined using the Large Ensemble; the text could be rephrased to show more clearly what was done with the Large Ensemble and what was done with the piControl simulations.
Section 3.3 relies on the piControl simulations. While the piControl simulation allows analysing 1100 years of simulations, the AMV/PDV Storylines are defined using 16 to 29 points. Would it be useful to also quantify the effects of internal climate variability from the CESM2 Large Ensemble (historical and SSP3-7.0) to add many more years to the analysis (after removing the externally forced response)? i.e., is the result sensitive to the number of years that is used?
2) Methods. The methods section could be improved. The four quadrants of the AMV-PDV phase space are shown in Figure 3, and this could be explained earlier in the text with a sentence stating why computing the AMV-PDV phase space (e.g. to show the co-variability between the drivers and selection of ensemble members or of the average across time periods, for the storylines, to link to Sahel precipitation). It could be clearer why you are using the piControl simulations while Liné et al. (2024) use the ensemble members of the large ensemble for the storyline. Why perform an ANOVA rather than another statistical method?
3) AMV and IPO as selected drivers. The authors motivate the use of AMV and IPO as drivers of Sahel precipitation, based on the literature. The decadal variability of the Atlantic and Pacific SSTs is a well-identified driver for Sahel precipitation. However, it is not immediately clear from Figure 1a that AMV (which is associated with strong warming in the North Atlantic subpolar gyre) is the most relevant driver here. See the large differences between Figure 1a and Figure 1b regarding the SST pattern in the North Atlantic. I suggest commenting/discussing the potential inability of the model to simulate Sahel precipitation decadal variability and the effects of the AMV on Sahel precipitation.
4) Forced NARI. I am not sure to follow the point about the high correlation between the forced NARI and forced change in Sahel precipitation (r=0.96). This could indicate that there is a strong link between NARI and Sahel precipitation, but the high correlation coefficient could just be due to the trend in both the forced NARI and Sahel precipitation. The negative trend in Sahel precipitation can be due to a weaker warming of the North Atlantic than of the tropics, with not necessarily a clearly identified IPO positive-like phase. In addition, I found this paragraph slightly confusing the forced NARI cannot be directly compared to internally driven NARI (effect of the GHG and anthropogenic aerosols) and the reader could be left thinking that both are equivalent. It could be best to rename it as fNARI (something like this). Eventually, is panel 6a necessary? Figure 6 shows a mix of forced and internal NARI, which is slightly confusing.
5) I understand that the PDV/AMV phase space is computed from the piControl simulations (e.g., Figure 3, Figure 4). Please explain what has been done then for Figure 6b (crosses), is that by superimposing the storyline computed from the piControl simulations or is it by computing the AMV/PDF phase space from the CESM2 Large ensemble, following the same approach than for when using piControl? The text was slightly confusing here and the caption of Figure 6 is not helpful. If the crosses are defined from only the NARI values then it would be best to also have the PDV/AMV storyline, which are the two drivers of the Storyline.
Technical corrections
PDV. The authors used the Tripole Index of Henley et al. (2015), which was used to represent the IPO. PDV refers to the variability in the Pacific, that is, the PDO and the IPO. Both PDO and IPO can have different impacts globally and over the Sahel. Please use IPO instead of PDV.
Results. Thank you for showing the cross-section of vertical velocity anomalies. It would also be useful to see cross sections in zonal wind (to document changes in the African Easterly Jet strength and location) and wind divergence (to see divergence associated with the Heat Low; Shekhar and Boos 2017) to further document the mechanisms.
Shekhar R, Boos WR (2017) Weakening and shifting of the Saharan shallow meridional circulation during wet years of the West African monsoon. J Clim 30:7399–7422. https://doi.org/10.1175/JCLI-D-16-0696.1
Line 30. Please add references on the effects of the external forcings on West African precipitation. See, for example, Hirasawa et al. (2020, 2022) and Monerie et al. (2022)
Hirasawa, H., P. J. Kushner, M. Sigmond, J. Fyfe, and C. Deser, 2020: Anthropogenic aerosols dominate forced multidecadal Sahel precipitation change through distinct atmospheric and oceanic drivers. J. Climate, 33, 10187–10204, https://doi.org/10.1175/JCLI-D-19-0829.1.
Hirasawa, H., Kushner, P. J., Sigmond, M., Fyfe, J., & Deser, C. (2022). Evolving Sahel rainfall response to anthropogenic aerosols driven by shifting regional oceanic and emission influences. Journal of Climate, 35(11), 3181–3193. https://doi.org/10.1175/JCLI-D-21-0795.1
Monerie, P., L. J. Wilcox, and A. G. Turner, 2022: Effects of Anthropogenic Aerosol and Greenhouse Gas Emissions on Northern Hemisphere Monsoon Precipitation: Mechanisms and Uncertainty. J. Climate, 35, 2305–2326, https://doi.org/10.1175/JCLI-D-21-0412.1.
Lines 35-36. Several authors have shown that internal climate variability can obscure the forced response in short-term projections. Adding and discussing these references would provide relevant information and help link the work to the existing literature. See Monerie et al. (2017) and Gaetani et al. (2020).
Gaetani, M., Janicot, S., Vrac, M. et al. Robust assessment of the time of emergence of precipitation change in West Africa. Sci Rep 10, 7670 (2020). https://doi.org/10.1038/s41598-020-63782-2
Monerie, P. A., Sanchez-Gomez, E., Pohl, B., Robson, J., & Dong, B. (2017). Impact of internal variability on projections of Sahel precipitation change. Environmental Research Letters, 12(11), 114003.
Lines 42-43: In addition to Sheen et al. (2017), the authors could consider citing Mohino et al. (2016), and probably Ndiaye et al. (2022).
Mohino, E., Keenlyside, N. & Pohlmann, H. Decadal prediction of Sahel rainfall: where does the skill (or lack thereof) come from?. Clim Dyn 47, 3593–3612 (2016). https://doi.org/10.1007/s00382-016-3416-9
Ndiaye, C. D., E. Mohino, J. Mignot, and S. M. Sall, 2022: On the Detection of Externally Forced Decadal Modulations of the Sahel Rainfall over the Whole Twentieth Century in the CMIP6 Ensemble. J. Climate, 35, 6939–6954, https://doi.org/10.1175/JCLI-D-21-0585.1.
Line 49: Please correct the reference: Findell et al.
Line 60: You could eventually also refer to Guilbert et al. (2024) to add potential effects of the Pacific SSTs, to also motivate the choice of the PDV as a driver.
Guilbert M, Terray P, Mignot J et al (2024) Interhemispheric temperature gradient and Equatorial Pacific SSTs drive Sahel monsoon uncertainties under global warming. J Clim 37:1033–1052
Lines 66-68: This is the first paper to apply an internal variability storyline to Sahel precipitation; the authors should highlight this here. There is a gap in the literature.
Line 72: I do not think the first question is the paper's scientific question. It is a key element and important information for applying the internal variability storyline.
Line 157: Please also note the northward shift of the shallow circulation. The Hadley cell is a zonal-mean quantity, so it would be best to refer to the Monsoon cell or the regional manifestation of the Hadley Cell.
Line 173: Northward displacement of the shallow circulation. In Figure 2d, I see positive anomalies north of the shallow circulation and negative anomalies around 15N; would this be a southward displacement rather than a northward displacement?
Line 176: vertical structure and static stability: can you please show the change in static stability, i.e. the change in the vertical temperature gradient?
Line 177: Neelin et al.
Lines 210-211: Is it possible to comment here on the relative importance of the AMV and PDV in the result (does one of the two modes of variability dominate the response)?
Lines 230-234: For motivating the use of the NARI is might best to first say that you assume NARI to capture the physical mechanisms that link both the North Atlantic (you account for the tropical North Atlantic in the NARI) and the Pacific (through changes in static stability). More than assessing how they effect of the North Atlantic and Pacific SSTs can be represented by a single index, you want to assess mechanisms. There is some of the information in lines 236-240 that could be used, to have a focus on mechanisms rather than on the use of an index.
Line 245: NARI = 0.58AMV – 0.54PDV + ε. It is interesting to see that the AMV and PDV have similar effects on Sahel precipitation. Have you accounted for potential co-variability between the AMV and the PDV here?
Line 253: Noyelle et al.
Line 271: storylines
Lines 290: The caption of Figure 6b and the text is confusing here. Are the crosses the two storylines computed from the PDV/AMV space phase? It is necessary to not just rely on the NARI and to be driver-oriented instead. What does “~2/3 of the members being equal and the resultant mean values and spatial patterns nearly indiscernible” mean? How was that defined, and how did you assess similarities in the spatial patterns?
Line 309: remove “here”
Line 311-312: Please note that Monerie et al. (2019) did not use a fully coupled model.
Lines 310-313: Please also add the reference to Mohino et al. (2026).
Mohino, E., Monerie, P.-A., Mignot, J., and Bordoni, S.: An energetic perspective on the impact of the Atlantic Multidecadal Variability on the West African Monsoon, Weather Clim. Dynam., 7, 1619–1639, https://doi.org/10.5194/wcd-7-1619-2026, 2026.
Lines 234 and 356: The reference is missing (?).
The authors discuss the reliability of CESM2. The focus is mostly on the Pacific and Atlantic SSTs. In addition, it would be useful to know how CESM2 performs relative to the other models for the representation of the monsoon system. Besides, the authors show that CESM2 simulate a future decrease in precipitation across the Sahel by the end of the 21st century. This is contrast to a majority of models that simulate a future increase in central Sahel precipitation. This could be discussed.
Another point for the discussion is the possibility that the results could be model-dependent.
Citation: https://doi.org/10.5194/egusphere-2026-5597-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 275 | 188 | 34 | 497 | 83 | 49 | 48 |
- HTML: 275
- PDF: 188
- XML: 34
- Total: 497
- Supplement: 83
- BibTeX: 49
- EndNote: 48
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Review of the manuscript “Explaining the Modulation of Forced Precipitation Changes with Internal Variability Storylines: The case of decadal Sahel Precipitation Changes” by Julius Mex, Julia Mindlin, and Marlene Kretschmer.
1. General comments
In this work, the authors address the uncertainty in decadal Sahel precipitation projections in CESM2 large ensemble by means of a storyline approach. To apply it, they use previously noted SST drivers of Sahel decadal precipitation variability, namely the Atlantic Multidecadal Variability (AMV) and the Pacific Decadal Variability (PDV), and investigate their combined effects in the preindustrial control run of the model. They show that their effects add most strongly when out of phase, and that this effect can be captured by just one index, the North Atlantic Relative Index (NARI). They apply this framework to the large ensemble of CESM2 SSP3-7.0 projection and show that for the mid-XXI century, the internal variability can cancel out or double the forced response, while for the end of the XXI-century, they can at most modulate the drying coming from the forced response.
Overall, I find the manuscript a good contribution to the current discussion of the uncertainty in projections of regional climate change and on the potential role of internal variability vs the forced response. However, there are still some specific comments and technical corrections I propose the authors address to improve the quality and readability of the manuscript.
2. Specific comments
- Model-dependent results: while the discussion nicely addresses this issue, I think the abstract or title should at least mention that the manuscript is based on just one model. Currently, this is not explained until the end of the introduction.
- Ocean-mediated vs direct response to greenhouse gases (GHG) and aerosols: Although the manuscript’s main thesis is about understanding the plausible impacts of internal variability mediated by SSTs, it also deals with the forced response. In fact, in fig. 6a, we are presented with the “forced NARI” and its strong link to the Total Sahel rainfall forced response (lines 272-275). However, there is little comment on the potential “direct” (this is, not mediated by the ocean) effect of GHG and aerosols (the only I found was in line 369-371), so the reader could be left with the idea that the forced response to GHG and aerosols are only mediated by SSTs. It might be wise to: 1) further explain this separation in the introduction (lines 29-34 again seem to only be related to the ocean-mediated response) and 2) note that the high correlations of the forced NARI and Total (and western) Sahel precipitation might be hiding more than SST→ Sahel rainfall impacts.
- Addition of technical details: I suggest adding some technical details to ease the reading of the manuscript:
- Statistical significance of anomalies/regressions: I find there is no evaluation of the statistical significance of many of the plotted anomalies, regressions and correlations in the manuscript. For instance, are all regressions in Fig. 1, Fig. 2, Fig. 5a (etc.) statistically significant? Could the authors somehow mask those regressions that are not? What about the storyline anomalies in Fig. 4? Can you mask somehow the anomalies that are too uncertain? Same with supplementary figures.
- Role of an enhanced/weakened shallow meridional circulation (SMC): in line 157, the lower tropospheric anomalous omega response to positive AMV is interpreted as a strengthening of the SMC. However, there is a clear weakening – positive omega anomalies at what seems to be the location of the climatological updraft of the SMC in the model: 850 hPa – 15ºN. This is a common response to a positive phase of the AMV in models, as shown in Mohino et al. (2026): the strengthened monsoon is accompanied by a weaker SMC over the Sahara that reduces the mid-level dry-air intrusion into the Sahel, in agreement with Shekhar and Boos (2017). The monsoon shifts towards a more top-heavy monsoon structure in response to the positive phase of AMV. Conversely to what the authors conclude for the positive phase of IPV (line 172-173), the omega anomalous response is suggestive of an enhancement of the SMC: a tendency towards a more bottom-heavy monsoon, where the excess of moist static energy in the column is exported through dry-air horizontal advection at mid levels instead of through deep convection (see also Hill et al. 2017). I suggest that the authors revise the cited references and critically review their results and conclusions with respect to the SMC. This comment is also applicable to the AMV+/PDV- and AMV-/PDV+ results (lines 195-196: the AMV+/PDV- in Fig. 4d also suggests a weakening of the SMC).
- Introduction of energetic considerations on the impacts of AMV on ITCZ shifts: in lines 159-162, when discussing the ITCZ northward shift following positive interhemispheric SST differences, you might want to include works exploring the link through energetic considerations (see, among others, Kang et al. 2008; Schneider et al., 2014; Monerie et al. 2023; Mohino et al. 2026).
- Discussion of the results in the light of previous works: I was missing the inclusion of the work of Park et al. 2015 in the discussion of the results, when addressing the uncertainty in the forced response. They already used a NARI-like index to explain it in CMIP5. In addition, even though the paper by Monerie et al. 2020a is in the reference section, I fail to find discussion of this paper in the discussion of the internal variability.
3. Technical corrections
- Line 45: revise the “furthermore”
- Several references do not have a year attached to them. Please revise carefully. Here are some instances: line 49 (Findell et al.), line 177 and 218 (Neelin et al.), line 253 and 360 (Noyelle et al.)
- Line 104: than → then?
- Line 166: “Regression of sea-level pressure confirms the results from the vertical cross-sections …” In which way? Could you expand the explanation?
- Line 180-181: Not sure why Mohino et al. 2024 and Monerie et 2019 are cited here. They mainly deal with AMV. I suggest better references in this line would be: (Joshi et al., 2022; Mohino et al., 2011a, b; Pomposi et al., 2016; Rodríguez-Fonseca et al., 2011; Villamayor and Mohino, 2015).
- Line 190: “(see 2)”: do you mean “(see section 2)”?
- Please, revise quality of fig S4: arrows are difficult to interpret, not sure of some directions even when zooming. Better use EPS or similar.
- Caption of Fig. 2: I found confusing referring to the pressure tendency (omega) as vertical velocity. I would suggest using omega directly to avoid confusion with the vertical component of the velocity.
- Caption of Fig. 2: omega is in shading but also in the vertical component of the arrows, right? Please specify in the caption.
- Caption of Fig. 2: Only the pi-Control run is used, right? Better specify it in the caption.
- Caption of Fig. 2: Is the reference arrow missing for panel b?
- Caption of Fig. 3: Is this the AMV and PDV standardized index? The one obtained by Trenberth and Shea's and TPI, respectively? Could you specify it in the caption, please?
- Line 204-205: Fig S4 are world maps; be precise about where the convergence/divergence are.
- Line 207: southward → northward?
- Line 208-209: I’m not sure what you mean by “convective” and “mixed” responses in this sentence. Could you clarify, please?
- Line 215-216: Do you mean over the continent? Over the ocean, it would be the other way round. Better clarify.
- Line 223-224: “Just as the circulation anomalies, the precipitation signal is stronger over land than over ocean.” Fig. 4i seems not consistent with this statement: rainfall negative anomalies are stronger over the tropical Atlantic than over the Sahel.
- Line 244: remove “)” in “Figure 5b) shows …”
- Caption of Fig. 4: Reference arrows are missing. Is it omega that you are showing?
- Caption of Fig. 5: IVS has not been defined in the text.
- I suggest revising the references used in lines 311-313, I think there are some errors: Mohino et al. 2011b investigates the influence of Pacific SSTs and it does not use coupled models but SST-forced AGCM simulations. Monerie et al. 2019 does not use a fully coupled model, but an AGCM coupled to a slab ocean and use a DCPP-C type of sensitivity experiment (PRIMAVERA-AMV protocol, I think), as do other models (these with complete ocean, not only slab model) in Mohino et al. 2024 and 2026. For observational studies on AMV→ Sahel you can also refer to Zhang and Delworth, 2006 and Villamayor et al. 2018.
- Line 344: Reference missing: see “?”
- Line 356: Reference missing: see “(?)”
- Line 370: The closing parenthesis is missing. Maybe you should also check the guidelines for referencing inside a parenthesis; I would expect: “(e.g., Herman et al., 2020; Ndiaye et al., 2022; Nnamchi and Fiedler, 2026)”
- Line 373: Remove extra space in “… North Atlantic . We ...”
References: