A model hierarchy to investigate heatwave persistence: From climate models to storm-resolving simulations

Dolores-Tesillos, Edgar; Pothapakula, Praveen K.; Zeman, Christian; Domeisen, Daniela I. V.

doi:10.5194/egusphere-2026-5758

Preprints

https://doi.org/10.5194/egusphere-2026-5758

Preprints

20 Oct 2026

| 20 Oct 2026

Status: this preprint is open for discussion and under review for Weather and Climate Dynamics (WCD).

A model hierarchy to investigate heatwave persistence: From climate models to storm-resolving simulations

Edgar Dolores-Tesillos, Praveen K. Pothapakula, Christian Zeman, and Daniela I. V. Domeisen

Abstract. Heatwave persistence is shaped by processes ranging from large-scale circulation to convection and land-atmosphere interactions. In conventional global climate models, deep convection remains parameterized, whereas storm-resolving simulations explicitly represent more of this variability. How grid spacing and convection treatment affect simulated heatwave persistence, however, remains unclear.

We investigate these sensitivities using a hierarchy of atmosphere-only ICON simulations comprising a storm-resolving 2.5 km configuration, two intermediate-resolution 10 km configurations that differ in the treatment of deep convection, and a coarser 40 km configuration with parameterized convection. Heatwave occurrence and persistence, land-atmosphere relationships, post-heatwave precipitation conditions, and blocking-heatwave associations are evaluated against ERA5. All ICON configurations reproduce the broad spatial structure of heatwave occurrence, but agreement with ERA5 does not improve systematically with finer grid spacing and remains strongly region-dependent. Differences are clearer for persistence: ICON-2.5 km overestimates very persistent heatwaves in several regions, with upper-tail exceedance probabilities reaching approximately 40 % above ERA5 in the tropics. The tropical Americas show the strongest sensitivity to convection treatment. There, the 2.5 km and 10 km configurations without parameterized deep convection exhibit positive persistence biases and substantially drier conditions following heatwave termination, whereas configurations with parameterized deep convection are wetter and closer to ERA5. The contrast between the two 10 km simulations shows that convection treatment can strongly affect heatwave persistence even at fixed grid spacing. In the extratropics, where large-scale circulation plays a stronger role, regional heatwave underestimation is more consistently associated with underestimated blocking frequency or a weaker blocking-heatwave relationship.

These results show that there is no single mechanism through which finer resolution or more explicit convection modifies heatwaves. Greater process explicitness does not necessarily reduce heatwave biases; instead, changes in convection, precipitation, land-atmosphere coupling, and persistent large-scale atmospheric circulation can reinforce or compensate each other in a regionally dependent manner. Storm-resolving simulations therefore provide important process insights, but improved process representation should not be equated with systematically improved climatological agreement.

Received: 22 Sep 2026 – Discussion started: 20 Oct 2026

Competing interests: At least one of the (co-)authors is a member of the editorial board of Weather and Climate Dynamics.

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Preprint (PDF, 7580 KB)

Supplement (4485 KB)

Download & links

Edgar Dolores-Tesillos, Praveen K. Pothapakula, Christian Zeman, and Daniela I. V. Domeisen

Status: open (until 15 Dec 2026)

Post a comment Subscribe to comment alert

RC1:
'Comment on egusphere-2026-5758', Anonymous Referee #1, 09 Nov 2026 reply
General comments
In this manuscript, the authors investigate how the representation of heatwaves and their persistence varies in numerical models depending on horizontal resolution and the treatment of convection. They examine a wide range of horizontal resolutions, from standard climate model grid spacing (40 km) to storm-resolving level (2.5 km). The authors suggest that heatwave characteristics do not exhibit systematic changes depending on horizontal resolution or the treatment of convection, rather show distinct regional dependence.
Recently, many studies have highlighted the advantages of storm-resolving models over conventional climate models. The present study offers an important caution by arguing that higher spatial resolution and explicit convection do not necessarily lead to improvement in the representation of heatwaves. However, because heatwaves are driven by the complex interplay of numerous physical processes, drawing such a conclusion based on datasets of only 4 years seems premature. Furthermore, the analyses presented here remain rather superficial. Thus, I suggest rejection at this stage. The datasets analyzed in this study are undoubtedly valuable, and I strongly encourage the authors to conduct further investigations that make better use of them and resubmit their work in the future.

Specific comments
I understand that conducting long-term simulations with a storm-resolving resolution is virtually impossible at this time. However, as noted in the introduction of this manuscript and in previous studies (e.g., Barriopedro et al., 2023), the onset and persistence of heatwaves involve complicated interactions among multiple physical processes. Furthermore, it has also been pointed out that results can significantly differ depending on the indices chosen (Barriopedro et al., 2023). Therefore, addressing this topic with a data period of only four years appears inadequate. A key finding of this manuscript is the lack of a systematic response to changes in horizontal resolution and the treatment of convection, but we can not deny the possibility that such systematic signals are simply masked by noises due to the small sample size. I suggest utilizing these valuable datasets to investigate simpler underlying processes first, and then move onto more complicated events, such as heatwaves. For instance, while De Luca et al. (2024) suggested in an idealized setup that storm-resolving models can exhibit a high performance in representing blocking highs, they did not conduct simulations with a realistic land-sea configuration. Also, they did not investigate the sensitivity to the presence of convective parameterization.

The manuscript addresses too many topics. In the conclusion section, as many as five separate conclusions are presented, which obscures the main focus of the study. Also, the authors describe the comparison with ERA5 too much in detail. The primary strength of this work lies in the straightforward identification of the effects of varying grid spacing and the treatment of convection, the manuscript should focus more on the comparison between the ICON simulations. Differences among the ICON experiments themselves should be thoroughly evaluated and the statistical significance should be evaluated, rather than describe their characteristics through using ERA5 as the reference field.

The analysis and discussion regarding the comparison among the ICON simulations are superficial. In-depth analyses are needed to deepen our understanding. Specifically, the manuscript should examine how convection and the associated latent heat release differ among the ICON simulations and give interpretations on the differences. Furthermore, although the future need for event-based analyses and process-oriented evaluations is mentioned in Section 4 "Summary and discussion", such analyses should be conducted in paper. Additionally, to clarify the model response to changes in resolution and convection schemes, I recommend carrying out the simulations suggested in lines 583–586 as part of the present study.

Barriopedro, D., García-Herrera, R., Ordóñez, C., Miralles, D. G., & Salcedo-Sanz, S. (2023). Heat waves: Physical understanding and scientific challenges. Reviews of Geophysics (Washington, D.C.: 1985), 61(2), e2022RG000780. https://doi.org/10.1029/2022rg000780
De Luca, P., Jiménez-Esteve, B., Degenhardt, L., Schemm, S., & Pfahl, S. (2024). Enhanced blocking frequencies in very‐high resolution idealized climate model simulations. Geophysical Research Letters, 51(22), e2024GL111016. https://doi.org/10.1029/2024gl111016
Stevens, B., Satoh, M., Auger, L., Biercamp, J., Bretherton, C. S., Chen, X., et al. (2019). DYAMOND: the DYnamics of the Atmospheric general circulation Modeled On Non-hydrostatic Domains. Progress in Earth and Planetary Science, 6(1), 1–17. https://doi.org/10.1186/s40645-019-0304-z

Minor comments and technical corrections:
Line 165: I am wondering if the authors could conduct longer-term simulations with the 10-km resolution configurations. The longer-term simulations might yield statistically robust insights regarding the effect of explicitly resolving convection, which is one of the primary objectives of this study.

Lines 17–19: Please also describe the impact of varying grid spacing and the treatment of convection for the extratropics.

Line 24: The abstract alone does not make it clear what the phrase “climatological agreement” means. Please rephrase this.

Lines 54–82: Considering the processes described here (e.g., stratification, inversions), vertical resolution can also be a key factor. While investigating different vertical resolutions may be beyond the scope of this paper, a discussion on the importance of vertical resolution should be included.

Line 90: Are these truly "first generation"? For example, comparison projects such as DYAMOND (Stevens et al., 2019) already exist. Please clarify what is meant by "first generation."

Lines 110–116: Please clarify how the 5 concluding points correspond to the 3 questions posed in the introduction.

Line 139: Although ERA5 is widely regarded as a reliable reanalysis, how large are the discrepancies among atmospheric reanalyses (e.g., differences between ERA5 and JRA 3Q)? Because heatwave persistence involves multiple processes, the discrepancies among reanalysis products could be non-negligible. In particular, are the characteristics of rare events (such as those analyzed in Figure 2) quantitatively consistent across reanalysis products? If the spread among reanalyses is non-negligible, relying solely on ERA5 as a reference might be inappropriate.

Lines 154–158, 173–175, 265–266: The discontinuity across the equator and the mismatch between the target period and the warmest season of each local region are mentioned repeatedly. Mentioning this once (or twice) should be sufficient.

Lines 195–199: It should be stated explicitly in this paragraph that the thresholds for identifying blocking are determined independently for each dataset (as mentioned later in line 213). This is essential when comparing datasets that exhibit different biases in Z500 variance (as noted in line 184).

Figure 3: Why are red colors used for negative values and blue colors for positive values? Unless there is a specific reason, reversing the color convention would avoid potential misinterpretation.

Figure 3 Caption: It would be, for example, “spatial distributions of temporal correlations” rather than “spatial correlation”

Reply
Citation: https://doi.org/10.5194/egusphere-2026-5758-RC1

Edgar Dolores-Tesillos, Praveen K. Pothapakula, Christian Zeman, and Daniela I. V. Domeisen

Supplement

https://doi.org/10.5194/egusphere-2026-5758-supplement

Edgar Dolores-Tesillos, Praveen K. Pothapakula, Christian Zeman, and Daniela I. V. Domeisen

Viewed

Total article views: 355 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
220	107	28	355	74	36	32

HTML: 220
PDF: 107
XML: 28
Total: 355
Supplement: 74
BibTeX: 36
EndNote: 32

Views and downloads (calculated since 20 Oct 2026)

Month	HTML	PDF	XML	Total
Oct 2026	191	88	23	302
Nov 2026	29	19	5	53

Cumulative views and downloads (calculated since 20 Oct 2026)

Month	HTML	PDF	XML	Total
Oct 2026	191	88	23	302
Nov 2026	29	19	5	53

Viewed (geographical distribution)

Total article views: 355 (including HTML, PDF, and XML) Thereof 355 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 10 Nov 2026

Short summary

We tested whether increasing climate-model resolution and changing how convection is represented improve the simulation of heatwaves. Using a hierarchy of global atmosphere-only simulations, we found that finer resolution does not consistently reduce heatwave biases. Heatwave persistence depends on regional interactions among convection, rainfall, land conditions, and large-scale circulation, showing that higher resolution alone is not enough.


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