Convective environments prone to downbursts, large hail, short-duration precipitation extremes and tornadoes in Spain

Barrio-Martin, Andres; Calvo-Sancho, Carlos; Plaza-Martin, Nuria P.; Azorin-Molina, Cesar; Prein, Andreas F.; McVicar, Tim R.; Vicente-Serrano, Sergio M.; Gimeno, Luis; Nieto, Raquel; Zeng, Zhenzhong; Morata, Ana; Chen, Deliang

doi:10.5194/egusphere-2026-5202

Preprints

https://doi.org/10.5194/egusphere-2026-5202

Preprints

09 Sep 2026

| 09 Sep 2026

Status: this preprint is open for discussion and under review for Natural Hazards and Earth System Sciences (NHESS).

Convective environments prone to downbursts, large hail, short-duration precipitation extremes and tornadoes in Spain

Andres Barrio-Martin, Carlos Calvo-Sancho, Nuria P. Plaza-Martin, Cesar Azorin-Molina, Andreas F. Prein, Tim R. McVicar, Sergio M. Vicente-Serrano, Luis Gimeno, Raquel Nieto, Zhenzhong Zeng, Ana Morata, and Deliang Chen

Abstract. Deep moist convection produces several hazards including downbursts, tornadoes, large hail, and short-duration precipitation extremes, resulting in fatalities and substantial economic losses. This study employs observational reports, lightning flashes and the ERA5 reanalysis over Spain (2015–2024) to summarize the key environmental differences between non-severe convection and each of these hazards, by ranking 203 convective parameters according to their discriminatory skill, and comparing composite vertical profiles. Summer downburst environments are markedly warmer than non-severe environments, with downdraft Convective Available Potential Energy (CAPE) emerging as the most discriminating parameter. Non-summer downbursts show enhanced wind shear relative to non-severe convection, though rarely reaching High-Shear–Low-CAPE conditions. Summer tornadoes and waterspouts display higher 0–3 km CAPE and a lower level of free convection than non-severe convection, but non-summer cases overlap substantially with non-severe environments. Hail over 5 cm in diameter occurs mostly in summer, in environments of elevated CAPE and shear, best distinguished by the WMAXSHEAR composite index. Short-duration precipitation extremes, evaluated during non-summer months, are associated with elevated moisture and some supercell signatures, supporting slow motions. These results provide a comprehensive framework for future studies on severe convective weather and climate model projections, representative of the Mediterranean severe convective environments.

Received: 31 Aug 2026 – Discussion started: 09 Sep 2026

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, 2439 KB)

Supplement (517 KB)

Download & links

Status: open (until 21 Oct 2026)

Post a comment Subscribe to comment alert

RC1:
'Comment on egusphere-2026-5202', Anonymous Referee #1, 19 Sep 2026 reply
The manuscript “Convective environments prone to downbursts, large hail, short-duration precipitation extremes and tornadoes in Spain” presents a useful analysis of severe convective environments during 2015-2024. The authors combine observational reports, lightning data, and ERA5 model-level profiles to compare different hazards using the same methodology. I particularly appreciate the geographical selection of the non-severe samples and the combination of convective parameters with composite vertical profiles.
The results are interesting, especially the seasonal differences in downburst environments, the better discrimination of large-hail environments, and the difficulties in separating tornado and waterspout environments from non-severe convection. The manuscript also discusses relevant limitations of the observations and reanalysis. Although the manuscript is well structured, interesting and as I have said the paper includes valuable results, some shortage can be noted. In general, the work is well described. The results section shows a good structure. Quality of figures must be enhanced, enlarging the legends in each figure.
The manuscript deserves publication after minor revision. Here are my main objections:
Major comments
At lines 218-228, the resampling procedure selects different non-severe subsets, while the severe sample seems to remain the same. Please indicate whether sampling is performed with replacement and confirm whether the severe sample is fixed. If this is the case, the reported standard deviation describes the variability associated with selecting the non-severe cases, but not the full sampling uncertainty of the area under the AUC. Please explain this in the Methods and table captions. Small AUC differences between the best parameters should be interpreted with caution. It would also be more precise to describe their ability to distinguish the samples analyzed here, since their forecasting performance has not been evaluated with independent data.

The use of lightning data and the geographical restriction in Sections 2.3 and 2.5 are useful aspects of the methodology. However, some details need further explanation. Please indicate how the ERA5 time before convection is selected for non-severe events. Does the geographical restriction use the locations of severe events in each season, or all locations for that hazard during the study period? Please also explain how multiple reports from the same storm or episode were treated. The problem of unreported severe events is already discussed in the manuscript. Still, it would be useful to explain from the beginning that “non-severe” refers to convection detected by lightning without a nearby severe weather report.

At lines 203-206, please explain what “statistically poor results” means for summer precipitation extremes and non-summer large hail. A small sample size and a low ability to distinguish environments are different limitations. The explanation for retaining summer tornadoes, based on their more physically reliable patterns, may give the impression that the categories were selected according to the results obtained. Please explain this decision more clearly. A short description of the excluded comparisons could be included in the supplement.

Some physical interpretations need more careful wording. At lines 284-289, the low discrimination of kinematic parameters does not necessarily mean that storm-scale pressure perturbations make a small contribution to summer downbursts. At lines 357-363 and 478-480, greater low-level CAPE and a lower LFC are consistent with conditions favorable for stretching pre-existing vertical vorticity, but they do not identify the main tornadogenesis mechanism. Similarly, the precipitation analysis at lines 415-438 and 485-489 suggests conditions favorable for supercells and slower storm motion, but does not directly evaluate storm type, persistence, or precipitation efficiency. These explanations are reasonable, but please present them as possible interpretations of the environmental differences. Also, please make clear that storm motion is estimated from the environmental profiles rather than observed.

Specific comments
Section 2.5, Figure 1, and Tables 5-8: Please check the sample sizes. Figure 1 reports 276 tornado/waterspout events, while Table 7 includes 66 summer and 211 non-summer events, giving a total of 277. Section 2.5 mentions 67 summer cases. Similarly, Figure 1 reports 161 precipitation events and Section 2.5 indicates 58 summer cases, which would leave 103 non-summer cases. However, Table 8 reports 105. If these differences result from filtering or missing profiles, please explain this and distinguish the initial and final sample sizes. Please also include the number of cases in each summer downburst subset in Table 5.

Lines 168-174: Clarify which profile is selected for severe reports without nearby lightning. The text sets (t_0) to the hour before the report, but the previous sentence indicates that the profile is extracted during the hour before (t_0). It is therefore unclear whether the profile corresponds to one or two hours before the report.

Section 2.5: Describe the Mann-Whitney U tests in the Methods. Indicate whether they are one-sided or two-sided, which samples are compared, and whether any correction for multiple comparisons was applied.

Lines 339-342: The text describes the DCAPE distributions of non-summer downbursts and non-severe environments as “identical” and gives “p < 0.05.” However, Figure 5e indicates that the difference is not statistically significant. Please check this result. If the difference is not significant, this should be stated instead of describing the distributions as identical.

Tables and figures: Please check the differences in some AUC values. For example, shear from the surface to the -10 °C level has an AUC of 0.69 in Table 6 and 0.71 in Figure 5a. Do the tables show the average from the resampling procedure and the figures the value for the full sample? If so, please explain this. Otherwise, please correct the values.

Lines 362-363: Check “the prevalence of this mechanism over stretching,” since the previous sentence already refers to stretching. The intended comparison is unclear.

Abstract: It would be useful to include one or two numerical results, for example the AUC of 0.80 for the complete summer downburst sample and 0.81 for summer large hail. Please also reconsider “rarely reaching” high-shear-low-CAPE conditions, since the Results indicate that 40% of non-summer downbursts meet these criteria.

Lines 491-495: The results are relevant for understanding Mediterranean severe convection, but the study does not evaluate whether they are representative of the whole Mediterranean basin. Please describe them as a reference for Spain that may also be useful for other Mediterranean regions.

Table 4: The table uses strict inequalities on both sides of the threshold, so values equal to the threshold are not assigned to either group. Please explain how these values are treated, especially for parameters with many zero values.

Figure 5: The caption refers to “non-severe summer thunderstorms,” although this figure presents non-summer downbursts. Please correct this.

Reply
Citation: https://doi.org/10.5194/egusphere-2026-5202-RC1

Supplement

https://doi.org/10.5194/egusphere-2026-5202-supplement

Viewed

Total article views: 362 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
223	106	33	362	28	33	32

HTML: 223
PDF: 106
XML: 33
Total: 362
Supplement: 28
BibTeX: 33
EndNote: 32

Views and downloads (calculated since 09 Sep 2026)

Month	HTML	PDF	XML	Total
Sep 2026	223	106	33	362

Cumulative views and downloads (calculated since 09 Sep 2026)

Month	HTML	PDF	XML	Total
Sep 2026	223	106	33	362

Viewed (geographical distribution)

Total article views: 362 (including HTML, PDF, and XML) Thereof 362 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 29 Sep 2026

Short summary

We identified the atmospheric conditions that distinguish ordinary thunderstorms from four severe weather hazards, highlighting numerical proxies for their modelling in the Mediterranean. Severe winds occur in warm, unstable air (summer) or with strong mid-level winds (other seasons); tornadoes are favored by instability near the ground; large hail occurs when instability and strong winds combine; and short-duration heavy precipitation is favored by high moisture and slow-moving thunderstorms.


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