Observations of wind ramps generated by thunderstorm outflows

Abraham, Aliza; Bodini, Nicola; Hamilton, Nicholas; Hirth, Brian; Schroeder, John; Moriarty, Patrick

doi:10.5194/wes-2026-82

Preprints

https://doi.org/10.5194/wes-2026-82

Preprints

08 May 2026

| 08 May 2026

Status: this preprint is currently under review for the journal WES.

Observations of wind ramps generated by thunderstorm outflows

Aliza Abraham, Nicola Bodini, Nicholas Hamilton, Brian Hirth, John Schroeder, and Patrick Moriarty

Abstract. Thunderstorms occur frequently in regions with strong wind resources, such as the U.S. Great Plains. The outflows from these storms generate wind ramps, where the wind speed and direction change dramatically over a short period. Wind turbines are designed to withstand wind ramps such as the "extreme coherent gust with direction change" (ECD) defined by the International Electrotechnical Commission (IEC), but this depiction is highly simplified and based on point measurements. In the current investigation, we characterize the wind ramps generated by six thunderstorm outflows observed in northern Oklahoma using dual-Doppler X-band radar measurements covering a 160 km³ volume, and compare their properties with the IEC-defined ECD. We show that, while the observed wind ramps exhibit wind speed and direction changes with similar magnitudes to the ECD, the observed rise times are an order of magnitude longer. Furthermore, the observed events are more complex than the idealized ECD profile, displaying spatial and temporal variations that cause significant variability in wind turbine power generation.

Received: 30 Apr 2026 – Discussion started: 08 May 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

Aliza Abraham, Nicola Bodini, Nicholas Hamilton, Brian Hirth, John Schroeder, and Patrick Moriarty

Status: final response (author comments only)

Subscribe to comment alert

RC1:
'Comment on wes-2026-82', Anonymous Referee #1, 06 Jul 2026
General comments

The manuscript proposes a novel method using remote sensing technology to determine Extreme Coherent Gust with Direction change (ECD) parameters. The manuscript provides adequate background and derives standardized parameters based on lidar, SCADA, and in-situ measurements. The proposed methodology and procedures are generally sound and well-described. The findings reveal the spatial variability of this type of complex flow and demonstrate how these events deviate from the IEC standard, posing unique challenges for wind farm power generation. We recommend this manuscript for publication following the adaptations listed below.

Specific comments

Because the manuscript relies on data from a large field campaign, detailed information regarding data treatment is essential. While the authors describe the filter thresholds adopted for the lidars, the manuscript lacks information on data filtering for the in-situ measurements (3.3), turbine anemometers (3.4) and, more importantly, the radars. As the radar measurements exhibit some data vacancies (shown already in Fig. 4), we suggest elaborating on the filtering and rejection criteria, as well as any interpolation methods if used. Please give more details than line 94.

Section 4 notes that 26 wind ramp events were identified but only 6 are analyzed in detail, with the selection based on concurrent precipitation observations and sufficient dual-Doppler radar coverage. While the authors acknowledge this is not a frequency assessment, we would like to suggest the authors briefly discuss whether this selection criterion could bias the reported ramp characteristics and clarify how many events were excluded due to radar data quality (see comment 1) versus the non-precipitation atmosphere.

There appear to be several Dual-Doppler radar synthesis techniques available, while the manuscript only briefly mentions that radial wind speeds are interpolated onto a three-dimensional Cartesian grid (lines 95–99). Extending from Comment 1, more elaboration should be appended here. Each individual radar generates radial wind speeds that are not uniform in spatial resolution. The further the measurement is from the radar, the greater the spacing between adjacent measurement points. The 0.5 degree across-beam resolution (line 91), the azimuthal spacing at the far edge of the ~30 km range is approximately 260 m (~ 5 times coarser than the 50 m horizontal grid resolution). Is it true that the synthesis data have a regular horizontal resolution of 50 m and a vertical resolution of 25 m across the full radar domain? How representative is a 50 m grid at far ranges? We suggest the authors elaborate on how the synthesis process accounts for this range-dependent resolution effect, and clarify whether the effective spatial resolution varies across the domain.

As noted in the manuscript line 100, the dual-Doppler radars are not synchronized and perform relatively slow scans, introducing a temporal ambiguity of 2 to 4 minutes. In the context of gust propagation, this time lag is not trivial (a gust moving at 15 m/s travels 1.9 km within this time period). While the naturally longer rise times observed make physical sense, calculating the rise time by combining a fit-derived spatial rise distance (Sect. 6.2) with an ensemble-averaged boundary propagation speed (Sect. 6.1) introduces uncertainty. We recommend adding a discussion paragraph to address these uncertainties or providing error estimates, particularly when comparing these space-time converted results to the standard IEC temporal method. In this regard, we note that the propagation speed is itself derived from volume-to- volume boundary displacements during this 2 to 4 minute interval (Sect. 6.1, lines 180–185). The temporal resolution limitation is therefore still present, as it is directly inherited through propagation into all reported rise time values in Table 2. Furthermore, propagation speed is computed as an averaged quantity while boundary displacement is a local measurement, applying a single mean propagation speed across the entire boundary implicitly assumes spatially uniform flow. More elaboration on this limitation and how it enters the resultant rise time should be given/quantified.

In Fig. 10(i), several bins exhibit near-zero Δt_{θ} values, which appear physically implausible given the spatial resolution and propagation speeds. The manuscript does not describe any quality control applied to the curve-fitting results such as goodness-of-fit thresholds or lower rejection bounds. It is unclear whether these near-zero values represent genuine sharp wind direction transitions, degraded fitting quality, or artifacts of using the wind speed boundary to define the perpendicular cross-section direction in a case where the wind speed and direction boundaries are non-coincident (as is the case for 2023-06-16, line 164). The authors should clarify the curve-fitting acceptance criteria and discuss whether these values are physically meaningful or should be excluded.

As a general comment for the methodology used, Fig. 8 and Fig. 10 (f)(g)(h)(i) demonstrate large variability in radar-derived ramp parameters along the outflow boundary. The manuscript attributes some of this variability to inhomogeneous flow features (lines 327–329), no metrics are provided to distinguish physically driven variability from methodological ones , such as the range-dependent increase in effective radar volume (noted in Comment 2) and the absence of quality control of the curve-fitting procedure (Sect. 6.2). We suggest the authors provide fitting quality metrics or discuss how these methodological factors may contribute to the observed variability.

The radar-based method extends gust quantification into the spatial domain, and it is clear this approach is not expected to compare one-on-one to the standard point measurement method. To strengthen the paper's practical impact, we suggest explicitly outlining the operational advantages of using this proposed spatial method (predictive wind farm control…etc.) alongside a discussion of its limitations regarding space-time volume averaging.

Technical corrections

Line 57: “Section 2” is inconsistent with “Sect. X” used in other texts.

Equation (3): should use smaller or equal (<=) for the first condition. Otherwise, 4 m/s has not been included in either condition.

Line 129 states that events are identified when the wind speed changes "at least 4 m/s" between adjacent 10-minute periods, while the Figure 3 caption describes a "ramp with wind speed change greater than 5 m/s". We suggest aligning these thresholds.

Line 164: … propagate (south) eastward…: need a space, or south-eastward

Line 190: “the 2024-05-24 (case study?) exhibits the largest deceleration”. This sentence seems to miss a word.

Fig. 10 captions: Site C1a and Site D seem to be inconsistent

Line 352-353: Suggest using standard terminology “cut-out” instead of cut-off.

Line 405 to 457: Several references have twice https://doi.org.
Citation: https://doi.org/10.5194/wes-2026-82-RC1
RC2:
'Comment on wes-2026-82', Anonymous Referee #2, 30 Jul 2026
This manuscript presents a detailed analysis of thunderstorm outflow-induced wind ramps based on multi-lidar observations. The dataset is valuable, and it is clear that a significant amount of work has gone into the data processing, event characterization, and analysis. In particular, the results concerning the spatial variability of the events and the apparent mismatch between changes in wind speed and wind direction are interesting and potentially novel. However, the manuscript would benefit from substantial restructuring, a clearer statement of its scientific objectives, and a more careful interpretation of the comparison with the IEC Extreme Coherent Gust with Direction Change (ECD).
Major comments
1. Clarify the research objectives and contribution

At present, I find it difficult to identify the main research question of the study and how it is specifically connected to wind energy applications. While the manuscript contains a large amount of analysis, the scientific focus is not sufficiently clear. I encourage the authors to formulate clear research questions early in the manuscript and ensure that the subsequent analysis is structured around answering them.

In my view, the most novel aspects of the work are the observed spatial variability of the events and the mismatch in timing and rise time between wind speed and wind direction changes. These findings deserve greater emphasis and could form the central focus of the paper. In contrast, the comparison with the IEC ECD appears less convincing in its current form and tends to dominate the discussion despite a limited basis for comparison.
2. Interpretation of the comparison with IEC extreme event

Several parts of the manuscript imply a direct comparison between the six observed thunderstorm outflow events and the ECD. I do not believe such a comparison is fully justified without additional context.

The IEC ECD is not intended to represent a thunderstorm outflow event. Rather, it represents an unspecified extreme wind event with a recurrence period of 50 years. By contrast, the events analyzed in this study are individual observed cases whose recurrence periods are unknown and are not demonstrated to correspond to 50-year events. This important distinction should be clearly acknowledged throughout the manuscript.

Related to this point, the statement that the ECD represents a "worst-case scenario" is not strictly correct. The ECD represents an extreme design event. Because the relevant parameters are modeled together and are not necessarily independent, the fact that one observed parameter exceeds the corresponding ECD value does not imply that the event exceeds the design condition as a whole. The manuscript should avoid such interpretations.

More generally, it is unclear what the authors intend to demonstrate through the ECD comparison. Are they attempting to show that the ECD is insufficiently conservative, or simply that real events exhibit different characteristics (which has already been done)? Such questions cannot be convincingly addressed using statistics from only six events and should therefore be discussed more cautiously.
3. Literature context

The discussion would benefit from additional references.

The manuscript cites Kelly et al. in a manner that could be interpreted as a study specifically of thunderstorm gust fronts. However, Kelly et al. investigated extreme wind ramps more broadly, a category that includes but is not limited to thunderstorm gust fronts. The corresponding text should be revised to better reflect the scope of that work.

Another relevant study is Hannesdóttir et al. (https://wes.copernicus.org/articles/8/231/2023/), which investigates extreme coherent gusts and associated turbine loading and includes extrapolation to 50-year return levels. This perspective is particularly relevant to the discussion of extreme-event characterization and design implications.

Furthermore, the statement that shorter-timescale changes may be more damaging to turbines because they provide less time for control-system response is supported by previous work, including Hannesdóttir et al.(https://wes.copernicus.org/articles/8/231/2023/). Adding references here would strengthen the discussion, which currently reads somewhat speculatively.
4. Methodological clarification

Several aspects of the methodology require further clarification:
The manuscript refers to an edge/boundary detection methodology. Is this an existing method from the literature, and if so, please provide an appropriate citation (other than the scikit Python library)? If it is a new adaptation, this should be clearly stated.
5. Characterization methodology

Line 271 states: "Having developed a method for characterizing thunderstorm outflow-induced wind ramps..."

This wording overstates the contribution. The authors appear to have modified and applied an existing characterization framework rather than developed an entirely new method. This statement needs rephrasing to more accurately reflect the contribution.
6. Discussion and conclusion

The discussion and the conclusions become somewhat unfocused. I suggest shortening these sections and concentrating on the primary findings that are directly supported by the presented analysis.

The conclusions introduce several new topics that are not sufficiently developed in the main manuscript.

The final paragraphs of the conclusion introduce several new concepts, including wind farm control strategies, forecasting applications, economic impacts, grid resilience. These topics have not been discussed in the preceding sections and therefore feel disconnected from the rest of the manuscript. I recommend removing most of this material from the conclusion. The conclusion should focus on the main findings of the present study and their direct implications.
Minor comments
Line 105: The sentence describing the lidar scan duration is unclear. It is not evident whether the approximately 5 s refers to the complete scan sequence or only the vertical beam measurement. Please rephrase for clarity.

The carrier-to-noise ratio threshold differs between the two lidars. Please explain the reason for using different thresholds.

In Table 2, some boundaries appear to be defined using wind direction and others using wind speed. The rationale for this choice is not clear and should be explained.

Overall assessment

The manuscript contains a valuable dataset and several interesting analyses. In particular, the spatial structure of the events, their variability, and the apparent mismatch in timing and rise rates between speed and direction changes appear to provide genuinely new insights. However, the manuscript would benefit from a clearer research focus, a more concise structure, and a more cautious treatment of comparisons with IEC design conditions. I encourage the authors to strengthen the motivation and novelty around the observed spatiotemporal characteristics of the events and reduce the emphasis on direct comparisons with the ECD, which are difficult to interpret given the limited sample size and the fundamentally different purposes of the observed events and the IEC design model.
Citation: https://doi.org/10.5194/wes-2026-82-RC2

Aliza Abraham, Nicola Bodini, Nicholas Hamilton, Brian Hirth, John Schroeder, and Patrick Moriarty

Viewed

Total article views: 278 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
205	61	12	278	19	17

HTML: 205
PDF: 61
XML: 12
Total: 278
BibTeX: 19
EndNote: 17

Views and downloads (calculated since 08 May 2026)

Month	HTML	PDF	XML	Total
May 2026	105	36	10	151
Jun 2026	41	12	0	53
Jul 2026	59	13	2	74

Cumulative views and downloads (calculated since 08 May 2026)

Month	HTML	PDF	XML	Total
May 2026	105	36	10	151
Jun 2026	41	12	0	53
Jul 2026	59	13	2	74

Viewed (geographical distribution)

Total article views: 251 (including HTML, PDF, and XML) Thereof 251 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 31 Jul 2026

Short summary

Thunderstorms can cause sudden changes in wind speed and direction called "wind ramps". While wind turbines are designed to withstand simplified versions of such wind ramps, this paper shows that real wind ramps are much more variable and complex than those prescribed in the design standard. This variability makes it difficult for wind farm operators to predict the conditions that each wind turbine will experience, adding uncertainty to the prediction of wind farm power.


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