the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 20 Apr 2026
RAAVEN data processing for TORUS and TORUS-LItE
Abstract. RAAVEN (Robust Autonomous Airborne Vehicle - Endurant and Nimble) uncrewed aircraft systems (UAS) were deployed in and around supercell thunderstorms during the Targeted Observation by Radars and UAS of Supercells (TORUS) and TORUS Left Flank Intensive Experiment (TORUS-LItE) field campaigns. On-board sensors measured temperature, humidity, pressure, and wind. Despite extensive predeployment testing, the demanding environments where data collection occurred presented numerous challenges to data quality. In this article, extensive quality control procedures adopted for these data are described. Many of these procedures aim to quantify data-quality uncertainty, in lieu of correcting questionable data. Procedures address the dependency of estimated wind on aircraft manoeuvring, periodically faulty sensors, questionable data induced by sensor wetting in rain, and sensor hysteresis and bias. Bulk data statistics are also presented, in part to assert data quality but also to highlight unique qualities of UAS data collected during TORUS and TORUS-LItE.
- Preprint
(3536 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- CC1: 'Comment on essd-2026-134', Weiteng Qiu, 28 Apr 2026
-
RC1: 'Comment on essd-2026-134', Anonymous Referee #1, 10 May 2026
This manuscript presents the processing, quality control, uncertainty characterization, and documentation of RAAVEN uncrewed aircraft system observations collected during the TORUS and TORUS-LItE field campaigns. The dataset includes in situ measurements of temperature, humidity, pressure, and three-dimensional wind in and around supercell thunderstorms, including challenging near-storm regions such as the near-inflow, left-flank, and right-flank sectors. The study is valuable because UAS observations in severe convective environments remain rare, technically difficult, and scientifically important. The manuscript clearly recognizes that the main contribution is not a new physical interpretation of supercell dynamics, but rather a careful description of data processing and quality control procedures for a unique observational dataset. The dataset is substantial, including 88 RAAVEN flights and 57.5 flight hours, with 71 flights directly targeting supercells. The authors also make both the data and processing code available, which strengthens the reproducibility and long-term utility of the work.
In general, I find this to be a useful and publishable data paper. The manuscript addresses an important need for transparent documentation of UAS data processing in severe-storm environments. The authors provide detailed procedures for handling sensor wetting, sensor hysteresis, humidity bias, faulty sensors, aircraft-motion effects on wind retrievals, and wind uncertainty estimation. These are all important issues for users of the dataset. The paper is therefore suitable for publication after revision. However, several aspects require clarification, expansion, and improved presentation before the dataset can be used confidently by a broad community.
- The authors correctly identify sensor wetting as a major challenge in thunderstorm UAS observations. The decision to flag questionable wetting periods rather than apply uncertain corrections is reasonable. However, the criteria for identifying wetting events remain partly qualitative. The manuscript states that typical wetting indicators include sudden temperature reductions, slower relaxation back to ambient conditions, increased relative humidity, and coincident radar-indicated precipitation. This is scientifically reasonable, but the exact thresholds or decision rules are not sufficiently formalized. The authors should clarify whether wetting identification was fully manual, semi-automatic, or automatic. If thresholds were used, they should be provided. If expert judgment was required, the manuscript should state how consistency was ensured across cases. Because wetting affects a substantial fraction of observations, especially in left-flank flights, this issue is important for downstream data users. Table 4 shows wet MHP occurrence of 18.4% for all storms and 30.7% for left-flank missions, and wet RSS-421 occurrence of 19.1% for all storms and 22.2% for left-flank missions.
- The manuscript notes that the dedicated iPTH temperature thermistor functioned only 26.6% of the time and that a backup RH temperature sensor was used after hysteresis correction when the dedicated thermistor was unavailable. This is a reasonable solution, but the resulting variable may have heterogeneous response characteristics. The manuscript should more explicitly discuss the possible consequences of switching between the dedicated temperature sensor and the corrected backup RH temperature sensor.
- The introduction could more clearly state why RAAVEN observations are complementary to mobile mesonet, radar, lidar, and balloon-borne observations. The manuscript mentions the broader TORUS observing system, but the unique role of UAS observations in filling near-surface and low-altitude sampling gaps should be emphasized more strongly. Table 1 is useful, but it would be helpful to add whether each sensor was exposed or shielded, and whether the sensor was used for final reported variables, backup variables, or only quality-control purposes. Table 2 is very large and difficult to read. Consider moving part of it to supplementary material and retaining a condensed summary table in the main manuscript.
- The humidity correction equation is important, but the notation should be checked carefully. The text should clearly define all variables, units, valid ranges, and whether temperature is in Kelvin or Celsius. Since RH values are clipped at 99.9%, the authors should discuss whether this may affect analyses of near-saturated air, especially in storm flank or precipitation-adjacent environments.
- The wind quality-control thresholds should be justified. For example, why were wind angles greater than 45 degrees, component magnitudes greater than 50 m s⁻¹, and uncertainties greater than 30 m s⁻¹ chosen as thresholds?
Citation: https://doi.org/10.5194/essd-2026-134-RC1 -
RC2: 'Comment on essd-2026-134', Anonymous Referee #2, 26 Jun 2026
This manuscript provides a valuable dataset of RAAVEN UAV observations in the core of thunderstorms. Its quality control process is scientific and transparent, and the data has been made openly accessible through a trusted data center. However, the manuscript has a significant structural flaw: it leans too heavily on an engineering-style description of quality checks and fails to fully explore and showcase the unique value of this dataset for addressing the core scientific questions in supercell meteorology. As a result, readers only understand the data processing workflow, but not the significance of the data itself.
1. The current introduction only vaguely mentions ‘studying near-surface rotation,’ which is too bland. It’s suggested to clearly point out the scientific bottleneck—for example, that existing radar and soundings cannot resolve centimeter-scale wind shear/temperature gradient evolution in the left-flank boundary layer before a tornado—and then logically justify the irreplaceability of UAS, finally leading to the critical nature of data quality. This way, the QC standards directly match scientific needs.
2. Currently, Section 6 and Figures 10/11 only list statistics showing that left-flank wind speed is higher than the inflow. It’s recommended not to just report uncertainty percentages, but to clearly tell readers that, based on this QC dataset, these observational constraints can be directly used to verify the reference value of convective parameterization in high-resolution numerical models.
3. For the ‘moisture flag’ and ‘questionable wind field flag,’ only the trigger conditions are defined, but users are not told how to use them. It’s suggested to add guidance at the end of the corresponding sections. For example, when RH is capped at 99.9%, advise users to treat this data only as a ‘saturation indicator’ rather than an exact humidity value. This kind of guidance can greatly enhance the usefulness of the dataset.
4. Was there any cross-calibration or consistency check performed between the 2019 and 2023 sensor configurations?
5. Section 5.1, Equation (2) and subsequent clipping description. The correction Equation (2) is based on humidity chamber experiments, but the humidity/temperature ranges of the experiments are not clearly stated. Users cannot judge the reliability of the correction results when input conditions exceed the experimental range. Values of 100.5% and 105% are all clipped to 99.9%, which affects the variance among samples.
6. According to the guidelines, reviewers should verify the dataset itself as much as possible. Due to time constraints, I was unable to fully download and test the complete dataset, but based on the statistics provided in the manuscript and the transparent disclosure of data quality flags by the authors, the overall quality of the dataset seems reliable.
-
RC3: 'Comment on essd-2026-134', Anonymous Referee #3, 27 Jun 2026
General Comments
The manuscript provides a detailed description of the RAAVEN UAS observations collected during the TORUS and TORUS-LItE field campaigns, including the instrument configuration, data processing procedures, quality-control methods, and uncertainty characterization. The resulting dataset contains in situ measurements in different storm-relative regions around supercell thunderstorms, which are highly valuable given the scientific importance and relative scarcity of publicly available observations in such hazardous and dynamically complex environments. Overall, the data processing and quality-control procedures appear scientifically sound and carefully implemented.
I recommend publication after revision. My main concern is related to the readability of the manuscript. In its current form, the paper reads somewhat more like a technical report than a self-contained scientific data paper. Although the manuscript is not overly long, the presentation could be improved by providing more background information on the scientific motivation, the comparison and advantages of these observations relative to other observational datasets, and how similar UAS or in situ datasets have been used in previous severe-storm studies. These additions would help make the manuscript more accessible to readers who are less familiar with this topic, while placing the dataset in a broader scientific context and better demonstrating its potential value for future research.
Major Comments
- The title contains several uncommon abbreviations but lacks more general terms that directly describe the dataset and its scientific content. This may make the manuscript less discoverable in literature searches and may also make it difficult for readers to immediately understand the main subject of the paper.
- The Introduction would benefit from a more comprehensive background description. At present, the manuscript moves rather quickly to the RAAVEN deployments and data-processing procedures, while the broader scientific context is not fully developed. The authors may consider adding some background paragraphs that introduce supercell thunderstorms and the importance of in situ observations, briefly compare conventional severe-storm observing platforms with UAS-based measurements, and explain the advantages and challenges of RAAVEN observations in hazardous near-storm environments. These additions would help readers better understand the motivation for the dataset and the specific contribution of the RAAVEN observations.
- In Section 3, the authors introduce the storm-relative mission areas used in the RAAVEN deployments, including the left flank, right flank, near-inflow, and far-field regions. However, the manuscript would benefit from a brief explanation of how these regions differ in terms of their dynamical and thermodynamical characteristics and how these differences motivated the deployment strategy. Such clarification would help readers better understand why the UAS observations were organized around these specific mission areas and how the resulting data may be used to examine different storm-scale processes.
- Since different sensor configurations were used in 2019 and 2023, the authors should clarify whether any consistency check was performed between the two in situ observational datasets.
- The criteria used for wind quality control and subsequent statistical analyses should be better justified. It would be helpful for the authors to clarify whether these thresholds are based on instrument limitations, previous studies, empirical testing, or sensitivity analyses, and whether the main conclusions are sensitive to these choices.
Minor Comments
- Table 2 is very long and difficult to read in the main text. The authors may consider moving it to the supplementary material or the data repository webpage. If possible, basic information on each observed storm, such as its lifecycle, development stage, and intensity, could also be added.
- Please check all abbreviations throughout the manuscript and ensure that they are defined at first use.
- The caption of Figure 8 could be made more detailed to improve readability.
- Please briefly explain the apparent abrupt change shown in Figure 11f.
- The summary section could include a brief outlook on future observational plans related to this project and discuss whether further intercomparisons with other TORUS observing platforms are planned to further evaluate the RAAVEN dataset.
- Please ensure that the Data Availability statement is presented as a separate section before the summary section, following the ESSD format requirements.
- Please carefully check the formatting of the references for consistency with the journal style.
Data sets
RAAVEN TORUS-LItE data Adam L. Houston et al. https://doi.org/10.26023/E5PJ-5CN7-VQ0T
RAAVEN TORUS data Adam L. Houston et al. https://doi.org/10.26023/FJD8-VMV2-XW0Y
Model code and software
Python functions for smoothing, hysteresis correction, and iPTH RH correction Mark De Bruin https://github.com/markdebrstorms/UAS_Processing
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 316 | 50 | 20 | 386 | 17 | 16 |
- HTML: 316
- PDF: 50
- XML: 20
- Total: 386
- BibTeX: 17
- EndNote: 16
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript presents a comprehensive description of data processing and quality control procedures for measurements collected by the RAAVEN uncrewed aircraft system during the TORUS and TORUS-LItE field campaigns. The dataset includes high-resolution in situ observations of temperature, humidity, pressure, and wind collected within and around supercell thunderstorms. This is a high-quality data paper. The dataset and quality control framework will be valuable for the community. I recommend this paper for minor revision.
Comments