The DOI matches its Crossref record; 28 of the 29 references it cites were found in a registry or an index; 6 of the 18 sources its claims cite were checked for support.
Advancing Data Quality Assurance with Machine Learning: A Case Study on Wind Vane Stalling Detection
Open accessAbstract
High-quality observational datasets are essential for climate research and models, but validating and filtering decades of meteorological measurements is an enormous task. Advances in machine learning provide opportunities to expedite and improve quality control while offering insight into non-linear interactions between the meteorological variables.
Overview
CoChat Summary
TL;DR
Machine-learning methods can help detect wind vane stalling at Cabauw, with KNN and Random Forest performing best among supervised methods and One-Class SVM promising because it avoids data imbalance. Wind vane stalling at 10 m occurred 4.54% of the time and was linked to low wind speeds under stable nocturnal conditions.
Key findings
- Wind vane stalling at 10 m occurred 4.54% of the time over 2001–2022; across all heights, 1456 events were reported (315 at 10 m; 206, 282, 213, 224, and 216 at 20–200 m), with roughly 3–4% stalling per height (cumulative ∼11–15 days annually).
- Stalling was most frequent at 10 m, predominantly nocturnal (18–6 UTC, approximately 90% of the time), peaking in September (∼20% of the month on average), and associated with low wind speeds (<1 m s⁻¹ at 10 m), stable stratification (lapse rate 0.10 K m⁻¹ below 80 m; mean surface pressure 1022 hPa), 94.43% of cases with Ta > 0 °C, and fog indicators (visibility decreasing from 25 km to 6 km; dew point depression 2.5 °C).
- Feature selection retained seven features above the 25th percentile MI-score: σWdir and Wdir at 10 m, U at 10 and 20 m, FSI at 20 and 40 m, and Γθ at 10 m.
- KNN and RF performed best among supervised methods: with balanced data (n = 2), KNN identified stalling with accuracy 0.76 and RF with 0.82; F1-scores were 0.77 and 0.82, respectively. GNB was worst, with stalling accuracy dropping from approximately 80% to 50% under imbalance. OCSVM with ν < 0.2 and γ ≤ 1 reached up to 90% accuracy and was not affected by data imbalance.
Methods
- Used 10-min averaged meteorological data from Cabauw from 2001 to 2022 (22 years), focusing on wind direction at 10 m plus variables measured at 1.5–200 m; stalling was identified by zero wind-direction gradient for two or more consecutive points (≥20 min) when U ≤ 6 m s⁻¹.
- Compared five supervised binary classifiers (Logistic Regression, K-Nearest Neighbour, Random Forest, Gaussian Naive Bayes, Support Vector Machine) and one semi-supervised One-Class SVM; features were selected by Mutual Information score above the 25th percentile, parameters tuned by grid search, data balanced by retaining n non-stalling points up to one hour before and after each stalling event, and training/test splits varied between 10 and 90%.
- Supervised methods were evaluated with accuracy, precision, recall, and F1-score; OCSVM was evaluated by relative mean classification error for stalling data points.
Limitations
- Stated in the paper: outlier detection used to define stalling is not entirely definitive and the underlying cause remains uncertain; supervised methods are affected by data imbalance; practical real-time implementation requires further development; OCSVM is designed and validated only on the stalling class and needs further exploration for practical application.
- Evident from the text: the analysis excludes outlier years 2011, 2012, and 2022 from further analysis, and focuses on one site and primarily the 10 m level, which limits representativeness and generalization; the manual heuristic labels may also misclassify exceptionally steady winds as stalling.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
5 Valid · 2 Not valid · 6 Not checked · 59 Own results and interpretations
Every source the paper cites for this claim was checked and supports it.
In the paper
“The historic observations from Cabauw reveal that atmospheric stability and low wind speed at this site are often linked to the occurrence of fog, in particular, radiation fog [17].”
What the sources say
- Supportedcites [17] Izett, 2018, Understanding and Reducing False Alarms in Observational Fog Prediction, Bound.-Layer Meteorol. source
In “Observations and Methodology > Prediction of Fog from Observations,” the paper reports that almost all observed radiation-fog events occurred with low wind speed and a stable boundary layer. This supports the claim about the link between those conditions and radiation fog at Cabauw. (LLM, conf=0.94)
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The K-Nearest Neighbour and Random Forest methods performed the best, identifying stalling with approximately 75% accuracy, while others were more affected by data imbalance (more non-stalling than stalling data points).”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The K-Nearest Neighbour and Random Forest methods performed the best, identifying stalling with approximately 75% accuracy, while others were more affected by data imbalance (more non-stalling than stalling data points).”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The K-Nearest Neighbour and Random Forest methods performed the best, identifying stalling with approximately 75% accuracy, while others were more affected by data imbalance (more non-stalling than stalling data points).”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“The semi-supervised method, avoiding the effects of the inherent data imbalance, also yielded promising results towards advancing data quality assurance.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: unclear — It is unclear why the paper says it holds.
33 more claims
Create a free account to see all 72 claims and to verify the 3 references still open.
- Probably not a substantial claim. 15
- It may state two separate claims as one. 21
- The quote only partly states this claim. 1
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated major: it would touch a result — a value a result rests on, reported and recomputed p-values on opposite sides of the significance level, two passages that conflict with no sign which is right, or a cited work retracted before this paper was published.
In the paper
§ 5. Conclusions
“annual distributions of wind vane stalling frequency at 10 m revealed a consistent increase in stalling in the last decade compared to the first decade”
§ 4.1. Meteorological Conditions During Wind Vane Stalling
“Although stalling appears more frequent after 2014, no clear trend emerges from the yearly distribution.”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 4.4. Semi-Supervised One-Class Evaluation
“Due to the fact that this machine-learning method is exclusively designed and validated using the majority class (stalling)”
§ 3.3. Machine-Learning Methods
“Fitting data-driven methods for binary classification to an imbalanced dataset would yield tools that are biased towards the majority class (in our case, non-stalling events).”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 4.1. Meteorological Conditions During Wind Vane Stalling
“The percentage of time that the wind vane at each height stalled throughout the entire 22 years is shown in Figure 3a.”
§ 4.1. Meteorological Conditions During Wind Vane Stalling
“Figures ( a , c , d ) exclude the outlier years: 2011, 2012, and 2022.”
Citations
Cites
Breiman, L. (2001) · Machine Learning · doi:10.1023/a:1010933404324 · Cited by 133,538
Atmospheric Modeling, Data Assimilation and Predictability
Kalnay, E. (2002) · doi:10.1017/cbo9780511802270 · Cited by 3,391
The quiet revolution of numerical weather prediction
Bauer, P., Thorpe, A., & Brunet, G. (2015) · Nature · doi:10.1038/nature14956 · Cited by 2,729
Support Vector Method for Novelty Detection
Schölkopf, B., Williamson, R. C., Smola, A. J., Shawe-Taylor, J. S., & Platt, J. C. (1999) · Cited by 2,040
Learning skillful medium-range global weather forecasting
Lam, R., Sanchez-Gonzalez, A., Willson, M., Wirnsberger, P., Fortunato, M., Alet, F., Ravuri, S., Ewalds, T., Eaton-Rosen, Z., Hu, W., et al. (2023) · Science · doi:10.1126/science.adi2336 · Cited by 1,466
Evaluation metrics and statistical tests for machine learning
Rainio, O., Teuho, J., & Klén, R. (2024) · Scientific Reports · doi:10.1038/s41598-024-56706-x · Cited by 1,338
Alibrahim, H., & Ludwig, S. A. (2021) · 2021 IEEE Congress on Evolutionary Computation (CEC) · doi:10.1109/cec45853.2021.9504761 · Cited by 489
MIFS-ND: A mutual information-based feature selection method
Hoque, N., Bhattacharyya, D. K., & Kalita, J. K. (2014) · Expert Systems with Applications · doi:10.1016/j.eswa.2014.04.019 · Cited by 442
Atmospheric blocking and weather extremes over the Euro-Atlantic sector – a review
Kautz, L.-A., Martius, O., Pfahl, S., Pinto, J. G., Ramos, A. M., Sousa, P. M., & Woollings, T. (2022) · Weather and Climate Dynamics · doi:10.5194/wcd-3-305-2022 · Cited by 337
Machine Learning Methods in Weather and Climate Applications: A Survey
Chen, L., Han, B., Wang, X., Zhao, J., Yang, W., & Yang, Z. (2023) · doi:10.20944/preprints202309.1764.v2 · Cited by 246
Research works printed but not linked: 1
- Holtslag, M., Steeneveld, G., and Holtslag, A. (2010, January 25–30). Fog forecasting: “Old fashioned” semi-empirical methods from radio sounding observations versus “modern” numerical models. Proceedings of the 5th International Conference on Fog, Fog Collection and Dew, Münster, Germany.
Cited by
Data Quality-Aware Federated Learning for Fake Review Detection
Xu, J., Zhang, C., Liu, J., et al. (2025) · International Conference on Software Engineering and Computer Science (CSECS) · doi:10.1109/csecs64665.2025.11009627 · Cited by 1
“Intraditional centralized approaches, data quality is generally ensured by manual filtering, statistical feature-based cleaning, or anomaly detection [13], [14].”
Jia, H., Pan, H., Qian, J., et al. (2026) · Atmosphere · doi:10.3390/atmos17040372
Revisiting Extreme Precipitation of European Summer 2012: The Role of Atlantic Tropical Storm Debby
Krichak, S. O. (2026) · Meteorology · doi:10.3390/meteorology5040035
Metadata
- Authors
- de Feiter, V. S., Strickland, J. M. I., & Garcia-Marti, I.
- Affiliations
- Meteorology & Air Quality Group, Wageningen University & Research, 6708 PB Wageningen, The Netherlands
- Royal Netherlands Meteorological Institute (NL)
- Royal Netherlands Meteorological Institute (KNMI), Utrechtseweg 297, 3731 GA De Bilt, The Netherlands
- Venue
- Atmosphere
- Publisher
- MDPI AG
- Published
- 2025-01-25
- Volume
- 16
- Issue
- 2
- Pages
- 129
- ISSN
- 2073-4433 (ISSN-L)
- OpenAlex
- W4406846458
- Cited by
- 3
- References
- 26
- Citations by year
- 2026: 2 · 2025: 1
- FWCI
- 2.72
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- Atmosphere (publishedVersion, cc-by)
- Socio-Environmental Systems Modeling (submittedVersion, cc-by)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Funders
- Wageningen University and Research
- Topics
- Physical Sciences › Environmental Science › Management, Monitoring, Policy and Law › Environmental Monitoring and Data Management
- Physical Sciences › Engineering › Control and Systems Engineering › Fault Detection and Control Systems
- Physical Sciences › Earth and Planetary Sciences › Atmospheric Science › Meteorological Phenomena and Simulations
- Keywords
- data quality assurance, meteorological observations, binary classification, semi-supervised learning, one-class support vector machine, class imbalance, in situ wind measurements, random forest, k-nearest neighbors
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Advancing Data Quality Assurance with Machine Learning: A Case Study on Wind Vane Stalling Detection. https://app.cochat.ai/papers/10.3390/atmos16020129
- The paper
- de Feiter, V. S., Strickland, J. M. I., & Garcia-Marti, I. (2025). Advancing Data Quality Assurance with Machine Learning: A Case Study on Wind Vane Stalling Detection. Atmosphere. https://doi.org/10.3390/atmos16020129
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.