Monerie et al. (2026) Global wind stilling and the role of sub-monthly variability in explaining deficiencies in atmospheric reanalyses

Identification

• Journal: Climate Dynamics
• Year: 2026
• Date: 2026-04-01
• Authors: Paul‐Arthur Monerie, R. Schiemann, David Brayshaw, Jon Robson
• DOI: 10.1007/s00382-026-08163-7

Research Groups

• National Centre for Atmospheric Science, Reading, UK
• University of Reading, Reading, UK

Short Summary

This study confirms the robustness of global wind stilling (1980-2010) in Northern Hemisphere land observations and reveals that most atmospheric reanalyses fail to reproduce this trend, primarily due to their inability to capture changes in sub-monthly wind speed variability. It also highlights the significant implications for wind power density assessments and reconciles contradictory findings in previous literature based on data processing methods.

Objective

• To assess the robustness of global near-surface (10-meter) wind stilling in station-based observations over the Northern Hemisphere (1980-2010).
• To evaluate the ability of multiple atmospheric reanalysis products to reproduce observed past changes in 10-meter wind speed.
• To explain the differences in wind speed trends between station-based datasets and reanalyses, specifically by decomposing wind speed variability into sub-monthly and monthly components.
• To reconcile contradictory findings from previous studies regarding the representation of global stilling in reanalyses.

Study Configuration

• Spatial Scale: Terrestrial Northern Hemisphere (0-70°N), with specific focus on Europe (35-71.5°N; 25°W-45.5°E) and China. Also mentions Eastern US, Central Asia, India, East Asia, South America, Africa, and Australia for broader trend patterns. Near-surface (10-meter) and 100-meter wind speeds are analyzed.
• Temporal Scale: Primary analysis period is 1980–2010. Some data used from 1979–2022 for specific calculations (e.g., wind shear exponent). Wind speed variability is decomposed into sub-monthly and monthly timescales.

Methodology and Data

• Models used:
  • ERA5 (ECMWF 5th-generation reanalysis)
  • JRA55 and JRA55C (Japanese 55-year Reanalysis, Japan Meteorological Agency)
  • MERRA2 (NASA’s Modern-Era Retrospective Analysis for Research and Applications-2)
  • NCEP-R1 and NCEP-R2 (National Centres for Environmental Prediction/National Centre for Atmospheric Research reanalysis)
  • 20CR (Twentieth Century Reanalysis, version 2, ensemble mean)
• Data sources:
  • Station-based observations:
    • HadISD3.4.1 (Met Office, global, sub-daily, ~8500 stations)
    • CN05.1 (China, gridded, based on 2416 meteorological stations)
    • E-OBS v28 (Europe, gridded, daily mean)
  • Ancillary data: European Space Agency (Climate Change Initiative) land cover (ESA, 2017) for urban/rural station classification.
  • Methods:
    • Wind speed calculation: $wspd = \sqrt{u^2 + v^2}$ using highest available temporal resolution.
    • Wind speed decomposition: Proxy $u^2 + v^2$ decomposed into monthly mean components ($\overline{u}^2 + \overline{v}^2$) and sub-monthly variability ($u'^2 + v'^2$).
    • Wind power density (WPD) calculation: $WPD = \frac{1}{2}\rho u{100}^{3}$, where air density $\rho = 1.213 \text{ kg m}^{-3}$ and $u{100}$ is estimated from $u_{10}$ using a power law with a spatially varying exponent $\alpha$ derived from ERA5.
    • Trend analysis: Linear trends computed for 1980–2010 after applying a 3-year running mean. Statistical significance assessed using a Student’s t-test at the 95% confidence level.
    • Wind event frequency: Changes in frequency of events exceeding 3 m s$^{-1}$, 10 m s$^{-1}$, and 15 m s$^{-1}$.

Main Results

• Robust Global Stilling: Terrestrial near-surface wind speed in HadISD3 decreased by approximately 5% over the Northern Hemisphere between 1980 and 2010. This 'global stilling' is robust and not an artifact of analysis methods, observational network characteristics, or urbanisation effects.
• Reanalysis Deficiencies: Most atmospheric reanalyses (ERA5, MERRA2, NCEP-R1, NCEP-R2, 20CR) fail to reproduce the observed global stilling trend over the Northern Hemisphere, Europe, and China, showing weaker or inconsistent trends.
• Role of Sub-monthly Variability: The observed global stilling in HadISD3 is primarily associated with a decrease in sub-monthly wind speed variability. The deficiencies in reanalyses are also primarily attributed to their misrepresentation of this sub-monthly component.
• Reconciling Contradictory Findings: The apparent alignment of reanalyses with observed stilling in some previous studies (e.g., Deng et al. 2021) is an artifact of their data-processing methodology, specifically using monthly mean zonal and meridional wind components to calculate wind speed, which ignores non-linear effects of sub-monthly variability.
• JRA55/JRA55C Performance: JRA55 and JRA55C are the only reanalyses that reproduce the magnitude and spatial pattern of observed global stilling with high skill. However, this skill is likely due to the assimilation of observed 10-meter wind speed and post-processing corrections, rather than an accurate representation of the underlying forecast system dynamics. These products also exhibit an anomalous high wind speed peak in the mid-1990s due to a unit conversion error.
• Wind Power Density Implications: Global stilling leads to a significant decrease in potential wind power density (WPD) in HadISD3, indicating reduced wind energy potential. Most reanalyses fail to accurately reproduce this WPD trend, with some even showing an increase, making them unreliable for assessing past WPD changes. The reduction in WPD is more pronounced than the wind speed decline because WPD is disproportionately influenced by high-wind events.
• Storm Track Activity: No significant trend in storm track activity (based on sea level pressure variance) was found over land in the Northern Hemisphere, suggesting it does not explain the discrepancies between reanalyses and observations regarding stilling.

Contributions

• Provides a robust confirmation of global near-surface wind stilling (1980-2010) in observational datasets, demonstrating its independence from methodological choices or network biases.
• Offers a comprehensive intercomparison of multiple atmospheric reanalyses, systematically evaluating their ability to capture observed wind stilling trends across key regions.
• Identifies and quantifies the critical role of sub-monthly wind speed variability as the primary driver of observed global stilling and the main source of error in reanalysis products.
• Resolves conflicting findings in existing literature by demonstrating how different wind speed calculation methodologies (using highest temporal resolution vs. monthly means) lead to divergent conclusions about reanalysis performance.
• Offers practical recommendations for the selection and cautious use of reanalysis products for historical wind trend analysis and wind energy assessments, highlighting the limitations of even seemingly accurate reanalyses like JRA55/JRA55C.
• Quantifies the significant implications of global stilling for the wind energy sector by showing a more pronounced reduction in potential wind power density than in wind speed itself.

Funding

• WISTERA (Wind Stilling and Energy in Europe and Asia) project under the Climate Science for Service Partnership China (CSSP China), as part of the International Science Partnerships Fund (ISPF), delivered by the Met Office.
• Natural Environment Research Council (NERC) CANARI (NE/W004984/1).
• Natural Environment Research Council (NERC) ALPACA project (NE/Y005279/1).

Citation

@article{Monerie2026Global,
  author = {Monerie, Paul‐Arthur and Schiemann, R. and Brayshaw, David and Robson, Jon},
  title = {Global wind stilling and the role of sub-monthly variability in explaining deficiencies in atmospheric reanalyses},
  journal = {Climate Dynamics},
  year = {2026},
  doi = {10.1007/s00382-026-08163-7},
  url = {https://doi.org/10.1007/s00382-026-08163-7}
}