pith. machine review for the scientific record. sign in

arxiv: 2604.26776 · v1 · submitted 2026-04-29 · ⚛️ physics.flu-dyn

Recognition: unknown

Conditional diffusion denoising probabilistic model for super-resolution of atmospheric boundary layer large eddy simulation

Omar Sallam , Mirjam F\"urth
Authors on Pith no claims yet

Pith reviewed 2026-05-07 11:37 UTC · model grok-4.3

classification ⚛️ physics.flu-dyn
keywords atmospheric boundary layerlarge eddy simulationdiffusion modelsuper-resolutionturbulent inflowwind energyReynolds stressesgenerative model
0
0 comments X

The pith

A conditional diffusion model reconstructs fine-scale turbulence and stresses from coarse atmospheric boundary layer simulations inside its training range.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper trains a conditional denoising diffusion probabilistic model on large-eddy simulation data of the atmospheric boundary layer to reconstruct high-resolution turbulent flow fields from coarser versions. This matters for wind-energy applications because full high-fidelity simulations remain too slow for large-scale or real-time use while still needing accurate shear and turbulence stresses. The model recovers small-scale structures, Reynolds stresses, and statistical properties when the coarse inputs come from wind speeds and roughness values inside the training set. Performance drops outside that set, producing more noise and overpredicted stresses at higher wind speeds. The work therefore shows that physics-informed generative models can lower the cost of inflow generation while remaining reliable only within a bounded range of atmospheric conditions.

Core claim

A conditional denoising diffusion probabilistic model is trained on high-fidelity LES datasets generated across varying geostrophic wind speeds and surface roughnesses aligned with IEC wind classes. In interpolation cases the model reconstructs fine-scale turbulent structures and maintains statistical consistency including Reynolds stresses. Extrapolation to higher wind speeds produces increased noise and overprediction of turbulent stresses.

What carries the argument

Conditional denoising diffusion probabilistic model conditioned on coarse-resolution LES fields to generate high-resolution atmospheric boundary layer turbulence.

If this is right

  • Reduces the computational cost of generating turbulent inflow conditions for wind-energy load and power predictions.
  • Preserves physical consistency in fine-scale structures and Reynolds stresses within the training domain.
  • Supports super-resolution at multiple scale factors while staying inside the trained atmospheric conditions.
  • Highlights the need for training data that covers the full range of expected wind speeds and roughnesses to avoid noise and stress overprediction.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Retraining on a wider set of wind speeds and roughnesses could extend reliable use to more variable atmospheric regimes.
  • The same conditioning approach could be tested on super-resolution tasks in other turbulent flows such as channel or ocean simulations.
  • Coupling the model with existing LES solvers might enable faster ensemble runs for uncertainty estimates in wind-farm planning.

Load-bearing premise

The distribution of coarse inputs created from the chosen geostrophic wind speeds and surface roughnesses is representative enough for the model to generalize.

What would settle it

Generate an independent high-resolution LES at a geostrophic wind speed outside the training range, downsample it to coarse resolution, pass it through the trained diffusion model, and check whether the output turbulence statistics and structures match the original high-resolution field.

Figures

Figures reproduced from arXiv: 2604.26776 by Mirjam F\"urth, Omar Sallam.

Figure 1
Figure 1. Figure 1: Conditional Convolutional UNet architecture for the tropical cyclone mesoscale to LES DDPM super-resolution problem. 1.4. Paper Organization The remainder of this paper is organized as follows. Section 2 describes the numerical configuration for the neutral atmospheric boundary layer, including the generation of LES datasets at multiple resolutions, validation of the numerical setup, and the train–test dat… view at source ↗
Figure 2
Figure 2. Figure 2: Validation and verification of the numerical simulation with previously reported numerical and field experiments. Figure 3a–3c compare the vertical profiles of the mean streamwise velocity and turbulent stresses obtained using the four grid resolutions for geostrophic wind speeds of 𝑈𝑔 = 5.85, 7.15, and 10.0 m s−1. The results demonstrate sensitivity of the mean velocity profiles to grid resolution, partic… view at source ↗
Figure 3
Figure 3. Figure 3: Parameters variation with grid resolutions at different geostrophic wind speeds. 2.3.2. Geostrophic wind speed The influence of geostrophic wind speed on the mean velocity profiles and turbulent stresses is examined across different surface roughness length scales view at source ↗
Figure 4
Figure 4. Figure 4: Parameters variation with geostrophic wind speeds at different surface roughness values. Omar Sallam et al.: Preprint submitted to Elsevier Page 11 of 38 view at source ↗
Figure 5
Figure 5. Figure 5: Parameters variation with surface roughness speeds at different geostrophic wind speeds 3. Diffusion-based super-resolution framework for ABL This section presents a diffusion-based generative modeling framework for super-resolving mesoscale wind fields to LES-resolution using a conditional denoising diffusion probabilistic model (DDPM). We begin with a general overview of the diffusion model fundamentals,… view at source ↗
Figure 6
Figure 6. Figure 6: Schematic overview of the proposed diffusion-based super-resolution framework for tropical cyclone wind field reconstruction. 4. Results This section presents the super-resolution performance of the proposed framework under both interpolation and extrapolation scenarios. The evaluation is conducted across multiple spatial scale factors (×2, ×4, and ×8), varying geostrophic wind speeds (𝑈𝑔 ), and different … view at source ↗
Figure 7
Figure 7. Figure 7: Results for the interpolation task, scale factor 2 with input grid 2, geostrophic wind speed 7.15 [m/sec], surface roughness coefficient 0.05 [m] Omar Sallam et al.: Preprint submitted to Elsevier Page 17 of 38 view at source ↗
Figure 8
Figure 8. Figure 8: Results for the interpolation task, scale factor 2 with input grid 2, geostrophic wind speed 10 [m/sec], surface roughness coefficient 0.05 [m] Omar Sallam et al.: Preprint submitted to Elsevier Page 18 of 38 view at source ↗
Figure 9
Figure 9. Figure 9: Results for the interpolation task, scale factor 4 with input grid 1, geostrophic wind speed 7.15 [m/sec], surface roughness coefficient 0.05 [m] Omar Sallam et al.: Preprint submitted to Elsevier Page 19 of 38 view at source ↗
Figure 10
Figure 10. Figure 10: Results for the interpolation task, scale factor 4 with input grid 1, geostrophic wind speed 10 [m/sec], surface roughness coefficient 0.05 [m] Omar Sallam et al.: Preprint submitted to Elsevier Page 20 of 38 view at source ↗
Figure 11
Figure 11. Figure 11: Results for the interpolation task, scale factor 8 with input grid 0, geostrophic wind speed 7.15 [m/sec], surface roughness coefficient 0.05 [m] Omar Sallam et al.: Preprint submitted to Elsevier Page 21 of 38 view at source ↗
Figure 12
Figure 12. Figure 12: Results for the interpolation task, scale factor 8 with input grid 0, geostrophic wind speed 10 [m/sec], surface roughness coefficient 0.05 [m] Omar Sallam et al.: Preprint submitted to Elsevier Page 22 of 38 view at source ↗
Figure 13
Figure 13. Figure 13: Isosurfaces of the 𝑄-criterion at different input grid resolution at geostrophic wind speed 7.15 [m/sec] [Interpolation task]. Omar Sallam et al.: Preprint submitted to Elsevier Page 23 of 38 view at source ↗
Figure 14
Figure 14. Figure 14: Isosurfaces of the 𝑄-criterion at different input grid resolution at geostrophic wind speed 10.0 [m/sec] [Interpolation task]. Omar Sallam et al.: Preprint submitted to Elsevier Page 24 of 38 view at source ↗
Figure 15
Figure 15. Figure 15: Results for the extrapolation task, scale factor 2 with input grid 2, geostrophic wind speed 7.15 [m/sec], surface roughness coefficient 0.1 [m] Omar Sallam et al.: Preprint submitted to Elsevier Page 27 of 38 view at source ↗
Figure 16
Figure 16. Figure 16: Results for the extrapolation task, scale factor 2 with input grid 2, geostrophic wind speed 10 [m/sec], surface roughness coefficient 0.1 [m] Omar Sallam et al.: Preprint submitted to Elsevier Page 28 of 38 view at source ↗
Figure 17
Figure 17. Figure 17: Results for the extrapolation task, scale factor 4 with input grid 1, geostrophic wind speed 7.15 [m/sec], surface roughness coefficient 0.1 [m] Omar Sallam et al.: Preprint submitted to Elsevier Page 29 of 38 view at source ↗
Figure 18
Figure 18. Figure 18: Results for the extrapolation task, scale factor 4 with input grid 1, geostrophic wind speed 10 [m/sec], surface roughness coefficient 0.1 [m] Omar Sallam et al.: Preprint submitted to Elsevier Page 30 of 38 view at source ↗
Figure 19
Figure 19. Figure 19: Results for the extrapolation task, scale factor 8 with input grid 0, geostrophic wind speed 7.15 [m/sec], surface roughness coefficient 0.1 [m] Omar Sallam et al.: Preprint submitted to Elsevier Page 31 of 38 view at source ↗
Figure 20
Figure 20. Figure 20: Results for the extrapolation task, scale factor 8 with input grid 0, geostrophic wind speed 10 [m/sec], surface roughness coefficient 0.1 [m] Omar Sallam et al.: Preprint submitted to Elsevier Page 32 of 38 view at source ↗
Figure 21
Figure 21. Figure 21: Isosurfaces of the 𝑄-criterion at different input grid resolution at geostrophic wind speed 7.15 [m/sec] [Extrapolation task]. Omar Sallam et al.: Preprint submitted to Elsevier Page 33 of 38 view at source ↗
Figure 22
Figure 22. Figure 22: Isosurfaces of the 𝑄-criterion at different input grid resolution at geostrophic wind speed 10.0 [m/sec] [Extrapolation task]. Omar Sallam et al.: Preprint submitted to Elsevier Page 34 of 38 view at source ↗
read the original abstract

Climate change necessitates rapid expansion of renewable energy, with wind energy offering a scalable and low-impact solution. However, accurate prediction of wind loads and power generation remains challenging due to uncertainties in wind shear and turbulence stresses under atmospheric boundary layer (ABL) conditions. High-fidelity Large Eddy Simulations (LES) are typically used to reduce these uncertainties but are computationally expensive and impractical for large-scale or real-time applications. This work addresses this limitation using generative AI, specifically Conditional Denoising Diffusion Probabilistic Models, to reconstruct high-resolution turbulent flow fields from coarse inputs. A high-fidelity dataset is generated using a parallel high-order finite-difference solver across varying geostrophic wind speeds, surface roughness conditions aligned with IEC wind classes, and multiple grid resolutions. The diffusion model is trained for super-resolution across different scale factors and evaluated under interpolation and extrapolation scenarios. Results show accurate recovery of fine-scale turbulent structures, Reynolds stresses, and statistical properties in interpolation cases, indicating strong physical consistency within the training domain. However, extrapolation to higher wind speeds leads to increased noise and overprediction of turbulent stresses, highlighting limitations in generalization. Overall, the study demonstrates that physics-informed generative models can significantly reduce computational cost while maintaining acceptable accuracy, enabling faster and more reliable turbulent inflow characterization for wind energy applications.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript proposes using conditional denoising diffusion probabilistic models (DDPMs) for super-resolving coarse-resolution large-eddy simulations (LES) of atmospheric boundary layer (ABL) turbulence to high-fidelity fields. It generates training data via a parallel high-order finite-difference solver across multiple geostrophic wind speeds, IEC-aligned surface roughnesses, and grid resolutions. The model is trained for various super-resolution scale factors and assessed in interpolation (within training parameter space) and extrapolation (to higher wind speeds) settings. Key results include accurate reconstruction of fine-scale turbulent structures, Reynolds stresses, and flow statistics for interpolation cases, demonstrating physical consistency, contrasted with degraded performance involving noise and stress overprediction in extrapolation. The work concludes that such AI models can substantially lower computational costs for turbulent inflow characterization in wind energy applications.

Significance. Should the reported physical consistency in interpolation scenarios be confirmed with rigorous quantitative validation, this approach could meaningfully advance renewable energy modeling by providing a computationally efficient alternative to full high-resolution LES for generating realistic turbulent inflows. The application of conditional diffusion models to ABL flows represents a timely integration of generative AI with fluid dynamics, potentially scalable to real-time predictions. The explicit acknowledgment of extrapolation limitations is a strength, though it underscores the importance of expanding the training distribution for broader utility.

major comments (2)
  1. [Abstract] Abstract (final paragraph): The claim of 'accurate recovery of fine-scale turbulent structures, Reynolds stresses, and statistical properties in interpolation cases' indicating 'strong physical consistency' lacks any reported quantitative metrics such as error norms on velocity fields or stresses, spectral comparisons, or baseline method results; without these, and given the documented degradation under extrapolation to higher wind speeds, it is unclear if the interpolation performance reflects robust recovery or simply close overlap with the limited training parameter combinations.
  2. [Abstract] Dataset generation (described in abstract): The high-fidelity dataset is generated 'across varying geostrophic wind speeds, surface roughness conditions aligned with IEC wind classes, and multiple grid resolutions' but no explicit ranges, discrete values, number of cases, or sampling strategy are provided; this detail is load-bearing for the interpolation claim because the abstract already shows sensitivity to shifts in wind speed, raising the possibility that reported success depends on dense coverage of the chosen conditions rather than generalization within the domain.
minor comments (1)
  1. [Abstract] The abstract would be clearer if it included at least one concrete quantitative indicator (e.g., a typical relative error on Reynolds stresses) alongside the qualitative statements of accuracy.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments on our manuscript. We address each major comment point by point below and have revised the abstract to improve clarity and support for the claims, as detailed in the responses.

read point-by-point responses

  1. Referee: [Abstract] Abstract (final paragraph): The claim of 'accurate recovery of fine-scale turbulent structures, Reynolds stresses, and statistical properties in interpolation cases' indicating 'strong physical consistency' lacks any reported quantitative metrics such as error norms on velocity fields or stresses, spectral comparisons, or baseline method results; without these, and given the documented degradation under extrapolation to higher wind speeds, it is unclear if the interpolation performance reflects robust recovery or simply close overlap with the limited training parameter combinations.

    Authors: We agree that the abstract, as a concise summary, would benefit from explicit reference to quantitative support for the interpolation results. The full manuscript reports these metrics (L2 norms on velocity and stress fields, spectral comparisons, and baseline comparisons) in the results section. We have revised the abstract to include a brief statement referencing the key quantitative findings for interpolation cases, such as low error norms and spectral agreement, to better substantiate the physical consistency claim while retaining the contrast with extrapolation performance. revision: yes

  2. Referee: [Abstract] Dataset generation (described in abstract): The high-fidelity dataset is generated 'across varying geostrophic wind speeds, surface roughness conditions aligned with IEC wind classes, and multiple grid resolutions' but no explicit ranges, discrete values, number of cases, or sampling strategy are provided; this detail is load-bearing for the interpolation claim because the abstract already shows sensitivity to shifts in wind speed, raising the possibility that reported success depends on dense coverage of the chosen conditions rather than generalization within the domain.

    Authors: We acknowledge that the abstract omits specific parameter details, which are provided in the methods section of the full manuscript. To address the concern about transparency for the interpolation claim, we have revised the abstract to include a concise summary of the ranges, discrete values, and number of cases used in dataset generation. This clarifies the sampling within the training domain without altering the manuscript's core findings. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation chain is self-contained

full rationale

The paper generates LES datasets from chosen geostrophic winds and IEC-aligned roughnesses, trains a conditional diffusion model for super-resolution, and reports empirical metrics on held-out interpolation/extrapolation cases. No equations reduce performance metrics (turbulent structures, Reynolds stresses) to quantities defined by the same fitted parameters; training and test splits are described as separate. No self-citations serve as load-bearing uniqueness theorems, no ansatzes are smuggled, and no predictions are statistically forced by construction. The central claims rest on independent evaluation rather than definitional equivalence.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on the assumption that a generative model trained on a finite set of LES runs can learn a general mapping from coarse to fine fields; the paper supplies no independent verification that this mapping is unique or physics-preserving outside the training distribution.

free parameters (1)
  • diffusion model architecture and training hyperparameters
    All weights and schedule parameters are fitted to the generated LES dataset; their values are not reported.
axioms (1)
  • domain assumption Coarse-grid fields contain sufficient statistical information to reconstruct fine-scale turbulence statistics
    Invoked when claiming physical consistency from super-resolved Reynolds stresses.

pith-pipeline@v0.9.0 · 5526 in / 1304 out tokens · 42898 ms · 2026-05-07T11:37:43.566724+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

57 extracted references · 2 canonical work pages · 1 internal anchor

  1. [1]

    Energy Information Administration, Where greenhouse gases come from, accessed: 2025-02-21 (n.d.)

    U.S. Energy Information Administration, Where greenhouse gases come from, accessed: 2025-02-21 (n.d.). URL https://www.eia.gov/energyexplained/energy-and-the-environment/ where-greenhouse-gases-come-from.php

    2025

  2. [2]

    URL https://www.nrel.gov/wind/offshore-supply-chain-road-map.html

    National Renewable Energy Laboratory, A supply chain road map for offshore wind energy in the united states, accessed: 2025-02-21 (January 2023). URL https://www.nrel.gov/wind/offshore-supply-chain-road-map.html

    2025

  3. [3]

    URL https://cleanpower.org/facts/wind-power/

    American Clean Power Association, Wind power facts and statistics, accessed: 2025-02-21 (n.d.). URL https://cleanpower.org/facts/wind-power/

    2025

  4. [4]

    A. N. Robertson, K. Shaler, L. Sethuraman, J. Jonkman, Sensitivity analysis of the effect of wind characteristics and turbine properties on wind turbine loads, Wind Energy Science 4 (3) (2019) 479–513

    2019

  5. [5]

    Jonkman, Definition of a 5-mw reference wind turbine for offshore system development, National Renewable Energy Laboratory (2009)

    J. Jonkman, Definition of a 5-mw reference wind turbine for offshore system development, National Renewable Energy Laboratory (2009)

    2009

  6. [6]

    B. J. Stanislawski, R. Thedin, A. Sharma, E. Branlard, G. Vijayakumar, M. A. Sprague, Effect of the integral length scales of turbulent inflows on wind turbine loads, Renewable Energy 217 (2023) 119218

    2023

  7. [7]

    35, 2021, pp

    T.Yu,Y.Yang,D.Li,T.Hospedales,T.Xiang,Simpleandeffectivestochasticneuralnetworks,in:Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 35, 2021, pp. 3252–3260

    2021

  8. [8]

    G. E. Karniadakis, I. G. Kevrekidis, L. Lu, P. Perdikaris, S. Wang, L. Yang, Physics-informed machine learning, Nature Reviews Physics 3 (6) (2021) 422–440

    2021

  9. [9]

    Cuomo, V

    S. Cuomo, V. S. Di Cola, F. Giampaolo, G. Rozza, M. Raissi, F. Piccialli, Scientific machine learning through physics–informedneuralnetworks:Whereweareandwhat’snext,JournalofScientificComputing92(3)(2022) 88

    2022

  10. [10]

    J. Ho, A. Jain, P. Abbeel, Denoising diffusion probabilistic models, Advances in neural information processing systems 33 (2020) 6840–6851

    2020

  11. [11]

    L. Yang, Z. Zhang, Y. Song, S. Hong, R. Xu, Y. Zhao, W. Zhang, B. Cui, M.-H. Yang, Diffusion models: A comprehensive survey of methods and applications, ACM Computing Surveys 56 (4) (2023) 1–39

    2023

  12. [12]

    M.Chen,S.Mei,J.Fan,M.Wang,Anoverviewofdiffusionmodels:Applications,guidedgeneration,statistical rates and optimization, arXiv preprint arXiv:2404.07771 (2024)

    work page arXiv 2024

  13. [13]

    Stoll, J

    R. Stoll, J. A. Gibbs, S. T. Salesky, W. Anderson, M. Calaf, Large-eddy simulation of the atmospheric boundary layer, Boundary-Layer Meteorology 177 (2020) 541–581

    2020

  14. [14]

    Smagorinsky, General circulation experiments with the primitive equations: I

    J. Smagorinsky, General circulation experiments with the primitive equations: I. the basic experiment, Monthly weather review 91 (3) (1963) 99–164

    1963

  15. [15]

    D.K.Lilly,Therepresentationofsmall-scaleturbulenceinnumericalsimulationexperiments,in:Proc.IBMsci. comput. symp. on environmental science, 1967, pp. 195–210

    1967

  16. [16]

    Kosović, Subgrid-scale modelling for the large-eddy simulation of high-reynolds-number boundary layers, Journal of Fluid Mechanics 336 (1997) 151–182

    B. Kosović, Subgrid-scale modelling for the large-eddy simulation of high-reynolds-number boundary layers, Journal of Fluid Mechanics 336 (1997) 151–182

    1997

  17. [17]

    F. K. Chow, R. L. Street, M. Xue, J. H. Ferziger, Explicit filtering and reconstruction turbulence modeling for large-eddy simulation of neutral boundary layer flow, Journal of the atmospheric sciences 62 (7) (2005) 2058– 2077

    2005

  18. [18]

    Wurps, G

    H. Wurps, G. Steinfeld, S. Heinz, Grid-resolution requirements for large-eddy simulations of the atmospheric boundary layer, Boundary-Layer Meteorology 175 (2) (2020) 179–201. Omar Sallam et al.:Preprint submitted to Elsevier Page 36 of 38

    2020

  19. [19]

    Ennis, C

    J.D.Mirocha,M.J.Churchfield,D.Muñoz-Esparza,R.K.Rai,Y.Feng,B.Kosović,S.E.Haupt,B.Brown,B.L. Ennis, C. Draxl, et al., Large-eddy simulation sensitivities to variations of configuration and forcing parameters in canonical boundary-layer flows for wind energy applications, Wind Energy Science 3 (2) (2018) 589–613

    2018

  20. [20]

    Watanabe, Large-eddy simulation of coherent turbulence structures associated with scalar ramps over plant canopies, Boundary-Layer Meteorology 112 (2004) 307–341

    T. Watanabe, Large-eddy simulation of coherent turbulence structures associated with scalar ramps over plant canopies, Boundary-Layer Meteorology 112 (2004) 307–341

    2004

  21. [21]

    Bohrer, G

    G. Bohrer, G. G. Katul, R. L. Walko, R. Avissar, Exploring the effects of microscale structural heterogeneity of forest canopies using large-eddy simulations, Boundary-layer meteorology 132 (2009) 351–382

    2009

  22. [22]

    W. C. Skamarock, J. B. Klemp, J. Dudhia, D. O. Gill, D. M. Barker, M. G. Duda, X.-Y. Huang, W. Wang, J. G. Powers, et al., A description of the advanced research wrf version 3, NCAR technical note 475 (125) (2008) 10–5065

    2008

  23. [23]

    Churchfield, S

    M. Churchfield, S. Lee, P. Moriarty, Overview of the simulator for wind farm application (sowfa), National Renewable Energy Laboratory 375 (2012)

    2012

  24. [24]

    J. A. Sauer, D. Muñoz-Esparza, J. M. Canfield, K. R. Costigan, R. R. Linn, Y.-J. Kim, A large-eddy simulation study of atmospheric boundary layer influence on stratified flows over terrain, Journal of the Atmospheric Sciences 73 (7) (2016) 2615–2632

    2016

  25. [25]

    P. P. Sullivan, E. G. Patton, The effect of mesh resolution on convective boundary layer statistics and structures generated by large-eddy simulation, Journal of the Atmospheric Sciences 68 (10) (2011) 2395–2415

    2011

  26. [26]

    Sallam, M

    O. Sallam, M. Fürth, Dual scale residual-network for turbulent flow sub grid scale resolving: A prior analysis, Computers & Fluids (2025) 106592

    2025

  27. [27]

    Fukami, K

    K. Fukami, K. Fukagata, K. Taira, Super-resolution reconstruction of turbulent flows with machine learning, Journal of Fluid Mechanics 870 (2019) 106–120

    2019

  28. [28]

    Fukami, K

    K. Fukami, K. Fukagata, K. Taira, Machine-learning-based spatio-temporal super resolution reconstruction of turbulent flows, Journal of Fluid Mechanics 909 (2021) A9

    2021

  29. [29]

    Fukami, K

    K. Fukami, K. Taira, Single-snapshot machine learning for super-resolution of turbulence, Journal of Fluid Mechanics 1001 (2024) A32

    2024

  30. [30]

    H. Kim, J. Kim, S. Won, C. Lee, Unsupervised deep learning for super-resolution reconstruction of turbulence, Journal of Fluid Mechanics 910 (2021) A29

    2021

  31. [31]

    Sanderson, W

    G.A.Meehl,W.M.Washington,J.M.Arblaster,A.Hu,H.Teng,J.E.Kay,A.Gettelman,D.M.Lawrence,B.M. Sanderson, W. G. Strand, Climate change projections in cesm1 (cam5) compared to ccsm4, Journal of Climate 26 (17) (2013) 6287–6308

    2013

  32. [32]

    Hersbach, B

    H. Hersbach, B. Bell, P. Berrisford, S. Hirahara, A. Horányi, J. Muñoz-Sabater, J. Nicolas, C. Peubey, R. Radu, D. Schepers, et al., The era5 global reanalysis, Quarterly journal of the royal meteorological society 146 (730) (2020) 1999–2049

    2020

  33. [33]

    Draxl, A

    C. Draxl, A. Clifton, B.-M. Hodge, J. McCaa, The wind integration national dataset (wind) toolkit, Applied Energy 151 (2015) 355–366

    2015

  34. [34]

    Baldauf, A

    M. Baldauf, A. Seifert, J. Förstner, D. Majewski, M. Raschendorfer, T. Reinhardt, Operational convective-scale numerical weather prediction with the cosmo model: Description and sensitivities, Monthly Weather Review 139 (12) (2011) 3887–3905

    2011

  35. [35]

    Stengel, A

    K. Stengel, A. Glaws, D. Hettinger, R. N. King, Adversarial super-resolution of climatological wind and solar data, Proceedings of the National Academy of Sciences 117 (29) (2020) 16805–16815

    2020

  36. [36]

    Buster, B

    G. Buster, B. N. Benton, A. Glaws, R. N. King, High-resolution meteorology with climate change impacts from global climate model data using generative machine learning, Nature Energy 9 (7) (2024) 894–906. Omar Sallam et al.:Preprint submitted to Elsevier Page 37 of 38

    2024

  37. [37]

    O.Miralles,D.Steinfeld,O.Martius,A.C.Davison,Downscalingofhistoricalwindfieldsoverswitzerlandusing generative adversarial networks, Artificial Intelligence for the Earth Systems 1 (4) (2022) e220018

    2022

  38. [38]

    H. Lin, J. Tang, S. Wang, S. Wang, G. Dong, Deep learning downscaled high-resolution daily near surface meteorological datasets over east asia, Scientific Data 10 (1) (2023) 890

    2023

  39. [39]

    Ronneberger, P

    O. Ronneberger, P. Fischer, T. Brox, U-net: Convolutional networks for biomedical image segmentation, in: Medical image computing and computer-assisted intervention–MICCAI 2015: 18th international conference, Munich, Germany, October 5-9, 2015, proceedings, part III 18, Springer, 2015, pp. 234–241

    2015

  40. [40]

    Sallam, A

    O. Sallam, A. Glaws, M. Sanchez-Gomez, G. Deskos, A. Cortiella, R. King, M. Fürth, Tropical cyclone super resolution using conditional diffusion denoising probabilistic model from mesoscale simulation to les, Environmental Research Communications 7 (12) (2025) 125025

    2025

  41. [41]

    Rybchuk, M

    A. Rybchuk, M. Hassanaly, N. Hamilton, P. Doubrawa, M. J. Fulton, L. A. Martínez-Tossas, Ensemble flow reconstruction in the atmospheric boundary layer from spatially limited measurements through latent diffusion models, Physics of Fluids 35 (12) (2023)

    2023

  42. [42]

    Y.Cheng,M.G.Giometto,P.Kauffmann,L.Lin,C.Cao,C.Zupnick,H.Li,Q.Li,Y.Huang,R.Abernathey,etal., Deep learning for subgrid-scale turbulence modeling in large-eddy simulations of the convective atmospheric boundary layer, Journal of Advances in Modeling Earth Systems 14 (5) (2022) e2021MS002847

    2022

  43. [43]

    Bartholomew, G

    P. Bartholomew, G. Deskos, R. A. Frantz, F. N. Schuch, E. Lamballais, S. Laizet, Xcompact3d: An open-source framework for solving turbulence problems on a cartesian mesh, SoftwareX 12 (2020) 100550

    2020

  44. [44]

    Laizet, E

    S. Laizet, E. Lamballais, High-order compact schemes for incompressible flows: A simple and efficient method with quasi-spectral accuracy, Journal of Computational Physics 228 (16) (2009) 5989–6015

    2009

  45. [45]

    International Electrotechnical Commission (IEC), Wind turbines - part 1: Design requirements, available at: https://www.iec.ch/(2019)

    2019

  46. [46]

    InternationalElectrotechnicalCommission(IEC),Windturbines-part3:Designrequirementsforoffshorewind turbines, available at:https://www.iec.ch/(2019)

    2019

  47. [47]

    Zheng, Y

    X. Zheng, Y. Yao, Z. Hu, Z. Yu, S. Hu, Influence of turbulence intensity on the aerodynamic performance of wind turbines based on the fluid-structure coupling method, Applied Sciences 13 (1) (2022) 250

    2022

  48. [48]

    M.J.Churchfield,S.Lee,J.Michalakes,P.J.Moriarty,Anumericalstudyoftheeffectsofatmosphericandwake turbulence on wind turbine dynamics, Journal of turbulence (13) (2012) N14

    2012

  49. [49]

    Y. Kang, S. Lee, Wake effect on floating offshore wind turbine fatigue load, Physics of Fluids 36 (12) (2024)

    2024

  50. [50]

    J. Berg, J. Bryant, B. LeBlanc, D. C. Maniaci, B. Naughton, J. A. Paquette, B. R. Resor, J. White, D. Kroeker, Scaled wind farm technology facility overview, in: 32nd ASME Wind Energy Symposium, 2014, p. 1088

    2014

  51. [51]

    S. B. Pope, Turbulent flows, CambridgeUniversityPress (2000) 558–638

    2000

  52. [52]

    S. B. Pope, Ten questions concerning the large-eddy simulation of turbulent flows, New journal of Physics 6 (1) (2004) 35

    2004

  53. [53]

    2256–2265

    J.Sohl-Dickstein,E.Weiss,N.Maheswaranathan,S.Ganguli,Deepunsupervisedlearningusingnonequilibrium thermodynamics, in: International conference on machine learning, pmlr, 2015, pp. 2256–2265

    2015

  54. [54]

    Dhariwal, A

    P. Dhariwal, A. Nichol, Diffusion models beat gans on image synthesis, Advances in neural information processing systems 34 (2021) 8780–8794

    2021

  55. [55]

    K. He, X. Zhang, S. Ren, J. Sun, Deep residual learning for image recognition, in: Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 770–778. Omar Sallam et al.:Preprint submitted to Elsevier Page 38 of 38

    2016

  56. [56]

    7354–7363

    H.Zhang,I.Goodfellow,D.Metaxas,A.Odena,Self-attentiongenerativeadversarialnetworks,in:International conference on machine learning, PMLR, 2019, pp. 7354–7363

    2019

  57. [57]

    D. P. Kingma, J. Ba, Adam: A method for stochastic optimization, arXiv preprint arXiv:1412.6980 (2014). Omar Sallam et al.:Preprint submitted to Elsevier Page 39 of 38

    work page internal anchor Pith review arXiv 2014