Pith. sign in

REVIEW 3 major objections 5 minor 21 references

Predicting Turbulence Structure In Street-Canyon Flows using Deep Generative Modeling

T0 review · 3 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A transformer-based generative model trained on wind-tunnel data from one street-canyon geometry reproduces the turbulence statistics, flow structures, and short-term evolution of a different geometry.

desk verdict Useful cross-aspect-ratio transfer for single-time statistics, but the temporal-forecasting claim collapses under the 7 Hz PIV sampling. read the letter →

arxiv 2501.13415 v1 pith:6ZAUQKGO submitted 2025-01-23 physics.flu-dyn

Tomek Jaroslawski , Aakash Patil , Beverley McKeon This is my paper · ORCID
classification physics.flu-dyn PACS 47.27.-i
keywords deepgenerativemodelingstreetcanyonturbulencepredictionparticleimagevelocimetryautoregressivetransformerquadrantanalysisproperorthogonaldecompositionurbanwindengineering
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper claims that a convolutional encoder-decoder transformer, trained autoregressively on 10,000 particle-image-velocimetry snapshots of a wide (aspect-ratio 3) street canyon in the wake-interference regime, can generate roof-level velocity fields in a different canyon (aspect-ratio 1, skimming regime) that match the measured mean flow, fluctuation levels, two-point correlations, quadrant-event statistics, and the spatial organization of the dominant proper-orthogonal-decomposition modes. The transfer matters because roof-level turbulence governs pollutant exchange between the canyon and the outer flow, so a model that crosses geometries could serve as a fast surrogate for wind and air-quality studies where experiments or simulations are expensive. The paper also claims the model tracks the temporal evolution of ejection events for about 500 generated snapshots, while noting that the 7 Hz sampling of the training data likely forces the model to rely mostly on spatial information and that generated fluctuation variances and modal energies run systematically low.

What carries the argument

The load-bearing component is the autoregressive loop in a convolutional encoder-decoder transformer. The encoder compresses a short window of velocity snapshots into a latent representation, the self-attention layers reweight features across channels and spatial positions, and the decoder up-samples back to full fields; at inference each predicted field is fed back as input for the next prediction. The training depth is two autoregressive steps, the loss is a mean-squared error summed over prediction steps, and zero-padding standardizes fields from the two canyon geometries to a common dimension.

What would settle it

Train the identical architecture on a velocity time series recorded at a much higher temporal rate (or on a temporally resolved large-eddy simulation at the same Reynolds number) and test whether the autoregressive horizon extends beyond 500 snapshots and whether the fluctuation standard deviations and POD eigenvalues stop falling systematically below the experimental reference; if the underestimates persist at high temporal resolution, the spatial-transfer claim would survive but the temporal-prediction claim would be falsified.

Watch

Extended reading notes

Core claim

The central claim is that the spatial structure of roof-level turbulence in one street-canyon flow regime carries enough information to predict the turbulence structure of another regime. Trained on the wide canyon, the model generates 500 time steps for the narrow canyon, and its predictions agree with the experimental reference in the time- and spanwise-averaged streamwise and vertical velocity profiles, in the two-point correlation function $R_{uu}$ near the canyon centre, in the quadrant hole-size contributions to the Reynolds shear stress, and in the shape of the first two proper-orthogonal-decomposition (POD) modes that represent the large-scale separated shear layer. The eigenvalues of the generated field are systematically smaller than the measured ones across all modes, and the predicted standard deviations fall slightly below the experiment, which the authors attribute to the model filtering small-scale energy. On the temporal side, the model initialized from four snapshots locates quadrant-analysis Q2 (ejection) events correctly in the early part of the sequence, deteriorating by snapshot 500, a horizon the authors link to the limited temporal resolution of the 7 Hz PIV training data.

Load-bearing premise

The load-bearing premise is that the 7 Hz PIV sampling, roughly one frame per seventeen eddy turnover times, still contains enough temporal dynamics for autoregressive training to learn physically meaningful time evolution; if the sampling is too sparse, the model is learning spatial statistics and the temporal-prediction claim lacks support.

Editorial extensions

If this is right

  • A model trained on one canyon configuration can generate realistic roof-level turbulence statistics for another configuration without retraining, offering a fast surrogate for wind-environment studies.
  • Continuing a flow from four initial snapshots suggests a route to state reconstruction from sparse sensor data, which could support forecasting transient pollutant releases.
  • Because the generated fields match two-point correlations and dominant POD modes, they could serve as synthetic turbulence inputs for dispersion models in geometries where experimental data are scarce.
  • The systematic underestimate of fluctuation variances and POD eigenvalues implies the model acts as a low-pass filter of turbulence; the paper's proposed fixes—training on fluctuation snapshots, adding layers, and improving temporal sampling—are concrete routes to extend the valid horizon.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The successful transfer across aspect ratios suggests the model has latched onto the roof-level separated shear layer, a structure common to both regimes, rather than memorizing canyon-specific statistics; this predicts transfer to intermediate aspect ratios or different roughness arrangements, a test the paper does not run.
  • The framework is called generative, but its objective is a deterministic mean-squared-error loss, so it cannot sample new turbulent realizations; replacing the loss with a distributional one (e.g., a diffusion or adversarial component) is a direct test of whether the low fluctuation variances are a modeling choice or a data limitation.
  • A horizon of 500 snapshots covers roughly 8,500 eddy turnover times while each training frame is about 17 eddy turnovers apart, which hints that the model is doing spatial pattern continuation as much as temporal prediction; training on temporally shuffled versus correctly ordered sequences would separate the two contributions.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes a convolutional encoder-decoder transformer with autoregressive training to predict spatio-temporal velocity fields in street-canyon flows. The model is trained on 10,000 PIV snapshots from the C3hR3h (wake-interference, aspect ratio 3) configuration and then applied to the C1hR1h (skimming, aspect ratio 1) configuration, a genuinely held-out case. The authors report agreement with experimental data for mean velocity profiles, two-point spatial correlations, quadrant/hole statistics, and the spatial organization of leading POD modes; they also claim strong agreement in the temporal evolution of Q2 events up to 500 generated snapshots, after which error propagation causes divergence. The manuscript is candid about limitations, including systematically low velocity fluctuations, low POD eigenvalues, and an explicit statement that temporal resolution may be insufficient.

Significance. If the spatial-generative result holds, the paper demonstrates a non-trivial cross-regime generalization: a model trained on one canyon geometry/flow regime reproduces single-time statistical structure in a different geometry/regime, and the test configuration is not used in training, so circularity is avoided. The comparison suite is appropriate (mean statistics, two-point correlations, quadrant analysis, POD) and the authors deserve credit for reporting the deficiencies they observe rather than only favorable metrics. However, the temporal-prediction claim, which appears in the abstract and in Section 4.3, is the load-bearing part of the paper's framing and is not supported by the evidence; the low sampling rate and the qualitative nature of the temporal comparison are central concerns. The paper would be significantly stronger as a spatial turbulence generator with honest scope limitations, or with rigorous quantitative temporal verification.

major comments (3)
  1. [§2, §4.1, Eq. (2)] The 7 Hz PIV sampling implies that consecutive snapshots are separated by roughly 17 eddy turnover times (10,000 snapshots over approximately 170,000 turnovers). For roof-level turbulence, velocity fields decorrelate on timescales of a few turnover times, so the training pairs (X_t, X_{t+Δt}) used in the MSE loss of Eq. (2) are nearly statistically independent. The optimal solution under such a loss is the conditional mean, which would produce systematically reduced fluctuations and lower POD eigenvalues, exactly the behavior reported in Sections 4.2 and 4.4. The authors' own statement in Section 4.1 — 'this issue may stem from the dataset's lack of temporal resolution, forcing the model to rely primarily on spatial information to infer flow dynamics' — acknowledges this mechanism. This undermines the abstract's claim of 'strong agreement with experimental data in capturing the temporal evolution of flow dynamics.' I ask the authors to quantify the temporal autocorrelation of the training data and to demonstrate predictive skill beyond baselines such as persistence or randomly reordered snapshots, using a temporal correlation or anomaly-correlation metric, or to explicitly reframe the contribution as spatial generative modeling.
  2. [§4.3, Fig. 8] The temporal comparison in Fig. 8 consists of selected Q2-event snapshots at N = 20, 21, 22, 200, and 500, with no pointwise temporal alignment, no quantitative error metric, and no baseline such as persistence or a randomly sampled sequence of experimental snapshots. Given the chaotic nature of turbulence and the fact that the model is initialized from only four snapshots, visual similarity of event locations in a handful of frames cannot establish temporal forecasting skill. I request quantitative metrics (e.g., field correlation at matched times, event-centroid tracking error, or spectral coherence) and a comparison against a non-predictive baseline. Without such evidence, the claim that the model 'accurately forecasting quadrant events well into the temporal evolution' is not supported.
  3. [§4.1, §4.4] All statistical analyses are restricted to the first 500 generated snapshots, a truncation chosen after observing that the model diverges afterward. This post hoc selection may bias the reported agreement, especially because 500 snapshots at Δt ≈ 17 T is still a small number of independent samples for statistics such as POD eigenvalues and quadrant hole analysis. The authors should report how the statistics vary with the number of snapshots used, provide results from multiple independent rollouts (which would also increase the effective sample size), and show the behavior beyond 500 steps rather than only the favorable window.
minor comments (5)
  1. [Throughout] The manuscript contains numerous duplications and remnants of a conference-paper format, including repeated figure captions, duplicated architecture diagrams, and duplicated results text. The manuscript needs a thorough editorial cleanup before publication.
  2. [Abstract and §3] The model is deterministic and is trained on a single configuration (C3hR3h), yet the text repeatedly calls it 'deep generative' and states that the training dataset 'contains diverse flow regimes.' These descriptions overstate the generative and multi-regime character of the method; please temper the wording to match the actual setup.
  3. [Eq. (3)] The formula for the two-point correlation Ruu omits ensemble averaging in the numerator and denominator; as written, it defines a pointwise product rather than a correlation coefficient. Please add the appropriate averaging notation.
  4. [§2] The definition of eddy turnover time T = h/Ue and the relation between the 10,000 snapshots and 170,000 turnovers should be stated more clearly, since the sampling-rate issue is central to the temporal claims.
  5. [Throughout] There are inconsistencies in notation, such as 'C1hC1h' versus 'C1hR1h' and 'Ch3R3h' versus 'C3hR3h', which should be unified.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the C1hR1h test is a held-out configuration, and the validation statistics are not optimization targets.

full rationale

The paper's central transfer claim is not circular. The model is trained on the C3hR3h wake-interference configuration, while the predictions are evaluated on the C1hR1h skimming-flow configuration, which is not used for parameter fitting. The only use of the test data is the first four snapshots as an autoregressive initial condition, which is an input, not a fitted target. The training loss in Eq. (2) minimizes squared error on instantaneous one-step velocity fields; the validation metrics--mean profiles, standard deviations, two-point correlations, quadrant hole statistics, and POD eigenvalues--are derived quantities not equal to this loss, so agreement on them is a genuine transfer result rather than a restatement of the training objective. The citations [13,14] provide the experimental wind-tunnel dataset; citing one's own or one's group's measured data as training data is not load-bearing circularity because the data are external measurements, and the test configuration provides a separate benchmark. The hand-chosen truncation of the analysis to the first 500 generated snapshots is a selection issue, not a circularity issue. The paper's own admission in Section 4.1 that limited temporal resolution forced the model to rely primarily on spatial information, and the resulting systematically low variances and POD eigenvalues, are correctness and evidence-quality concerns about the temporal-evolution claim; they do not reduce the derivation to its inputs by definition. No equation or fitted parameter was found to be equivalent to the quantity it is claimed to predict.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The model is an empirical ML surrogate: network weights are fitted to PIV snapshots and all physics enters through the training data and standard post-processing assumptions. The main domain assumptions are Reynolds-number independence, representativeness of the 0.9h plane, sufficiency of 7 Hz sampling for temporal learning, and zero-padding preserving geometry; none are demonstrated within the paper.

free parameters (4)
  • Neural network weights (conv-transformer encoder-decoder) = Fitted to 10,000 PIV snapshots (C3hR3h); values not released
    Trained with MSE loss (Eq. 2); the transfer claim is carried entirely by these learned weights.
  • Autoregressive training sequence length = 2
    Set to two in order to limit computational cost (Section 3); determines how temporal consistency is learned.
  • Number of generated snapshots used for evaluation = 500
    Analysis restricted to the first 500 snapshots after observing significant error propagation (Section 4.1); post-hoc choice.
  • Training hyperparameters (learning rate, batch size, epochs, early-stopping patience, filter counts) = Not reported
    Hand-chosen but omitted; required for exact reproduction.
assumptions (5)
  • domain assumption PIV sampling at 7 Hz is sufficient for the autoregressive model to learn temporal dynamics
    Authors admit the dataset's lack of temporal resolution forces reliance on spatial information (Section 4.1); the temporal claim depends on this being adequate.
  • domain assumption Reynolds number independence holds at Reh approximately 3e4, so 1:200 scale wind-tunnel results represent real urban flows
    Invoked with citation [15] in Section 2.
  • domain assumption The wall-parallel plane at 0.9h is the critical region for pollutant exchange and is sufficient to characterize canyon turbulence
    Stated in abstract and introduction; not validated against other measurement planes.
  • ad hoc to paper Zero-padding standardizes inputs without destroying the geometric information that lets the model generalize from AR=3 to AR=1
    Section 2; no ablation or geometric conditioning is provided to justify transfer.
  • standard math Standard PIV, POD, and quadrant-analysis methods provide unbiased ground truth for training and evaluation
    Assumed background in experimental fluid mechanics.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Predicting Turbulence Structure In Street-Canyon Flows using Deep Generative Modeling." pith.science (2026). https://pith.science/paper/6ZAUQKGO

@misc{pith2026250113415,
  author       = {Pith},
  title        = {Pith review of: Predicting Turbulence Structure In Street-Canyon Flows using Deep Generative Modeling},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6ZAUQKGO}},
  note         = {Machine review of arXiv:2501.13415}
}
read the original abstract

The high dimensionality and complex dynamics of turbulent flows in urban street canyons present significant challenges for wind and environmental engineering, particularly in addressing air quality, pollutant dispersion, and extreme wind events. This study introduces a deep learning framework to predict spatio-temporal flow behavior in street canyons with varying geometric configurations and upstream roughness conditions. A convolutional encoder-decoder transformer model, trained on particle image velocimetry (PIV) data from wind tunnel experiments, is employed with autoregressive training to predict flow characteristics. The training dataset contains diverse flow regimes, with a focus on the wall-parallel plane near the canyon roof, a critical region for pollutant exchange between the outer flow and the canyon interior. The model accurately reproduces key flow features, including mean turbulent statistics, two-point correlations, quadrant analysis, and dominant flow structures, while demonstrating strong agreement with experimental data in capturing the temporal evolution of flow dynamics. These findings demonstrate the potential of deep learning models to enhance predictive capabilities for urban canyon flows, offering a pathway toward improved urban design, sustainable environmental management, and more effective pollutant dispersion modeling.

Figures

Figures reproduced from arXiv: 2501.13415 by the authors.

Figure 1
Figure 1. (Experimental setup (top) and configurations (bottom). [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Streamwise velocity field snapshots of experimental data for both wide ( [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Time- and spatially- averaged turbulent statistics: (a) streamwise velocity component and (b) vertical velocity component observed in both the training data and the predicted data. ML represents the model’s prediction, while PI’ references the experimental data. 3.1. Mean flow The ML model is trained using the C3hR3h configuration, and its predictive capabilities are evaluated by applying it to the C1hR1h configurat… view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Convergence of the mean velocity, standard deviation, and skewness at a single point [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 3
Figure 3. Figure 3: Time- and spatially- averaged turbulent statistics: (a) streamwise velocity component and (b) vertical velocity component observed in both the training data and the predicted data. ML represents the model’s prediction, while PI’ references the experimental data. 3.1. M…
Figure 3
Figure 3. Figure 3: Time- and spatially- averaged turbulent statistics: (a) streamwise velocity component and (b) vertical velocity component observed in both the training data and the predicted data. ML represents the model’s prediction, while PI’ references the experimental data. 3.1. M…
Figure 5
Figure 5. Figure 5: (a) Two-point correlation fields, Ruu, of the ML model and PIV dat Convolutional encoder-decoder transformer deep learning architecture : Model archi (a) c con con con 05 0.6 in the x-z p a canyon w F3Tid tilld tblt ttiti() t CONCLU In this (a) The canonical four-stag…
Figure 3
Figure 3. Figure 3: Time- and spatially- averaged turbulent statistics: (a) d (b) til litt bd ibth thtii 1 In th that inte a remarkable agreement is observed between the model’s re MODEL The ML model is trained using the C3hR3h configuration, and its predictive capabilities modes 1, 2, 4,…
Figure 3
Figure 3. Figure 3: Time- and spatially- averaged turbulent statistics: (a) streamwise velocity component and (b) vertical velocity component observed in both the training data and the predicted data. ML represents the model’s prediction, while PI’ references the experimental data. 3.1. M…
Figure 3
Figure 3. Figure 3: Time- and spatially- averaged turbulent statistics: (a) streamwise velocity component and (b) vertical velocity component observed in both the training data and the predicted data. ML represents the model’s prediction, while PI’ references the experimental data. 3.1. M…
Figure 3
Figure 3. Figure 3: Time- and spatially- averaged turbulent statistics: (a) streamwise velocity component and (b) vertical velocity component observed in both the training data and the predicted data. ML represents the model’s prediction, while PI’ references the experimental data. 3.1. M…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

21 extracted references · 20 canonical work pages

  1. [1]

    Assessment of inner–outer interactions in the urban boundary layer using a predictive model

    Karin Blackman, Laurent Perret, and Romain Mathis. Assessment of inner–outer interactions in the urban boundary layer using a predictive model. Journal of Fluid Mechanics, 875:44–70, 2019

    work page 2019

  2. [2]

    Data-driven assessment of arch vortices in simplified urban flows

    Álvaro Martínez-Sánchez, Eneko Lazpita, Adrián Corrochano, Soledad Le Clainche, Sergio Hoyas, and Ricardo Vinuesa. Data-driven assessment of arch vortices in simplified urban flows. International Journal of Heat and Fluid Flow, 100:109101, 2023

    work page 2023

  3. [3]

    The transformative potential of machine learning for experiments in fluid mechanics

    Ricardo Vinuesa, Steven L Brunton, and Beverley J McKeon. The transformative potential of machine learning for experiments in fluid mechanics. Nature Reviews Physics, 5(9):536–545, 2023

    work page 2023

  4. [4]

    Pedestrian exposure to black carbon and pm2

    Honey Dawn Alas, Almond Stöcker, Nikolaus Umlauf, Oshada Senaweera, Sascha Pfeifer, Sonja Greven, and Alfred Wiedensohler. Pedestrian exposure to black carbon and pm2. 5 emissions in urban hot spots: new findings using mobile measurement techniques and flexible bayesian regression models. Expo Environ Epidemiol, 32:604–614, 2022. 14 Predicting Turbulence ...

    work page 2022

  5. [5]

    Study of interscale interactions for turbulence over the obstacle arrays from a machine learning perspective

    Wanting Liu, Yajun Zou, and Xuebo Li. Study of interscale interactions for turbulence over the obstacle arrays from a machine learning perspective. Phys Fluids, 35, 2023

    work page 2023

  6. [6]

    Using machine learning to predict urban canopy flows for land surface modeling

    Yanle Lu, Xu-Hui Zhou, Heng Xiao, and Qi Li. Using machine learning to predict urban canopy flows for land surface modeling. Geophys Res Lett, 50:e2022GL102313, 2023

    work page 2023

  7. [7]

    A reduced order model for turbulent flows in the urban environment using machine learning

    Dunhui Xiao, CE Heaney, L Mottet, F Fang, W Lin, IM Navon, Y Guo, OK Matar, AG Robins, and CC Pain. A reduced order model for turbulent flows in the urban environment using machine learning. Build Environ, 148:323–337, 2019

    work page 2019

  8. [8]

    Machine learning accelerated turbulence modeling of transient flashing jets

    David Schmidt, Romit Maulik, and Konstantinos Lyras. Machine learning accelerated turbulence modeling of transient flashing jets. Phys Fluids, 33, 2021

    work page 2021

Show all 21 references
  1. [9]

    A novel spatial-temporal prediction method for unsteady wake flows based on hybrid deep neural network

    Renkun Han, Yixing Wang, Yang Zhang, and Gang Chen. A novel spatial-temporal prediction method for unsteady wake flows based on hybrid deep neural network. Phys Fluids, 31, 2019

    2019

  2. [10]

    Convolutional neural network and long short-term memory based reduced order surrogate for minimal turbulent channel flow

    Taichi Nakamura, Kai Fukami, Kazuto Hasegawa, Yusuke Nabae, and Koji Fukagata. Convolutional neural network and long short-term memory based reduced order surrogate for minimal turbulent channel flow. Phys Fluids, 33, 2021

    2021

  3. [11]

    Predictive models for flame evolution using machine learning: apriori assessment in turbulent flames without and with mean shear

    Jiahao Ren, Haiou Wang, Guo Chen, Kun Luo, and Jianren Fan. Predictive models for flame evolution using machine learning: apriori assessment in turbulent flames without and with mean shear. Phys Fluids, 33, 2021

    2021

  4. [12]

    Identifying regions of importance in wall- bounded turbulence through explainable deep learning

    Andrés Cremades, Sergio Hoyas, Rahul Deshpande, Pedro Quintero, Martin Lellep, Will Junghoon Lee, Jason P Monty, Nicholas Hutchins, Moritz Linkmann, Ivan Marusic, et al. Identifying regions of importance in wall- bounded turbulence through explainable deep learning. Nature Com...

    2024

  5. [13]

    The spanwise variation of roof-level turbulence in a street-canyon flow

    Thomas Jaroslawski, Laurent Perret, Karin Blackman, and Eric Savory. The spanwise variation of roof-level turbulence in a street-canyon flow. Bound-Layer Meteorol, 170:373–394, 2019

    2019

  6. [14]

    Roof-level large-and small-scale coherent structures in a street canyon flow

    Thomas Jaroslawski, Eric Savory, and Laurent Perret. Roof-level large-and small-scale coherent structures in a street canyon flow. Environ Fluid Mech, 20:739–763, 2020

    2020

  7. [15]

    The flow around a surface-mounted cube in uniform and turbulent streams

    IP Castro and AG Robins. The flow around a surface-mounted cube in uniform and turbulent streams. Journal of fluid Mechanics, 79(2):307–335, 1977

    1977

  8. [16]

    Adam: Method for stochastic optimization

    Diederik P Kingma and Jimmy Ba. Adam: Method for stochastic optimization. arXiv Preprint, 2014

    2014

  9. [17]

    Rectified linear units improve restricted boltzmann machines

    Vinod Nair and Geoffrey E Hinton. Rectified linear units improve restricted boltzmann machines. In Proc 27th ICML-10, pages 807–814, 2010

    2010

  10. [18]

    Tensorflow: a system for large-scale machine learning

    Martín Abadi, Paul Barham, Jianmin Chen, Zhifeng Chen, Andy Davis, Jeffrey Dean, Matthieu Devin, Sanjay Ghemawat, Geoffrey Irving, Michael Isard, et al. Tensorflow: a system for large-scale machine learning. In 12th USENIX Sym Oper Syst Des Implement (OSDI 16), pages 265–283, 2016

    2016

  11. [19]

    Street design and urban canopy layer climate

    Tim R Oke. Street design and urban canopy layer climate. Ener Build, 11:103–113, 1988

    1988

  12. [20]

    Quadrant analysis in turbulence research: history and evolution

    James M Wallace. Quadrant analysis in turbulence research: history and evolution. Annual Review of Fluid Mechanics, 48:131–158, 2016

    2016

  13. [21]

    Turbulence and the dynamics of coherent structures

    Lawrence Sirovich. Turbulence and the dynamics of coherent structures. i. coherent structures. Quarterly of applied mathematics, 45(3):561–571, 1987. 15

    1987

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.