Pith. sign in

REVIEW 4 major objections 4 minor 30 references

Radiosonde measurement uncertainty can be propagated through state-space Monte Carlo simulation into planetary boundary layer height estimates, yielding point values and uncertainty intervals where standard methods give only point values.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

A state-space and Monte Carlo framework propagates GRUAN radiosonde uncertainties into planetary boundary layer height estimates and reports 95% uncertainty ranges.

T0 review reviewed 2026-08-02 challenge →

load-bearing objection A genuinely useful first attempt at propagating GRUAN measurement uncertainty into PBLH estimates, but a concrete covariance inconsistency and missing validation keep the central claim conditional. the 4 major comments →

arxiv 2607.14960 v2 pith:PB66MSS4 submitted 2026-07-16 stat.AP

Statistical Modelling of Planetary Boundary Layer Height and Its Measurement Uncertainty Using GRUAN Profiles

Tommaso Locatelli , Alessandro Fasso , Fabio Madonna This is my paper
classification stat.AP
keywords planetary boundary layer heightmeasurement uncertainty propagationstate-space modellocal linear trendKalman smoothingsimulation smoothingMonte Carlo uncertaintyradiosonde profiles
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

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 asks whether the traceable per-height measurement uncertainties attached to reference radiosonde profiles can be carried through to the derived quantity everyone actually uses: the height of the planetary boundary layer (PBLH). It argues that standard plug-in retrievals cannot do this, because the diagnostics that define PBLH methods are non-differentiable, so the usual error-propagation law stops at the gradient stage. The proposed fix embeds the profile variables in a local-linear-trend state-space model, smooths them with a recursive filter, and then draws many simulated profiles consistent with both observations and their uncertainties. Each draw is passed through the PBLH criteria, so the result is a full distribution of plausible boundary-layer heights. A sympathetic reader would care because it converts a long-standing missing quantity — the measurement uncertainty of PBLH — into a computable interval, and it shows preliminary signs of correcting spurious low heights that noisy near-surface gradients produce in standard gradient-based methods.

Core claim

The paper's claim is that measurement uncertainty can be propagated into PBLH estimates by treating the observed vertical profile as a local-linear-trend state-space process with known observation noise (from the reference radiosonde data product) and unknown state-noise variances estimated by maximum likelihood. After recursive state-space smoothing, gradients are formed as ratios of smoothed rates of change rather than finite differences of raw data, and whole-state trajectories are simulated from their conditional distribution. Applying each PBLH criterion to hundreds (M=200) of simulated profiles yields a Monte Carlo distribution whose median is the revised PBLH estimate and whose 95% pe

What carries the argument

The carrying mechanism is the local-linear-trend state-space model: each measured variable (altitude, virtual potential temperature, relative humidity, wind components) evolves as a random walk in level plus a stochastic slope, with measurement noise fixed to the profile's reported standard uncertainties and slope-noise variances estimated by maximum likelihood. A recursive smoother delivers posterior means and covariances for every level and slope; simulation smoothing then generates whole vertical profiles drawn from the posterior state distribution. These draws are converted, via smoothed slope ratios, into the diagnostic variables — virtual-potential-temperature gradient, humidity gradie

Load-bearing premise

The load-bearing assumption is that the local-linear-trend state-space model with independent Gaussian disturbances represents the real atmospheric profile structure closely enough that the smoothed gradients are unbiased and the simulated draws are physically plausible; the paper itself postpones the residual diagnostics and oversmoothing checks that would verify this.

What would settle it

Take a set of synthetic temperature, humidity, and wind profiles with a known injected inversion height, add noise of the magnitude reported by the reference radiosonde product, run the procedure, and check whether the Monte Carlo median tracks the true height and whether the 95% intervals cover it at the nominal rate over many replicates. If the smoother flattens the sharp capping inversion, the intervals will be systematically misplaced; a residual test for non-Gaussian or serially correlated smoothing errors would similarly expose a misspecified noise model.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • For any single radiosonde profile, each PBLH method now returns a median height and a 95% uncertainty interval instead of a bare number, with no ground-truth reference required.
  • Gradient-based PBLH retrievals become less prone to the spurious very-low heights caused by noise in finite-difference gradients near the surface; the aggregated distributions shift upward and become smoother.
  • Method-specific ambiguity becomes visible: humidity-gradient estimates show wide, often multimodal uncertainty whenever several humidity drops exist in the profile.
  • The framework generalizes to other PBLH diagnostics beyond the four tested, since it only requires a criterion function applied to simulated diagnostic profiles.
  • Preliminary seasonal and diurnal patterns (lower PBLH at night, summer maximum at mid-latitudes, weaker tropical cycle, hemisphere flip at a southern site) survive the Monte Carlo treatment, so the added uncertainty does not erase known climatology.

Where Pith is reading between the lines

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

  • If the gain in robustness comes from smoothing rather than from the specific state-space prior, then simpler smoothers might deliver similar PBLH medians at lower computational cost; comparing against spline or wavelet pre-filtering would isolate the source of the improvement.
  • The uncertainty intervals assume independent, Gaussian measurement errors with a diagonal covariance; if the reference profile errors are correlated in height or heavier-tailed, the 95% ranges may be miscalibrated — a testable prediction, since the paper defers residual diagnostics to future work.
  • A direct validation path is to build synthetic profiles with a known injected inversion height and realistic noise, then check whether the Monte Carlo median tracks the truth and the 95% intervals achieve nominal coverage over many replicates.
  • If adopted by users of reanalysis or air-quality models, the output distribution could be used as an ensemble of boundary-layer heights rather than a single value, letting downstream products carry measurement uncertainty through the whole chain — an extension the paper does not develop.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. The paper proposes a Monte Carlo framework for propagating GRUAN radiosonde measurement uncertainties into PBLH retrievals. It replaces the standard plug-in application of the parcel, θv-gradient, RH-gradient, and Richardson-number methods with a local-linear-trend state-space model; after MLE estimation of the disturbance covariance Q_t, simulation smoothing draws profiles from p(S_n | X̄_n), and the median and 95% percentile range of the resulting PBLH ensemble are reported as estimate and uncertainty. The method is illustrated on one year of 00/12 UTC soundings from Lindenberg, Hong Kong, and Lauder, with case studies and aggregated tables by time of day and season. The authors position the work as a first step and explicitly defer model-validation tasks.

Significance. The paper addresses a real gap: PBLH retrievals from radiosondes rarely report quantitative uncertainties, and GRUAN's traceable uncertainties are well suited to propagation. If the state-space model were validated, the MC approach would provide both a principled uncertainty interval and a noise-robust alternative to finite-difference gradient retrievals. The manuscript ships reproducible code and uses publicly available GRUAN data, which is a strength. The contribution is, however, not yet established: the reported intervals are conditional on an unvalidated LLT+Gaussian model, and several specification errors prevent reproduction and undermine the standard-versus-MC comparisons.

major comments (4)
  1. [Section 3.3, observation equation and diag(H_t)] The observation vector has five elements (z, θv, RH, u, v), but the stated diag(H_t) lists seven variances: σ²_z, σ²_T, σ²_p, σ²_RH, σ²_r, σ²_u, σ²_v. The T, p, and r variances cannot be inserted into a 5×5 measurement-noise covariance, and the text says σ_θv is obtained by propagation. Replace the displayed vector with the five variances actually used (including σ²_θv), otherwise the model cannot be reproduced.
  2. [Sections 3.1, 3.2, and Eq. (2)] The Richardson diagnostic is defined three incompatible ways. Section 3.1 gives a local gradient Richardson number, (g/θ_v)(∂θ_v/∂z)/[(∂u/∂z)²+(∂v/∂z)²]; Section 3.2's plug-in uses a bulk form from the surface, (g/θ_v)(θ_v,t−θ_v,1)(z_t−z_1)/[(u_t−u_1)²+(v_t−v_1)²]; and Eq. (2) reverts to the local gradient form. The comparison of 'standard' and MC methods is therefore not a comparison of the same retrieval, and the method name is ambiguous.
  3. [Section 3.3, state equation for u and v] The state equation for u_t uses Λ_u,t (current slope) while all other variables use the lagged slope Λ_·,t−1; the same applies to v_t. As written this is not an LLT for the wind components and introduces an asymmetric temporal indexing. State whether this is intentional; if a typo, correct it and confirm the implementation used the same indexing.
  4. [Sections 4.1 and 5] The central claim—that this methodology 'quantifies' PBLH uncertainty—requires that the MC intervals be calibrated. The intervals are draws from p(S_n | X̄_n) under the LLT+Gaussian model with Q_t estimated by MLE from the same profile; they do not include model or parameter uncertainty. The paper explicitly defers residual checks and oversmoothing assessment (Section 4.1), and no synthetic-truth or independent-reference experiment demonstrates coverage or shows that MC medians are closer to a reference than plug-in estimates. Without such validation, the reported widths are only a model-conditional spread, and the abstract's claim to quantify uncertainty and refine the retrieval is not established.
minor comments (4)
  1. [Throughout] Typos: 'Exisiting' (Introduction), 'ballon burst' (Sec. 3.1), 'indipendent' and 'potential virtual temperature' (Sec. 3.3), 'algoorithm' (Introduction). A language pass is needed.
  2. [Section 3.4] M=200 is justified by 'empirical analyses' but no convergence or sensitivity analysis is reported; a brief study of M (e.g., 50/100/500) would support the choice.
  3. [Section 4.2 and Tables] The tables after the references use 'T od' as a column header; this should be 'TOD' or 'Time of day'. Also check figure callouts: 'figures 11 and Table 2' for Lindenberg, and later uses of Figures 12 and 13, so numbering should be verified.
  4. [Section 2] The GCOS/GRUAN acronym expansion is slightly awkward; check for consistency with the standard naming conventions.

Circularity Check

1 steps flagged

No load-bearing circularity; MC PBLH uncertainty is genuine model-based propagation, though one self-consistency check is tautological and external validation is deferred.

specific steps
  1. other [Section 4.1, Figure 4 caption and following paragraph]
    "As expected, the simulated profiles remain contained within these narrower uncertainty envelopes, confirming that the algorithm behaves consistently with the underlying state-space model assumptions."

    The simulated profiles are generated by simulation smoothing, which draws from p(S_n | Xbar_n). Their containment within the smoothed-state posterior envelopes is therefore a mathematical consequence of the sampling algorithm, not an independent confirmation of the LLT/Gaussian model. The paper itself defers the needed checks: 'Further analysis will include a detailed assessment of the residual distribution as well as checks for potential oversmoothing.' This is a self-referential validation, but it is not load-bearing for the main PBLH propagation result, which remains a genuine function of the observations, their GRUAN uncertainties, and the fitted state-space model.

full rationale

The paper's central derivation chain is not circular. GRUAN measurement uncertainties H_t enter the observation equation; the LLT state-space disturbance covariance Q_t is estimated by maximum likelihood; the Kalman smoother gives p(S_n | Xbar_n); simulation smoothing draws synthetic state profiles; the diagnostic function fbar and PBLH criterion g are applied to each draw; and the median and 95% percentile range summarize the resulting PBLH distribution. This is a Monte Carlo propagation of measurement uncertainty through a measurement function, consistent with GUM Supplement 1. No PBLH value is ever fitted as a parameter and then renamed as a prediction: the MC PBLH estimates are posterior sample statistics of g applied to simulated profiles, not direct fits to PBLH data. The MLE-fitted Q_t is a nuisance parameter, and ignoring its estimation uncertainty is a limitation rather than a circular reduction. The claimed 'increased robustness' is preliminary and is not externally benchmarked; the paper explicitly defers residual diagnostics and oversmoothing checks and acknowledges the absence of a PBLH ground truth, which is a validation gap, not circularity. Self-citations (e.g., Madonna et al. 2021, Summa et al. 2022) are contextual and not load-bearing; there is no author-imported uniqueness theorem or ansatz smuggled via self-citation. One minor self-referential consistency check is tautological by construction, but it does not support the central derivation, yielding a low circularity score. A separate internal inconsistency (diag(Ht) lists seven variances while the observation equation has five variables) is a reproducibility bug to correct, not evidence of circular reasoning.

Axiom & Free-Parameter Ledger

4 free parameters · 5 axioms · 0 invented entities

The paper introduces no new physical entities. Its contribution is a statistical uncertainty-propagation framework, which relies on fitted state-disturbance variances, Gaussian/diagonal noise assumptions, GRUAN uncertainty budgets, and standard PBLH criteria. The free parameters are model-engineering choices rather than physics parameters.

free parameters (4)
  • State disturbance covariance Q_t (10 diagonal variances per profile)
    Estimated by maximum likelihood for each sounding. Controls smoothness of the LLT and therefore the smoothed gradients and all downstream PBLH estimates and uncertainty widths.
  • Initial state variance values for MLE optimization
    Chosen for numerical stability. Authors say they are not critical, but they influence which local optimum is found.
  • Monte Carlo sample size M = 200
    Chosen based on 'empirical analyses' with no convergence diagnostics shown. The reported uncertainty widths depend on this choice.
  • PBLH estimator definition (median and 95% percentile range)
    Adopted because MC distributions may be skewed or multimodal, but the choice affects the reported h_MC and u(h_MC).
axioms (5)
  • ad hoc to paper Atmospheric profiles follow a local-linear-trend process with independent Gaussian state disturbances.
    Section 3.3 state equation. The central smoothing and simulation-smoothing machinery rests on this; oversmoothing is explicitly left for future work.
  • domain assumption Measurement disturbances are independent, Gaussian, and have diagonal covariance Ht equal to the GRUAN uncertainties.
    Section 3.3 observation equation. Correlations between variables and along the profile are ignored, which can understate uncertainty.
  • domain assumption GRUAN RS41-GDP.1 uncertainties are complete and unbiased.
    Section 2. If GRUAN uncertainties omit systematic components, the propagated PBLH uncertainty is incomplete.
  • domain assumption The Richardson-number criterion with threshold 0.25, using a local gradient definition, is a valid PBLH diagnostic.
    Section 3.1. The plug-in implementation in Section 3.2 uses a bulk Richardson number instead, creating an internal inconsistency.
  • domain assumption PBLH can be treated as an indirect measurement whose uncertainty is fully described by propagating measurement error through the deterministic criteria g.
    Section 3.1 and 3.4. This excludes structural method uncertainty and model-form uncertainty.

reviewed 2026-08-02 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Statistical Modelling of Planetary Boundary Layer Height and Its Measurement Uncertainty Using GRUAN Profiles." pith.science (2026). https://pith.science/paper/PB66MSS4

@misc{pith2026260714960,
  author       = {Pith},
  title        = {Pith review of: Statistical Modelling of Planetary Boundary Layer Height and Its Measurement Uncertainty Using GRUAN Profiles},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PB66MSS4}},
  note         = {Machine review of arXiv:2607.14960}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

The Planetary Boundary Layer (PBL) governs the exchange of energy and moisture and hosts the highest concentrations of pollutants before they mix into the free troposphere. The height of the PBL (PBLH) is therefore a key variable in meteorological and air-quality applications. Despite the wide range of methods available to derive PBLH from atmospheric observations, the associated uncertainties are rarely quantified. This study presents a methodology for propagating radiosonde measurement uncertainty into PBLH estimates obtained from state-of-the-art retrieval methods, including the parcel method, gradient-based methods, and the Richardson-number method. The framework relies on three components. First, it uses the GCOS Reference Upper-Air Network (GRUAN) Data Product, which provides traceable uncertainty estimates for all variables required in PBLH retrievals. Second, it employs a state-space model that captures the structure of atmospheric profiles and enables the generation of physically plausible simulated vertical profiles consistent with both observations and their uncertainties. Third, a Monte Carlo approach is used to propagate measurement uncertainty into the PBLH estimates, refining the retrieval and quantifying its uncertainty. Beyond providing uncertainty estimates, the methodology also shows preliminary signs of increased robustness in PBLH detection across several case studies, particularly in situations where standard gradient-based methods exhibit sensitivity to measurement uncertainty.

Figures

Figures reproduced from arXiv: 2607.14960 by Alessandro Fasso, Fabio Madonna, Tommaso Locatelli.

Figure 1
Figure 1. Figure 1: Profile of GRUAN RS41-GDP.1 sounding from the Lindenberg [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Day/night/twilight (left) and seasonal (right) distribution of ra [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Flow chart of the PBLH Monte Carlo procedure. [PITH_FULL_IMAGE:figures/full_fig_p017_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Example profile from the Lindenberg site. Comparison between [PITH_FULL_IMAGE:figures/full_fig_p019_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Profile from LIN during daytime (Local Launch Time: 2024-01-02 [PITH_FULL_IMAGE:figures/full_fig_p020_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Profile from LIN during nighttime (Local Launch Time: 2024-01-01 [PITH_FULL_IMAGE:figures/full_fig_p021_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Profile from HKO during daytime (Local Launch Time: 2024-01-03 [PITH_FULL_IMAGE:figures/full_fig_p022_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Profile from HKO during nighttime (Local Launch Time: 2024-01-01 [PITH_FULL_IMAGE:figures/full_fig_p023_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Profile from LAU during daytime (Local Launch Time: 2024-01-06 [PITH_FULL_IMAGE:figures/full_fig_p024_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Profile from LAU during nighttime (Local Launch Time: 2024- [PITH_FULL_IMAGE:figures/full_fig_p025_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Violin plots for all Lindenberg profiles, grouped by time of day. [PITH_FULL_IMAGE:figures/full_fig_p026_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Violin plots for all Hong Kong profiles, grouped by time of day. [PITH_FULL_IMAGE:figures/full_fig_p027_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Violin plots for all Lauder profiles, grouped by time of day. There [PITH_FULL_IMAGE:figures/full_fig_p028_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: Violin plots for all Lindenberg profiles, grouped by season. Each [PITH_FULL_IMAGE:figures/full_fig_p029_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: Violin plots for all Hong Kong profiles, grouped by season. For [PITH_FULL_IMAGE:figures/full_fig_p030_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: Violin plots for all Lauder profiles, grouped by season. For details, [PITH_FULL_IMAGE:figures/full_fig_p032_16.png] view at source ↗

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

30 extracted references · 7 canonical work pages

  1. [1]

    and Zhang, J

    Guo, J. and Zhang, J. and Shao, J. and Chen, T. and Bai, K. and Sun, Y. and Li, N. and Wu, J. and Li, R. and Li, J. and Guo, Q. and Cohen, J. B. and Zhai, P. and Xu, X. and Hu, F. , title =. Earth System Science Data , volume =. 2024 , number =

    2024

  2. [2]

    and Li, C

    Zhao, Z. and Li, C. and Xiao, Y. and others , title =. Environmental Fluid Mechanics , volume =. 2021 , doi =

    2021

  3. [3]

    and Angevine, Wayne M

    Grimsdell, Alison W. and Angevine, Wayne M. , title =. Journal of Applied Meteorology , volume =. 2002 , doi =

    2002

  4. [4]

    and Waliser, Duane E

    Ao, Chi O. and Waliser, Duane E. and Chan, Steven K. and Li, Jui-Lin and Tian, Baijun and Xie, Feiqin and Mannucci, Anthony J. , title =. Journal of Geophysical Research: Atmospheres , volume =. doi:https://doi.org/10.1029/2012JD017598 , url =. https://agupubs.onlinelibrary.wiley.com/doi/pdf/10.1029/2012JD017598 , abstract =

    work page doi:10.1029/2012jd017598

  5. [5]

    Park and Song-you Hong and Do-Hyeon Park and Sang-Woo Kim and Yujin J

    Hyeonmin Kim and Rokjin J. Park and Song-you Hong and Do-Hyeon Park and Sang-Woo Kim and Yujin J. Oak and Xu Feng and Haipeng Lin and Tzung-May Fu , keywords =. A mixed layer height parameterization in a 3-D chemical transport model: Implications for gas and aerosol simulations , journal =. 2024 , issn =. doi:https://doi.org/10.1016/j.scitotenv.2024.17683...

    arXiv 2024

  6. [6]

    , title =

    Stull, Roland B. , title =. 1988 , publisher =

    1988

  7. [7]

    , title =

    Holzworth, George C. , title =. Journal of Applied Meteorology , year =

  8. [8]

    Quarterly Journal of the Royal Meteorological Society , year =

    Stevens, Bjorn , title =. Quarterly Journal of the Royal Meteorological Society , year =

  9. [9]

    and Ao, Chi O

    Seidel, Dian J. and Ao, Chi O. and Li, Kun , title =. Journal of Geophysical Research: Atmospheres , volume =. doi:https://doi.org/10.1029/2009JD013680 , url =. https://agupubs.onlinelibrary.wiley.com/doi/pdf/10.1029/2009JD013680 , abstract =

    work page doi:10.1029/2009jd013680

  10. [10]

    and Zhang, Yehui and Beljaars, Anton and Golaz, Jean-Christophe and Jacobson, Andrew R

    Seidel, Dian J. and Zhang, Yehui and Beljaars, Anton and Golaz, Jean-Christophe and Jacobson, Andrew R. and Medeiros, Brian , title =. Journal of Geophysical Research: Atmospheres , volume =. doi:https://doi.org/10.1029/2012JD018143 , url =. https://agupubs.onlinelibrary.wiley.com/doi/pdf/10.1029/2012JD018143 , abstract =

    doi:10.1029/2012jd018143

  11. [11]

    and Puccioni, M

    Abraham, A. and Puccioni, M. and Jordan, A. and Maric, E. and Bodini, N. and Hamilton, N. and Letizia, S. and Klein, P. M. and Smith, E. N. and Wharton, S. and Gero, J. and Jacob, J. D. and Krishnamurthy, R. and Newsom, R. K. and Pekour, M. and Rad\"unz, W. and Moriarty, P. , TITLE =. Wind Energy Science , VOLUME =. 2025 , NUMBER =

    2025

  12. [12]

    Atmosphere , volume =

    Madonna, Fabio and Summa, Donato and Di Girolamo, Paolo and Marra, Fabrizio and Wang, Yuanzu and Rosoldi, Marco , title =. Atmosphere , volume =. 2021 , publisher =. doi:10.3390/atmos12030301 , url =

    work page doi:10.3390/atmos12030301 2021

  13. [13]

    and Beyrich, F

    Seibert, P. and Beyrich, F. and Gryning, S.-E. and Joffre, S. and Rasmussen, A. and Tercier, P. , title =. Atmospheric Environment , volume =. 2000 , doi =

    2000

  14. [14]

    Atmospheric Measurement Techniques , volume =

    Summa, Donato and Madonna, Fabio and Franco, Noemi and De Rosa, Benedetto and Di Girolamo, Paolo , title =. Atmospheric Measurement Techniques , volume =. 2022 , doi =

    2022

  15. [15]

    and Liu, B

    Li, H. and Liu, B. and Ma, X. and Jin, S. and Ma, Y. and Zhao, Y. and Gong, W. , TITLE =. Atmospheric Measurement Techniques , VOLUME =. 2021 , NUMBER =

    2021

  16. [16]

    and Hobbs, Peter V

    Wallace, John M. and Hobbs, Peter V. , title =. 2006 , publisher =

    2006

  17. [17]

    2011 , publisher =

    Burton, Tony and Jenkins, Nick and Sharpe, David and Bossanyi, Ervin , title =. 2011 , publisher =

    2011

  18. [18]

    Large-Eddy Simulation of Wind-Turbine Wakes: Evaluation of Turbine Parametrisations , journal =

    Wu, Yu-Ting and Port. Large-Eddy Simulation of Wind-Turbine Wakes: Evaluation of Turbine Parametrisations , journal =. 2011 , volume =. doi:10.1007/s10546-010-9569-x , url =

    work page doi:10.1007/s10546-010-9569-x 2011

  19. [19]

    Journal of Climate , year =

    Liu, Shuhua and Liang, Xin-Zhong , title =. Journal of Climate , year =. doi:10.1175/2010JCLI3552.1 , url =

    work page doi:10.1175/2010jcli3552.1

  20. [20]

    Wijnands, J. S. and Apituley, A. and Gouveia, D. A. and Noteboom, J. W. , TITLE =. Atmospheric Measurement Techniques , VOLUME =. 2024 , NUMBER =

    2024

  21. [21]

    Atmospheric Measurement Techniques , volume =

    Chen, Xi and Yang, Ting and Wang, Zifa and Wang, Futing and Wang, Haibo , title =. Atmospheric Measurement Techniques , volume =. 2023 , doi =

    2023

  22. [22]

    Chapter 2 - Balloon-borne radiosondes , editor =

    Holger Vömel and Bruce Ingleby , keywords =. Chapter 2 - Balloon-borne radiosondes , editor =. Field Measurements for Passive Environmental Remote Sensing , publisher =. 2023 , isbn =. doi:https://doi.org/10.1016/B978-0-12-823953-7.00010-1 , url =

    work page doi:10.1016/b978-0-12-823953-7.00010-1 2023

  23. [23]

    2023 , howpublished =

    ERA5 hourly data on single levels from 1940 to present , author =. 2023 , howpublished =

    1940

  24. [24]

    2023 , month =

    Sommer, Michael and von Rohden, Christoph and Simeonov, Tzvetan and Oelsner, Peter and Naebert, Tatjana and Romanens, Gonzague and Jauhiainen, Hannu and Survo, Petteri and Dirksen, Ruud , title =. 2023 , month =

    2023

  25. [25]

    GCOS Reference Upper-Air Network (GRUAN) , year =

  26. [26]

    Joint Committee for Guides in Metrology , title =

  27. [27]

    2008 , url =

    Evaluation of measurement data --- Supplement 1 to the ``Guide to the expression of uncertainty in measurement'' (GUM:1995): Propagation of distributions using a Monte Carlo method , institution =. 2008 , url =

    1995

  28. [28]

    and Koopman, S

    Durbin, J. and Koopman, S. J. , title =. Biometrika , volume =. 2002 , month =. doi:10.1093/biomet/89.3.603 , url =

    work page doi:10.1093/biomet/89.3.603 2002

  29. [29]

    and Stoffer, David S

    Shumway, Robert H. and Stoffer, David S. , biburl =

  30. [30]

    2012 , month =

    Durbin, James and Koopman, Siem Jan , title =. 2012 , month =. doi:10.1093/acprof:oso/9780199641178.001.0001 , url =

    arXiv 2012

This paper was first reviewed by deepseek-v4-flash on August 2, 2026.