Pith. sign in

REVIEW 4 major objections 7 minor 34 references

Deep Learning Surrogates for Real-Time Gas Emission Inversion

T0 review · 4 major / 7 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A neural-network surrogate for computational fluid dynamics, embedded in a sequential Monte Carlo filter, locates methane sources from sparse sensor measurements in unsteady wind fields at a fraction of the solver's cost.

desk verdict A sensible surrogate-in-the-loop Bayesian inversion with a real data win over a plume baseline, but the real-time claim outruns the reported runtimes and the validation is narrower than the abstract implies. read the letter →

arxiv 2506.14597 v1 pith:QIHNVO2C submitted 2025-06-17 cs.LG stat.APstat.ML

Thomas Newman , Christopher Nemeth , Matthew Jones , Philip Jonathan This is my paper · ORCID
classification cs.LGstat.APstat.ML
keywords gasemissioninversionsourcelocalizationBayesianstate-spacemodelparticlefilterCFDsurrogatemultilayerperceptronmethanemonitoringspatio-temporalinference
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

The paper proposes a way to find and quantify gas leaks in near real time from sparse, high-frequency concentration measurements. It replaces the expensive computational-fluid-dynamics solver inside a Bayesian particle filter with a fast neural-network surrogate trained on CFD output, so thousands of candidate source locations can be scored in milliseconds. On the Chilbolton controlled methane-release dataset, the surrogate-based filter locates the source to within about 5.8 metres, compared to about 11.1 metres for a Gaussian plume filter, while cutting total inversion time roughly in half. The method also tracks time-varying emission rates and works in simulated obstructed flows, which suggests it could support continuous industrial emissions monitoring.

What carries the argument

The load-bearing object is the trained MLP surrogate $C_{\mathrm{MLP}}(\tilde{x}, \tilde{y})$, which maps candidate source coordinates to the 20-second-averaged concentration expected at each sensor and is trained by regression on CFD runs of the time-dependent Navier-Stokes and advection-diffusion equations. It sits inside the SIR particle filter's likelihood evaluation, where the measurement model is evaluated once per particle per time step. Sliding time windows keep the surrogate tied to the current wind regime, and the random-walk state equation maintains particle diversity. The surrogate is what turns a computationally prohibitive forward model into a millisecond evaluation.

What would settle it

Run the trained surrogate and the CFD solver against the full Chilbolton sensor record for a held-out release period: if the surrogate matches the CFD but the CFD misses the measured concentration time series by large margins when local wind deviates from the uniform assumption, the surrogate-based posterior will be biased even though it emulates its teacher perfectly. A direct check is to compute the particle-filter localization error on a known ground-truth release under non-uniform wind and see whether the 5.82 m error degrades beyond the Gaussian plume baseline.

Watch

Extended reading notes

Core claim

The central claim is that a multilayer perceptron trained on outputs of a two-dimensional CFD solver can stand in for the physics within a sequential importance resampling particle filter, making Bayesian inference of methane source location and emission rate feasible in unsteady wind fields. The observation model $\hat{d}_t = C(\dot{x},\dot{y},\dot{z}|\tilde{x},\tilde{y},\tilde{z})\, s_{\kappa:t} + \beta_{\kappa:t} + \epsilon_t$ is retained, but $C$ is evaluated by the MLP instead of the numerical solver; separate MLPs are trained on sliding time windows so each sees roughly homogeneous wind. Validation on the Chilbolton releases shows the surrogate's concentration predictions closely match the CFD solver (MAPE 13.38% versus 13.28% on Source 1) and its inverted source location is more accurate than the Gaussian plume baseline (5.82 m versus 11.09 m), at less than half the total compute. The paper also demonstrates on synthetic obstructed scenarios that the filter recovers hidden sources and follows fluctuating emission rates.

Load-bearing premise

The whole chain inherits its physics from a two-dimensional CFD model driven by a spatially uniform, time-varying wind, and the paper does not compare that model's concentration predictions against independently measured plume fields, so the surrogate can only be as faithful as the simulator it learns from.

Editorial extensions

If this is right

  • Continuous monitoring becomes practical: because each likelihood evaluation costs milliseconds, the filter can ingest high-frequency sensor streams and update source estimates minute by minute.
  • The CFD cost is paid once offline to create training data, so replacing the plume model as the inner loop more than halves total inversion time in the reported experiments.
  • The particle posterior provides a full distribution over source location and emission rate, not just a point estimate, so uncertainty can be propagated into downstream decisions.
  • Synthetic obstructed-flow experiments show the filter localizes sources occluded by obstacles and tracks increasing, decreasing, and fluctuating emission rates, extending the approach beyond flat open terrain.

Reading between the lines

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

  • The same pattern of training a static surrogate on an expensive simulator and embedding it in a sequential Monte Carlo filter should transfer to other sparse-sensor inverse problems, such as groundwater contaminant source identification or indoor pollutant tracing.
  • Because the surrogate is retrained for each wind window, the framework implicitly treats the wind field as quasi-static; conditioning a single surrogate on wind speed and direction as additional inputs could remove the retraining overhead and allow true online drift.
  • The reported accuracy gain over Gaussian plume models suggests that transient physics captured by CFD matters even at a flat, open site; if that holds, plume-model-based inversions elsewhere may be underestimating localization uncertainty.
  • The delayed posterior adjustment after Source 2's sharp emission-rate drop means the raw filter is better suited to gradual leakage than sudden events; a regime-switching layer would be needed for detecting abrupt leak onset or shutdown.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 7 minor

Summary. The paper proposes a Bayesian source-inversion framework for gas emissions in which a multilayer perceptron (MLP) is trained to emulate a CFD-based concentration model, and the resulting surrogate is embedded in a sequential importance resampling (SIR) particle filter. The method is applied to the Chilbolton controlled methane release dataset, where the authors report surrogate predictions with MAPE close to the full numerical solver and lower than a Gaussian plume model, and a source-localization error of 5.82 m versus 11.09 m for the plume baseline. The paper also includes a synthetic obstructed-flow study with time-varying emission rates. The central claims are that the surrogate provides accuracy comparable to full CFD at a fraction of the computational cost and enables real-time, near-real-time inversion.

Significance. The core idea—replacing expensive CFD evaluations inside a Monte Carlo inversion with a trained neural emulator—is timely and sensible, and the paper demonstrates a working instance on a real controlled-release dataset with clearly reported numbers. The state-space formulation is standard, and the comparison against an independent Gaussian plume baseline is a useful design choice. However, the headline claims outrun the evidence: the reported end-to-end runtime of 83.3 minutes for 10 minutes of data contradicts the 'real-time' framing, the surrogate is trained on CFD simulations driven by the same wind conditions used for evaluation, only two 'ideal wind' windows are tested, and no full CFD-based inversion is run. The paper's value will be substantially higher if these issues are addressed by reframing the claims and adding temporal held-out validation.

major comments (4)
  1. [Section 4.2, Table 2] The reported end-to-end runtime of 83.3 minutes to process 10 minutes of Chilbolton data, including roughly 8 minutes of CFD data generation and training per MLP window, is incompatible with the abstract's 'real-time' and 'near-real-time' claims and with Section 4.2's statement that the framework enables 'real-time spatio-temporal inference'. The comparison to the Gaussian plume baseline is only a factor-of-two saving (83.3 vs 173.6 minutes), not the 'orders-of-magnitude faster runtimes' asserted in the abstract and conclusion.
  2. [Sections 3.1 and 4.1] The MLP is trained on CFD simulations that use the same wind boundary time series as the two 'ideal wind' evaluation windows, and the CFD itself is a 2D approximation with a spatially uniform, temporally varying wind. As a result, the close agreement between MLP and CFD in Table 1 is expected by construction, and the paper does not validate the CFD against measured concentration fields on held-out time periods or non-ideal conditions. The claim of 'comparable accuracy to full CFD' therefore rests on an unvalidated simulator and on interpolation over source locations only, not over flow conditions.
  3. [Section 4.2] No full CFD-based SIR inversion is ever run, so the abstract's 'comparable accuracy to full CFD solvers' for source localization is inferred only from surrogate-versus-CFD prediction errors, not from a direct inversion comparison. If the CFD is biased, the 5.82 m localization result may reflect simulator error rather than a genuine advantage over the Gaussian plume baseline; a full CFD inversion on at least one window, or a sensitivity analysis with perturbed CFD inputs, is needed to support the claim.
  4. [Section 5] The synthetic obstructed-flow results are presented only as posterior density plots and qualitative statements, with no quantitative metrics, no error bars, and no comparison to a baseline such as the Gaussian plume model or a full CFD inversion. Since the abstract claims 'robustness in complex environments', the main text should report numerical localization errors, emission-rate tracking errors, and runtimes for each of the three scenarios.
minor comments (7)
  1. [Section 2, Eq. (1)] The notation s_k:t and beta_k:t is introduced only verbally as 'history'; please define the window length k and the exact dependence of the observation model on it, since the particle filter state described in the text contains only the current emission rate and location.
  2. [Section 3.1] The phrase 'ground truth' is used for the CFD output; this overstates the status of a numerical simulator. Suggest using 'simulator reference' or 'high-fidelity simulation output' throughout.
  3. [Table 1] MAPE is not defined in the table or text; please state the formula and explain how zero or near-zero concentration readings are handled, since these can dominate the percentage error.
  4. [Section 4.2] The 'mean distance from all particles at the SIR last iteration' is not a standard posterior summary; please also report the posterior mean location, a credible region, and the standard deviation across independent filter runs.
  5. [Sections 4.1 and 5] The training-set sizes are reported as 484 simulations in Section 4.1 and 499 CFD-based training simulations in Section 5; please explain the difference or correct the inconsistency.
  6. [Figures 1 and 3] The figures would be easier to interpret with color bars, axis labels with units, and quantitative annotations; the claim in Figure 1 that one posterior is 'closer to the true source location' should be supported by the numerical metric in Table 2.
  7. [Supplementary Materials] Several load-bearing details, including CFD solver settings, supplementary results on long CFD runs, and the vertical Gaussian plume correction, are deferred to a supplement that was not available for review; please ensure the supplement is included or move essential details into the main text.

Circularity Check

1 steps flagged · score 2.0 of 10

Surrogate-vs-CFD closeness is expected from training; the real-data comparison gives independent content, so only minor circularity.

  1. self definitional [Section 4.1 (Table 1 discussion); cf. Section 3.2 loss definition]
    "The MLP surrogate achieved accuracy close to the numerical solver, outperforming the plume model, while requiring only milliseconds per prediction – faster than even the plume model."

    The MLP is trained by minimizing the MSE between C_MLP and the CFD solver output C_ns, as stated in Section 3.2: 'We define a loss function L as the Mean Squared Error (MSE) between the MLP’s prediction CMLP and the ground-truth solver output Cns over all training samples.' Therefore, reporting that the surrogate is 'close to the numerical solver' measures the quality of the emulator fit on CFD-simulated cases, including held-out source locations, rather than providing independent evidence of physical accuracy. The abstract's 'comparable accuracy to full CFD solvers' is thus guaranteed up to network capacity and training convergence, not established by an independent test.

full rationale

The core methodology is not circular: the SIR particle filter is a standard algorithm, the MLP is trained on CFD simulations, and the Chilbolton source-localization result is evaluated against real sensor data with an independent Gaussian plume baseline. The one notable circular element is that the surrogate's closeness to the CFD solver is expected by construction, since the CFD output is the training target; this makes 'comparable accuracy to full CFD' a fit-quality statement rather than an independent validation. The comparison to the Gaussian plume model uses the authors' own prior work [22] as a baseline, but that is a competitor implementation, not a load-bearing justification of the proposed method. The deeper scientific risk—that the 2D uniform-wind CFD is itself unvalidated against measured concentration fields—is a correctness/robustness concern rather than a circularity, so it does not raise the score beyond the minor self-definitional issue.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The method's central results depend on the CFD surrogate being a good emulator of the chosen CFD solver, and on the CFD solver itself being an adequate representation of the real dispersion. The listed hyperparameters affect the reported accuracy and runtimes.

free parameters (6)
  • Sliding time-window length = 4 minutes (Chilbolton), 3 minutes (synthetic)
    Chosen to match gas transport time across the site; controls how many MLPs are trained and which data window is used for each likelihood evaluation.
  • Number of SIR particles and iterations = N=1000, 700 iterations (Chilbolton); N=1000, 100 iterations per minute (synthetic)
    Standard particle filter tuning parameters; the reported posterior and localization error depend on them.
  • Number of CFD training simulations per MLP = 484 (Chilbolton), 499 per window (synthetic)
    The size of the training set for each surrogate; affects surrogate accuracy and total compute time.
  • MLP architecture = 4 hidden layers, 100 neurons (Chilbolton) / 500 neurons (synthetic), SeLU activations
    Chosen by hand; the paper reports no architecture search, and accuracy depends on this choice.
  • Background concentration and noise variance = Not specified in text
    The observation model (1) includes background beta and Gaussian noise sigma; neither is estimated or reported, yet both enter every likelihood evaluation.
  • Process noise covariance W = Not specified
    The random walk in Eq. (2) requires a covariance W; its value controls particle diversity and tracking speed for time-varying emission rates.
assumptions (5)
  • domain assumption Concentration is linear in emission rate: the source's influence on the flow field is negligible, so C can be computed for unit emission and scaled by s_kappa:t (Eq. 1).
    Invoked in Section 2 to justify the product C x s_kappa:t; fails for strong plumes with buoyancy or momentum effects.
  • domain assumption Two-dimensional flow approximation: all sources and sensors lie in the same horizontal plane, allowing 2D CFD.
    Stated in Section 3.1 and acknowledged as a limitation in the Conclusion.
  • domain assumption Spatially uniform but temporally varying wind field across the site.
    Stated in Section 3.1; ignores wind shear and obstacle-induced spatial variability in the real-data validation.
  • domain assumption The CFD solver (PhiFlow) produces accurate ground-truth concentrations for the Chilbolton site.
    The entire training dataset and the 13-14% MAPE benchmark come from this simulator (Section 4.1); no independent validation of the CFD is provided.
  • domain assumption Gaussian, independent measurement noise with zero mean.
    Assumed in Section 2 for the likelihood p(d_t|theta_t); sensor noise may be correlated or heavy-tailed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Deep Learning Surrogates for Real-Time Gas Emission Inversion." pith.science (2026). https://pith.science/paper/QIHNVO2C

@misc{pith2026250614597,
  author       = {Pith},
  title        = {Pith review of: Deep Learning Surrogates for Real-Time Gas Emission Inversion},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QIHNVO2C}},
  note         = {Machine review of arXiv:2506.14597}
}
read the original abstract

Real-time identification and quantification of greenhouse-gas emissions under transient atmospheric conditions is a critical challenge in environmental monitoring. We introduce a spatio-temporal inversion framework that embeds a deep-learning surrogate of computational fluid dynamics (CFD) within a sequential Monte Carlo algorithm to perform Bayesian inference of both emission rate and source location in dynamic flow fields. By substituting costly numerical solvers with a multilayer perceptron trained on high-fidelity CFD outputs, our surrogate captures spatial heterogeneity and temporal evolution of gas dispersion, while delivering near-real-time predictions. Validation on the Chilbolton methane release dataset demonstrates comparable accuracy to full CFD solvers and Gaussian plume models, yet achieves orders-of-magnitude faster runtimes. Further experiments under simulated obstructed-flow scenarios confirm robustness in complex environments. This work reconciles physical fidelity with computational feasibility, offering a scalable solution for industrial emissions monitoring and other time-sensitive spatio-temporal inversion tasks in environmental and scientific modeling.

Figures

Figures reproduced from arXiv: 2506.14597 by the authors.

Figure 1
Figure 1. Gaussian plume model and surrogate model-based SIR particles’ posterior density of Source [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Case study results for surrogate-based SIR filter source location estimation in unsteady and [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Case study results for temporally varying emission rate estimation. The surrogate-based [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

34 extracted references · 31 canonical work pages

  1. [1]

    State space modeling of multiple time series.Econometric Reviews, 10(1):1–59, 1991

    Masanao Aoki and Arthur Havenner. State space modeling of multiple time series.Econometric Reviews, 10(1):1–59, 1991

    work page 1991

  2. [2]

    A tutorial on particle filters for online nonlinear/non-Gaussian Bayesian tracking.IEEE Transactions on Signal Processing, 50(2):174–188, 2002

    M Sanjeev Arulampalam, Simon Maskell, Neil Gordon, and Tim Clapp. A tutorial on particle filters for online nonlinear/non-Gaussian Bayesian tracking.IEEE Transactions on Signal Processing, 50(2):174–188, 2002. 9

    work page 2002

  3. [3]

    The interacting multiple model algorithm for systems with Markovian switching coefficients.IEEE Transactions on Automatic Control, 33(8):780– 783, 2002

    Henk AP Blom and Yaakov Bar-Shalom. The interacting multiple model algorithm for systems with Markovian switching coefficients.IEEE Transactions on Automatic Control, 33(8):780– 783, 2002

    work page 2002

  4. [4]

    CRC press, 2011

    Steve Brooks, Andrew Gelman, Galin Jones, and Xiao-Li Meng.Handbook of Markov Chain Monte Carlo. CRC press, 2011

    work page 2011

  5. [5]

    Stochastic problems in physics and astronomy.Reviews of Modern Physics, 15(1):1, 1943

    Subrahmanyan Chandrasekhar. Stochastic problems in physics and astronomy.Reviews of Modern Physics, 15(1):1, 1943

    work page 1943

  6. [6]

    Comparison of resampling schemes for particle filtering

    Randal Douc and Olivier Cappé. Comparison of resampling schemes for particle filtering. In ISPA 2005. Proceedings of the 4th International Symposium on Image and Signal Processing and Analysis, 2005., pages 64–69. Ieee, 2005

    work page 2005

  7. [7]

    Scalable Monte Carlo for Bayesian learning.arXiv preprint arXiv:2407.12751, 2024

    Paul Fearnhead, Christopher Nemeth, Chris J Oates, and Chris Sherlock. Scalable Monte Carlo for Bayesian learning.arXiv preprint arXiv:2407.12751, 2024

    arXiv 2024

  8. [8]

    Understanding the difficulty of training deep feedforward neural networks

    Xavier Glorot and Yoshua Bengio. Understanding the difficulty of training deep feedforward neural networks. InProceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, pages 249–256. JMLR Workshop and Conference Proceedings, 2010

    2010

Show all 34 references
  1. [9]

    Novel approach to nonlinear/non- Gaussian Bayesian state estimation

    Neil J Gordon, David J Salmond, and Adrian FM Smith. Novel approach to nonlinear/non- Gaussian Bayesian state estimation. InIEE Proceedings F (Radar and Signal Processing), volume 140, pages 107–113. IET, 1993

    1993

  2. [10]

    State-space models.Handbook of Econometrics, 4:3039–3080, 1994

    James D Hamilton. State-space models.Handbook of Econometrics, 4:3039–3080, 1994

    1994

  3. [11]

    Methane emissions: remote mapping and source quantification using an open-path laser dispersion spectrometer.Geophysical Research Letters, 47(10), 2020

    Bill Hirst, David Randell, Matthew Jones, Johnny Chu, Arun Kannath, Neil Macleod, Marcella Dean, and Damien Weidmann. Methane emissions: remote mapping and source quantification using an open-path laser dispersion spectrometer.Geophysical Research Letters, 47(10), 2020

    2020

  4. [12]

    Mapping CO2 and CH4 emissions: field-trial evaluation of LightSource for remotely estimating the locations and mass emission rates of sources

    Bill Hirst, David Randell, Matthew Jones, Damien Weidman, and Marcella Dean. Mapping CO2 and CH4 emissions: field-trial evaluation of LightSource for remotely estimating the locations and mass emission rates of sources. InFifth CO2 Geological Storage Workshop, volume 2018, pag...

    2018

  5. [13]

    ΦFlow (PhiFlow): differentiable simulations for PyTorch, TensorFlow and Jax

    Philipp Holl and Nils Thuerey. ΦFlow (PhiFlow): differentiable simulations for PyTorch, TensorFlow and Jax. InInternational Conference on Machine Learning. PMLR, 2024

    2024

  6. [14]

    Parameterizing state–space models for infectious disease dynamics by generalized profiling: measles in ontario

    Giles Hooker, Stephen P Ellner, Laura De Vargas Roditi, and David JD Earn. Parameterizing state–space models for infectious disease dynamics by generalized profiling: measles in ontario. Journal of The Royal Society Interface, 8(60):961–974, 2011

    2011

  7. [15]

    Academic Press, 1970

    Andrew H Jazwinski.Stochastic Processes and Filtering Theory. Academic Press, 1970

    1970

  8. [16]

    A new approach to linear filtering and prediction problems.Journal of Basic Engineering, 1960

    Rudolph E Kalman. A new approach to linear filtering and prediction problems.Journal of Basic Engineering, 1960

    1960

  9. [17]

    Monte Carlo filter and smoother for non-Gaussian nonlinear state space models.Journal of Computational and Graphical Statistics, 5(1):1–25, 1996

    Genshiro Kitagawa. Monte Carlo filter and smoother for non-Gaussian nonlinear state space models.Journal of Computational and Graphical Statistics, 5(1):1–25, 1996

    1996

  10. [18]

    Self-normalizing neural networks.Advances in Neural Information Processing Systems, 30, 2017

    Günter Klambauer, Thomas Unterthiner, Andreas Mayr, and Sepp Hochreiter. Self-normalizing neural networks.Advances in Neural Information Processing Systems, 30, 2017

    2017

  11. [19]

    Sur les lois des mouvement des fluides, en ayant égard à l’adhesion des molecules

    Claude-Louis Navier. Sur les lois des mouvement des fluides, en ayant égard à l’adhesion des molecules. InAnnales de Chimie et de Physique, volume 19, page 1821. Lavoisier Paris, France, 1821

  12. [20]

    Sur les lois du mouvement des fluides

    Claude-Louis Navier. Sur les lois du mouvement des fluides. InMemoires de l’Academie Royale des Sciences, volume 6, pages 389–440, 1827

  13. [21]

    Particle learning methods for state and parameter estimation

    Christopher Nemeth, Paul Fearnhead, Lyudmila Mihaylova, and Dave V orley. Particle learning methods for state and parameter estimation. In9th IET Data Fusion & Target Tracking Conference (DF&TT 2012): Algorithms & Applications, pages 12–1. IET, 2012. 10

    2012

  14. [22]

    Probabilistic inversion modeling of gas emissions: a gradient-based MCMC estimation of Gaussian plume parameters.arXiv preprint arXiv:2408.01298, 2024

    Thomas Newman, Christopher Nemeth, Matthew Jones, and Philip Jonathan. Probabilistic inversion modeling of gas emissions: a gradient-based MCMC estimation of Gaussian plume parameters.arXiv preprint arXiv:2408.01298, 2024

    2024 arXiv

  15. [23]

    The Oil & Gas Methane Partnership 2.0

    OMGP2.0. The Oil & Gas Methane Partnership 2.0. https://www.ogmpartnership.org/, 2024

    2024

  16. [24]

    Christopher D. Prashad. State-space modelling for infectious disease surveillance data: dynamic regression and covariance analysis.Infectious Disease Modelling, 10(2):591–627, 2025

    2025

  17. [25]

    Springer, 1999

    Christian P Robert, George Casella, and George Casella.Monte Carlo Statistical Methods, volume 2. Springer, 1999

    1999

  18. [26]

    The perceptron: a probabilistic model for information storage and organiza- tion in the brain.Psychological Review, 65(6):386, 1958

    Frank Rosenblatt. The perceptron: a probabilistic model for information storage and organiza- tion in the brain.Psychological Review, 65(6):386, 1958

    1958

  19. [27]

    Global methane budget 2000–2020.Earth System Science Data, 17(5):1873–1958, 2025

    Marielle Saunois, Adrien Martinez, Benjamin Poulter, Zhen Zhang, Peter A Raymond, Pierre Regnier, Josep G Canadell, Robert B Jackson, Prabir K Patra, Philippe Bousquet, et al. Global methane budget 2000–2020.Earth System Science Data, 17(5):1873–1958, 2025

    2000

  20. [28]

    Über Brownsche Molekularbewegung unter Einwirkung äußerer Kräfte und deren Zusammenhang mit der verallgemeinerten Diffusionsgleichung.Annalen der Physik, 353(24):1103–1112, 1916

    Marian V Smoluchowski. Über Brownsche Molekularbewegung unter Einwirkung äußerer Kräfte und deren Zusammenhang mit der verallgemeinerten Diffusionsgleichung.Annalen der Physik, 353(24):1103–1112, 1916

    1916

  21. [29]

    Cambridge University Press, 2014

    Thomas Stocker.Climate change 2013: the physical science basis: working group I contribution to the fifth assessment report of the Intergovernmental Panel on Climate Change. Cambridge University Press, 2014

    2013

  22. [30]

    The mathematics of atmospheric dispersion modeling.Siam Review, 53(2):349– 372, 2011

    John M Stockie. The mathematics of atmospheric dispersion modeling.Siam Review, 53(2):349– 372, 2011

    2011

  23. [31]

    Artech House, 2013

    Lawrence D Stone, Roy L Streit, Thomas L Corwin, and Kristine L Bell.Bayesian Multiple Target Tracking. Artech House, 2013

    2013

  24. [32]

    Antti V oss, Elisa Vänskä, Damien Weidmann, Aki Pulkkinen, and Aku Seppänen. Multi-open- path laser dispersion spectroscopy combined with Bayesian state estimation for localizing and quantifying methane emissions.Atmospheric Environment: X, page 100260, 2024

    2024

  25. [33]

    A survey of transfer learning.Journal of Big Data, 3:1–40, 2016

    Karl Weiss, Taghi M Khoshgoftaar, and DingDing Wang. A survey of transfer learning.Journal of Big Data, 3:1–40, 2016

    2016

  26. [34]

    Springer Science & Business Media, 2006

    Mike West and Jeff Harrison.Bayesian Forecasting and Dynamic Models. Springer Science & Business Media, 2006. 11

    2006

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.