arxiv: 2605.03454 · v1 · submitted 2026-05-05 · ⚛️ physics.geo-ph

Recognition: 3 theorem links

· Lean Theorem

Estimating noise for airborne electromagnetic data from repeat flight lines or inversion residuals

Tim Scarr , Anandaroop Ray , Ross C. Brodie

Authors on Pith no claims yet

Pith reviewed 2026-05-08 18:26 UTC · model grok-4.3

classification ⚛️ physics.geo-ph

keywords airborne electromagneticAEMnoise estimationrepeat flight linesdata covarianceinversionGaussian likelihoodmultiplicative noise

0 comments

The pith

Repeat flight lines over the same ground separate multiplicative and additive noise in AEM data and support a diagonal covariance matrix for regularised inversions.

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

The paper develops practical methods to estimate noise levels in airborne electromagnetic surveys by using repeat flight lines flown over identical terrain. It separates the observed variability into multiplicative and additive noise components and applies a non-linear correction to account for small altitude differences between passes. The approach further transforms the noise statistics into a Gaussian form suitable for inversion likelihood functions and calculates the full data covariance, including correlations between time channels. The central result is that these steps often justify simplifying the covariance to a diagonal matrix when performing regularised time-domain AEM imaging.

Core claim

Repeat flight lines provide a way for geophysicists to calculate the statistical variability in AEM data acquired over the same ground and therefore estimate the levels of noise to propagate into the inversion. The total noise can be separated into multiplicative and additive components. The multiplicative noise is derived by repeat lines at survey altitude with a non-linear altitude correction. The paper details a methodology to Gaussianise the data noise and provide a statistically valid Gaussian data misfit or likelihood function, along with methods for estimating off-diagonal elements in the data covariance matrix that account for time-channel correlation. For regularised time-domain AEM

What carries the argument

Non-linear altitude correction applied to repeat-line data to isolate multiplicative noise, combined with Gaussianisation of the resulting distribution to produce a valid likelihood and covariance for the inversion misfit function.

If this is right

Estimated noise levels from repeat lines can be used directly to set stopping criteria in deterministic inversions or to define model likelihoods in probabilistic ones.
Separation of multiplicative and additive noise components improves the accuracy of propagating system noise into subsurface conductivity models.
Gaussianisation allows standard least-squares or Bayesian inversion algorithms to be applied to AEM data with a statistically justified misfit function.
The finding that a diagonal covariance matrix suffices reduces the computational cost of large-scale AEM inversions while retaining statistical rigor.

Where Pith is reading between the lines

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

The same repeat-line protocol could be adapted to other airborne or marine geophysical methods where multiple passes over identical ground are feasible.
In operational surveys, adopting a diagonal covariance would simplify software implementation and speed up inversion of very large datasets without major loss of accuracy.
If geological heterogeneity contributes to differences between repeat passes, the resulting noise estimate becomes an upper bound, leading to more conservative but still valid inversion results.

Load-bearing premise

Repeat flight lines over the same ground capture only system noise without significant contributions from unaccounted geological or operational variations, and the Gaussianisation step produces a statistically valid likelihood for the specific data distribution encountered.

What would settle it

Perform independent repeat-line noise estimates on new survey data, run regularised inversions, and test whether the observed data residuals are statistically consistent with the estimated diagonal covariance or show significant unaccounted time-channel correlations.

Figures

Figures reproduced from arXiv: 2605.03454 by Anandaroop Ray, Ross C. Brodie, Tim Scarr.

**Figure 1.** Figure 1: Repeat line section over Menindee Lakes, New South Wales, Australia. The green line shows view at source ↗

**Figure 2.** Figure 2: Plot of receiver heights and average of AEM system across all repeat lines flown over Lake view at source ↗

**Figure 3.** Figure 3: Plot of corrections ∆j for the differing receiver heights across all time channels (window #). Note how when the actual, acquired receiver height is above the desired height (given by the mean from all flights), we need to add to the amplitude of the signal (z is positive downwards into the earth). This is because the secondary field amplitude at higher (than desired) height is lower, and requires positive… view at source ↗

**Figure 4.** Figure 4: Comparison cross-plot of standard deviation vs mean of Z-component AEM data. Green view at source ↗

**Figure 5.** Figure 5: Plot of standard deviation divided by mean amplitude of the HC data, plotted against view at source ↗

**Figure 6.** Figure 6: Correlation matrix from height corrected repeat line data (a,b) or inversion residuals (c,d), view at source ↗

**Figure 7.** Figure 7: Cross-sections of inverted conductivity with depth of the same line across Menindee Lake view at source ↗

**Figure 8.** Figure 8: Pixel-wise standard deviation of inverted conductivity from all inversion tests in Figure 7. view at source ↗

**Figure 9.** Figure 9: Residuals from all approaches (a) in Table 1, and profiles of conductivity (b) from each view at source ↗

**Figure 10.** Figure 10: An analysis of standardised inversion residuals arranged from top-bottom according to view at source ↗

**Figure 11.** Figure 11: A line search over c in equation (18) for a synthetic set of 5,000 samples with variance ( view at source ↗

**Figure 12.** Figure 12: If we have a set of residuals x, akin to our height corrected AEM residuals divided by mean amplitude shown in view at source ↗

read the original abstract

Characterising the noise of an airborne electromagnetic (AEM) system is critical in correctly imaging the earth's subsurface conductivity. Deterministic and probabilistic geophysical inversion algorithms require foreknowledge of the system noise to specify stopping criteria or a valid model likelihood. Repeat flight lines provide a way for geophysicists to calculate the statistical variability in AEM data acquired over the same ground, and therefore estimate the levels of noise to propagate into the inversion. The total noise can be separated into multiplicative and additive components. The multiplicative noise is derived by repeat lines at survey altitude. The method to calculate the multiplicative noise is scarcely documented and usual methods for height correcting acquired data require a linear trend removal. This study will outline the algorithm used to estimate multiplicative noise of an AEM system, and non-linearly correct for varying altitudes during repeat flights. Additionally, this paper details a methodology to Gaussianise the data noise and provide a statistically valid Gaussian data misfit or likelihood function. Significantly, we provide methods for estimating the off-diagonal elements in the data covariance matrix used within the misfit function, taking into account the time-channel data correlation that is usually neglected. While our methodology is general, our study of a rotary-wing system leads us to conclude that for regularised time-domain AEM imaging, a diagonal data covariance suffices -- an important implication for rigorous yet practical AEM inversion.

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.

Desk Editor's Note private letter to a colleague

This paper gives a clear, implementable workflow for pulling multiplicative and additive noise estimates from AEM repeat lines, with a non-linear altitude correction and an empirical check showing diagonal covariance is often enough for their rotary-wing data.

read the letter

The paper's main contribution is a practical recipe for noise characterisation in airborne EM surveys using repeat flight lines. It separates multiplicative noise (estimated at survey altitude) from additive noise, applies a non-linear correction for altitude differences instead of the usual linear detrend, Gaussianises the residuals to support a valid likelihood, and computes the full data covariance including time-channel correlations before testing the diagonal approximation. The explicit algorithm for the height correction and the step-by-step covariance estimation are the parts that stand out as new relative to the cited literature. The finding that off-diagonal terms do not materially improve regularised time-domain inversions on this dataset is useful because it supports simpler, faster inversions without sacrificing much rigour in their case. The methods are described in enough detail that a reader could code them up and apply them to similar data. The central assumption—that repeat lines mainly capture system noise once altitude is accounted for—is addressed by the workflow rather than left hanging, and the results are tied directly to one rotary-wing system. That limits how far the diagonal-covariance conclusion can be generalised, but the paper does not overclaim it as universal. The title mentions inversion residuals as an alternative route, yet the body focuses on repeats, which is reasonable but leaves a direct head-to-head for follow-up work. This is aimed at geophysicists who run deterministic or probabilistic AEM inversions and need defensible noise models. Anyone working on resource or environmental applications would get concrete tools they can adapt. It deserves peer review because the documented steps and the empirical test fill a gap in standard practice even if the advance is incremental rather than transformative.

Referee Report

2 major / 2 minor

Summary. The manuscript presents methods to estimate multiplicative noise in AEM data from repeat flight lines at survey altitude, including a non-linear altitude correction to account for height variations. It separates total noise into multiplicative and additive components, details a Gaussianisation procedure to produce a statistically valid Gaussian likelihood for the data misfit, and provides an approach to estimate the full data covariance matrix including off-diagonal time-channel correlations. Based on analysis of data from one rotary-wing AEM system, the authors conclude that a diagonal covariance matrix suffices for regularised time-domain AEM imaging.

Significance. If the empirical result holds, the work has practical significance for AEM inversion by enabling simpler yet statistically grounded data misfit functions that neglect time-channel correlations without material loss of accuracy. The manuscript earns credit for supplying the explicit algorithm for multiplicative noise estimation, the Gaussianisation transform, and the direct test of the diagonal approximation on the reported dataset, all of which support reproducibility and falsifiability of the central claim.

major comments (2)

[Abstract and conclusion] Abstract and conclusion: the claim that a diagonal data covariance suffices for regularised time-domain AEM imaging rests on results from a single rotary-wing system. Because this is the load-bearing empirical finding, the manuscript should include a brief discussion of the conditions (e.g., system type, survey geometry, or geological variability) under which the off-diagonal terms remain negligible, to clarify the scope of the implication.
[Results] The weakest assumption—that repeat lines isolate system noise without significant unaccounted geological or operational contributions—is addressed by the workflow, but the results section should report a quantitative check (e.g., comparison of noise estimates against independent measures or across multiple repeat pairs) to confirm the Gaussianised residuals yield a valid likelihood for the encountered data distribution.

minor comments (2)

[Abstract] The abstract uses future tense ('this study will outline') in a completed manuscript; rephrase to present tense for consistency.
[Methods] Notation for the multiplicative and additive noise components should be defined explicitly at first use (e.g., in the methods section) to aid readers unfamiliar with AEM processing conventions.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive review and the recommendation for minor revision. We address each major comment point by point below, agreeing where revisions will strengthen the manuscript and providing the strongest honest defense of our approach.

read point-by-point responses

Referee: [Abstract and conclusion] Abstract and conclusion: the claim that a diagonal data covariance suffices for regularised time-domain AEM imaging rests on results from a single rotary-wing system. Because this is the load-bearing empirical finding, the manuscript should include a brief discussion of the conditions (e.g., system type, survey geometry, or geological variability) under which the off-diagonal terms remain negligible, to clarify the scope of the implication.

Authors: We agree that the empirical finding is based on a single rotary-wing system and that clarifying its scope strengthens the manuscript. We will revise the abstract and conclusion to include a concise discussion noting that the diagonal covariance approximation holds for the studied rotary-wing AEM system under the encountered survey geometry and geological conditions. We will further indicate that the result may extend to similar systems but caution against generalisation without additional validation, thereby delineating the implication without overstatement. revision: yes
Referee: [Results] The weakest assumption—that repeat lines isolate system noise without significant unaccounted geological or operational contributions—is addressed by the workflow, but the results section should report a quantitative check (e.g., comparison of noise estimates against independent measures or across multiple repeat pairs) to confirm the Gaussianised residuals yield a valid likelihood for the encountered data distribution.

Authors: The referee correctly notes the importance of validating the isolation of system noise. While the repeat-line workflow and Gaussianisation procedure are designed to minimise geological contributions, we will augment the results section with a quantitative check. This will comprise a comparison of multiplicative noise estimates across multiple repeat pairs and an evaluation of the Gaussianised residuals (including distributional diagnostics) to confirm that the transformed data support a valid Gaussian likelihood for the reported dataset. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper defines noise estimates directly from observed statistical variability across repeat flight lines, separating multiplicative and additive components via explicit algorithmic steps including non-linear altitude correction and Gaussianisation. These are data-driven computations rather than parameters fitted to a subset and then renamed as predictions. The conclusion that a diagonal covariance matrix suffices is an empirical observation drawn from the specific rotary-wing dataset and full covariance computation described in the workflow, not a reduction by construction or via load-bearing self-citation. No equations or steps equate outputs to inputs tautologically; the methods remain externally falsifiable against the repeat-line data.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The approach rests on the domain assumption that repeat-line variability isolates system noise and that noise can be usefully decomposed and transformed into a Gaussian form for inversion likelihoods.

axioms (2)

domain assumption Total noise separates into multiplicative and additive components
Stated directly in the abstract as the basis for the estimation procedure.
domain assumption Repeat flight lines provide a statistically valid sample of system noise
Central premise for deriving both multiplicative and additive noise estimates.

pith-pipeline@v0.9.0 · 5544 in / 1285 out tokens · 32746 ms · 2026-05-08T18:26:33.759295+00:00 · methodology

Review history (2 revisions) →

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost (J(x)=½(x+x⁻¹)−1) washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

σ_j = √((kd_j)² + a_j²) ... k is the multiplicative noise factor

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

34 extracted references · 32 canonical work pages

[1]

doi: 10.1111/j.1365-246X.2010.04530.x

ISSN 0956-540X. doi: 10.1111/j.1365-246X.2010.04530.x. URLhttps://doi.org/10.1111/j.1365-246X.2010.04530.x. Esben Auken, Anders Vest Christiansen, Casper Kirkegaard, Gianluca Fiandaca, Cyril Schamper, Ahmad Ali Behroozmand, Andrew Binley, Emil Nielsen, Flemming Effersø, Niels Bøie Christensen, Kurt Sørensen, Nikolaj Foged, and Giulio Vignoli. An overview ...

work page doi:10.1111/j.1365-246x.2010.04530.x 2010
[2]

URLhttps://doi.org/10

doi: 10.1071/EG13097. URLhttps://doi.org/10. 1071/EG13097. Jeff Bezanson, Alan Edelman, Stefan Karpinski, and Viral B. Shah. Julia: A fresh approach to numerical computing.SIAM Review, 59(1):65–98,

work page doi:10.1071/eg13097
[3]

Bezanson, A

doi: 10.1137/141000671. URLhttps: //doi.org/10.1137/141000671. Daniel Blatter, Kerry Key, Anandaroop Ray, Neil Foley, Slawek Tulaczyk, and Esben Auken. Trans- dimensional bayesian inversion of airborne transient em data from taylor glacier, antarctica.Geophys- ical Journal International, 214(3):1919–1936, 06

work page doi:10.1137/141000671 1919
[4]

doi: 10.1093/gji/ggy255

ISSN 0956-540X. doi: 10.1093/gji/ggy255. URLhttps://doi.org/10.1093/gji/ggy255. T. Bodin, M. Sambridge, K. Gallagher, and N. Rawlinson. Transdimensional inversion of receiver functions and surface wave dispersion.Journal of Geophysical Research: Solid Earth, 117(B2),

work page doi:10.1093/gji/ggy255
[5]

URLhttps://agupubs.onlinelibrary.wiley.com/ doi/abs/10.1029/2011JB008560

doi: https://doi.org/10.1029/2011JB008560. URLhttps://agupubs.onlinelibrary.wiley.com/ doi/abs/10.1029/2011JB008560. R. C. Brodie. Galeisbstdem: A deterministic algorithm for 1d sample by sample inversion of time- domain aem data – theoretical details,

work page doi:10.1029/2011jb008560
[6]

A. Davis. Estimating noise in aem data.AEM 2023: Short abstracts, 2023(225):52–66,

2023
[7]

URLhttps://doi.org/10.1080/14432471.2023.2236352

doi: 10.1080/14432471.2023.2236352. URLhttps://doi.org/10.1080/14432471.2023.2236352. Jan Dettmer and Stan E Dosso. Trans-dimensional matched-field geoacoustic inversion with hierarchi- cal error models and interacting Markov chains.The Journal of the Acoustical Society of America, 132(4):2239–2250, October

work page doi:10.1080/14432471.2023.2236352 2023
[8]

doi: 10.1121/1.4746016

ISSN 1520-8524. doi: 10.1121/1.4746016. Jan Dettmer, Sheri Molnar, Gavin Steininger, Stan E. Dosso, and John F. Cassidy. Trans- dimensional inversion of microtremor array dispersion data with hierarchical autoregressive error models.Geophysical Journal International, 188(2):719–734, February

work page doi:10.1121/1.4746016
[9]

doi: 10.1111/j.1365-246X.2011.05302.x

ISSN 0956540X. doi: 10.1111/j.1365-246X.2011.05302.x. Wilmut MJ Dosso, Nielsen PL. Data error covariance in matched-field geoacoustic inversion.J Acoust Soc Am, 119(1):208–19, Jan

work page doi:10.1111/j.1365-246x.2011.05302.x 2011
[10]

Romina A S Gehrmann, Amir Haroon, McKinley Morton, Axel T Djanni, and Timothy A Minshull

doi: 10.1121/1.2139625. Romina A S Gehrmann, Amir Haroon, McKinley Morton, Axel T Djanni, and Timothy A Minshull. Seafloor massive sulphide exploration using deep-towed controlled source electromagnetics: Naviga- tional uncertainties.Geophysical Journal International, 220(2):1215–1227, November

work page doi:10.1121/1.2139625
[11]

doi: 10.1093/gji/ggz513

ISSN 0956-540X. doi: 10.1093/gji/ggz513. Wences P. Gouveia and John A Scales. Bayesian seismic waveform inversion: Parameter estimation and uncertainty analysis.Journal of Geophysical Research: Solid Earth, 103(B2):2759–2779,

work page doi:10.1093/gji/ggz513
[12]

doi: 10.1029/97jb02933

ISSN 21699356. doi: 10.1029/97jb02933. Andrew Green and Richard Lane. Estimating noise levels in aem data.ASEG Extended Abstracts, 2003, 08

work page doi:10.1029/97jb02933 2003
[13]

doi: 10.1071/ASEG2003ab093. S. Kang, R. Knight, and M. Goebel. Improved imaging of the large-scale structure of a groundwater system with airborne electromagnetic data.Water Resources Research, 58:e2021WR031439,

work page doi:10.1071/aseg2003ab093
[14]

Kerry Key

doi: 10.1029/2021WR031439. Kerry Key. Mare2dem: a 2-d inversion code for controlled-source electromagnetic and magnetotelluric data.Geophysical Journal International, 207(1):571–588, 08

work page doi:10.1029/2021wr031439
[15]

doi: 10.1093/ gji/ggw290

ISSN 0956-540X. doi: 10.1093/ gji/ggw290. URLhttps://doi.org/10.1093/gji/ggw290. Reinhard Kirsch. Groundwater geophysics: A tool for hydrogeology.Groundwater Geophysics: A Tool for Hydrogeology, Edited by Reinhard Kirsch. 2006 XVII, 493 p, 300 illus. 3-540-29383-3. Berlin: Springer, 2006., 01

work page doi:10.1093/gji/ggw290 2006
[16]

Alan Yusen Ley-Cooper, Ross C

doi: 10.1111/1365-2478.70021. Alan Yusen Ley-Cooper, Ross C. Brodie, and Murray Richardson and. Ausaem: Australia’s airborne electromagnetic continental-scale acquisition program.Exploration Geophysics, 51(1):193–202,

work page doi:10.1111/1365-2478.70021
[17]

URLhttps://doi.org/10.1080/08123985.2019.1694393

doi: 10.1080/08123985.2019.1694393. URLhttps://doi.org/10.1080/08123985.2019.1694393. Alberto Malinverno. Parsimonious bayesian markov chain monte carlo inversion in a nonlinear geo- physical problem.Geophysical Journal International, 151(3):675–688, 12

work page doi:10.1080/08123985.2019.1694393 2019
[18]

doi: 10.1046/j.1365-246X.2002.01847.x

ISSN 0956-540X. doi: 10.1046/j.1365-246X.2002.01847.x. URLhttps://doi.org/10.1046/j.1365-246X.2002.01847.x. William Menke. Copyright. InGeophysical Data Analysis: Discrete Inverse Theory (Third Edition), page iv. Academic Press, Boston, third edition edition,

work page doi:10.1046/j.1365-246x.2002.01847.x 2002
[19]

doi: https:// doi.org/10.1016/B978-0-12-397160-9.00018-7

ISBN 978-0-12-397160-9. doi: https:// doi.org/10.1016/B978-0-12-397160-9.00018-7. URLhttps://www.sciencedirect.com/science/ article/pii/B9780123971609000187. Burke J. Minsley, Nathan Leon Foks, and Paul A. Bedrosian. Quantifying model structural uncertainty using airborne electromagnetic data.Geophysical Journal International, 224(1):590–607, 2021a. ISSN ...

work page doi:10.1016/b978-0-12-397160-9.00018-7
[20]

S Mule, Camilla Soerensen, and Tim Munday and

doi: 10.1029/94JB03097. S Mule, Camilla Soerensen, and Tim Munday and. Handling noise in aem inversion – implications for subsurface characterisation.ASEG Extended Abstracts, 2019(1):1–4,

work page doi:10.1029/94jb03097 2019
[21]

2019.12073183

doi: 10.1080/22020586. 2019.12073183. URLhttps://doi.org/10.1080/22020586.2019.12073183. James Murr, Roger Skirrow, Anthony Schofield, James Goodwin, R. Coghlan, Lindsay Highet, Michael Doublier, Jingjie Duan, and K. Czarnota. Tennant creek -mount isa iocg mineral potential assess- ment.Exploring for the Future: Extended Abstracts, 06

work page doi:10.1080/22020586 2019
[22]

Jorge Nocedal and Stephen Wright

doi: 10.26186/5d6c90ebb0ae9. Jorge Nocedal and Stephen Wright. Numerical optimization,

work page doi:10.26186/5d6c90ebb0ae9
[23]

doi: 10.1190/1.1826719. R. L. Parker.Geophysical Inverse Theory. Princeton series in Geophysics. Princeton University Press, Princeton,

work page doi:10.1190/1.1826719
[24]

doi: 10.1093/jge/gxad043

ISSN 1742-2132. doi: 10.1093/jge/gxad043. URL https://doi.org/10.1093/jge/gxad043. V.A. Profillidis and G.N. Botzoris. Chapter 5 - statistical methods for transport demand modeling. In V.A. Profillidis and G.N. Botzoris, editors,Modeling of Transport Demand, pages 163–224. Elsevier,

work page doi:10.1093/jge/gxad043
[25]

doi: https://doi.org/10.1016/B978-0-12-811513-8.00005-4

ISBN 978-0-12-811513-8. doi: https://doi.org/10.1016/B978-0-12-811513-8.00005-4. A. Ray, T. Bodin, and K. Key. Hierarchical bayesian inversion of marine csem data over the scarborough gas field - a lesson in correlated noise,

work page doi:10.1016/b978-0-12-811513-8.00005-4
[26]

Anandaroop Ray, Yusen Ley-Cooper, Ross C Brodie, Richard Taylor, Neil Symington, and Negin F Moghaddam

URLhttps://perso.ens-lyon.fr/thomas.bodin/ Publi/Ray_etal_SEG2013.pdf. Anandaroop Ray, Yusen Ley-Cooper, Ross C Brodie, Richard Taylor, Neil Symington, and Negin F Moghaddam. An information theoretic bayesian uncertainty analysis of aem systems over menindee lake, australia.Geophysical Journal International, 235(2):1888–1911, 09 2023a. ISSN 0956-540X. doi...

work page doi:10.1093/gji/ggad337 1911
[27]

Malcolm Sambridge and Klaus Mosegaard

URLhttp://dx.doi.org/10.11636/Record.2015.029. Malcolm Sambridge and Klaus Mosegaard. Monte carlo methods in geophysical inverse problems. Reviews of Geophysics, 40(3):3–1,

work page doi:10.11636/record.2015.029 2015
[28]

doi: 10.1093/gji/ggz520

ISSN 0956-540X. doi: 10.1093/gji/ggz520. URLhttps://doi.org/10.1093/gji/ggz520. Marc A. Vallee and Richard S. Smith. Application of occam’s inversion to airborne time-domain electromagnetics.The Leading Edge, 28(3):284–287, 03

work page doi:10.1093/gji/ggz520
[29]

doi: 10.1190/1

ISSN 1070-485X. doi: 10.1190/1. 3104071. URLhttps://doi.org/10.1190/1.3104071. K. Vozoff. Electromagnetic methods in applied geophysics.Geophysical surveys, 4(1):9–29,

work page doi:10.1190/1
[30]

doi: 10.1007/BF01452955

ISSN 1573-0956. doi: 10.1007/BF01452955. URLhttps://doi.org/10.1007/BF01452955. Brent Wheelock, Steven Constable, and Kerry Key. The advantages of logarithmically scaled data for electromagnetic inversion.Geophysical Journal International, 201(3):1765–1780,

work page doi:10.1007/bf01452955
[31]

doi: 10.1093/gji/ggv107

ISSN 1365246X. doi: 10.1093/gji/ggv107. Sebastian Wong, Ian Roach, M Nicoll, P English, Marie-Aude Bonnardot, R Brodie, N Rollet, and Alan Ley-Cooper. Interpretation of the ausaem1: insights from the world’s largest airborne electro- magnetic survey.Exploring for the Future: Extended Abstracts, 06

work page doi:10.1093/gji/ggv107
[32]

Sihong Wu, Qinghua Huang, and Li Zhao

doi: 10.11636/134283. Sihong Wu, Qinghua Huang, and Li Zhao. Physics-guided deep learning-based inversion for airborne electromagnetic data.Geophysical Journal International, 238, 07

work page doi:10.11636/134283
[33]

Xin Wu, Guo-qiang Xue, He Yiming, and Junjie Xue

doi: 10.1093/gji/ggae244. Xin Wu, Guo-qiang Xue, He Yiming, and Junjie Xue. Removal of the multi-source noise in airborne electromagnetic data based on deep learning.GEOPHYSICS, 85:1–72, 08

work page doi:10.1093/gji/ggae244
[34]

As expected, the empirical CDF ofx∼ N(0,( 1 3)2), matches CDF N(0,1) (3x), as demonstrated in Figure

The estimate from a standard Broyden-Fanno-Goldfarb-Shanno gradient descent algorithm [Nocedal and Wright, 2006] is shown with the blue triangle. As expected, the empirical CDF ofx∼ N(0,( 1 3)2), matches CDF N(0,1) (3x), as demonstrated in Figure

2006