An electromagnetic response is not a picture of a reservoir
Did you know that a marine controlled-source electromagnetic survey can detect a response consistent with resistive material without being able to name that material? Marine CSEM uses a known electromagnetic transmitter, often a deep-towed electric dipole, and receivers on or near the seabed to measure field responses. The measured electric and sometimes magnetic fields depend on source strength, frequency, geometry, conductivity distribution and noise. Inversion estimates plausible electrical properties beneath the seafloor. An anomaly is therefore evidence about electromagnetic propagation under a model, not a direct sample of oil, gas, freshwater or mineral composition.
A scientific report might show field amplitude against transmitter–receiver offset, phase against frequency, an anomaly relative to a background model, or a coloured resistivity cross-section. Those are different products. The raw recorded quantity may be a small voltage across a receiver electrode pair; converting it to electric field requires the electrode geometry and calibration. A resistivity section results from fitting a numerical propagation model to many observations. Translating the diagram well means preserving the boundary between measured voltage, derived field and inferred subsurface structure.
This chapter is an educational first-principles explanation and not an exploration guarantee. All arithmetic cases below are constructed illustrations, not undocumented field observations. The guiding habit is simple: for every colourful resistive patch, ask which field amplitudes and phases required it, which geometry illuminated it, and whether a different conductivity structure could explain those same observations. That question makes the interpretation more useful to decision-makers than a confident label unaccompanied by uncertainty.
Electrical resistivity and conductivity are reciprocals
Resistivity, represented by rho, describes how strongly a material resists current flow and is conventionally expressed in ohm metres, Ω·m. Conductivity, sigma, describes ease of current flow and is expressed in siemens per metre, S/m. In an isotropic scalar approximation, sigma equals 1/rho. A low conductivity is a high resistivity, not a stronger measured electric field by definition. Effective bulk properties of marine sediments depend on connected pore water, salinity, porosity, clay, temperature and mineral conduction pathways. The numerical values require context before material interpretation.
A model cell with resistivity 2 Ω·m has conductivity 0.5 S/m. Another at 20 Ω·m has conductivity 0.05 S/m. The latter is ten times more resistive, but the measured response does not simply increase tenfold because wave diffusion samples a volume and depends on transmitter geometry. If an axis is labelled log10 resistivity, a change of one unit means a factor of ten. Confusing logarithmic colours with linear values can turn a modest comparison into a fictitious thousandfold geological contrast.
Why seawater is part of the experiment
The ocean is an electrically conductive layer and shapes how source-generated fields reach seabed receivers. Water conductivity varies with salinity, temperature and other conditions. A model of the seabed that omits seawater or assumes an inappropriate thickness can misattribute part of the observed response to subsurface geology. Transmitter altitude above bottom, receiver position and the air–sea interface also influence pathways. The water column is not just a blank space above the geological target; it forms part of the full electromagnetic boundary-value problem.
A good interpretation states which seawater conductivity profile, bathymetry and tidal or depth reference were used. If a constant water value is assumed, test whether plausible change materially alters the anomaly. For shallow water, energy coupling to the air can become important, especially at certain offsets and frequencies. A strong far-offset field should not automatically be interpreted as a deep resistive layer. The correct comparison is with a forward model containing all relevant boundaries, not a land-survey intuition transferred unchanged offshore.
The transmitter dipole and source moment
Marine CSEM commonly uses a horizontal electric dipole whose electrode separation and transmitted current determine a dipole moment, ideally current multiplied by source length in ampere metres, A·m. That moment describes source strength and geometry under a simplified representation. It is not the electric field already measured at a receiver. Current waveform, towing speed, source altitude, electrode placement and orientation can all affect actual response. A record without reliable source logs can produce apparent anomalies that merely track changes in transmitter behaviour.
Suppose a transmitter drives a hypothetical 500 A current across an effective 200 m dipole. Its nominal source moment is 100,000 A·m. If current falls by ten percent while geometry stays fixed, the nominal moment falls to 90,000 A·m. Raw field amplitudes may fall correspondingly even if geology is unchanged in a linear response regime. Normalizing or accounting for actual source moment is important before comparing receiver stations. Avoid converting source strength into a claim of deeper penetration without considering frequency, noise and subsurface conductivity.
Electrode voltage is converted into an electric field
A receiving pair of electrodes measures a potential difference, subject to instrument response and other effects. Dividing by the effective electrode spacing yields a projected electric-field component under appropriate geometry. Units matter because instruments may record microvolts while published plots use volts per metre, microvolts per metre or normalized quantities. Orientation matters too: the same physical vector field produces different components along differently aligned receiver dipoles. The translator should never compare numbers from unlike units or vector components as if they represented one scalar property.
A measured 4 microvolts across a 20 m aligned electrode baseline corresponds to 4 × 10⁻⁶ V / 20 m = 2 × 10⁻⁷ V/m, or 0.2 microvolt per metre, for that projected component. If another figure uses 0.2 microvolt per metre and the first uses 4 microvolts, the two may describe the same measurement in different representations. Check calibration, spacing and orientation before drawing a geological comparison. This simple conversion often prevents bigger errors than a more sophisticated inversion.
Inline, broadside and vertical geometry
A transmitter and receiver separated along the dipole axis form a broadly inline arrangement; a receiver positioned crosswise forms a broadside configuration. Real deployments may include vertical and oblique components. Electric and magnetic fields vary with orientation because the dipole radiates a three-dimensional source field. A response at one geometry should not be treated as interchangeable with another even at equal scalar offset. Forward models must preserve transmitter bearing, receiver orientation and position. A plot labelled Ex can be a component, not the full electric-field magnitude.
Imagine two receivers each 2 km from the transmitter: one located along the tow direction and one 2 km to the side. They need not measure the same amplitude or phase through identical geology because source–receiver geometry differs. A proper report identifies components and survey geometry in every comparison. Modelled inline and broadside responses can have complementary sensitivity, helping distinguish some subsurface structures. Averaging them indiscriminately may erase information about anisotropy or lateral variation.
Frequency and diffusion rather than ordinary ray travel
At the low frequencies commonly used in marine CSEM, electromagnetic propagation through conductive seawater and sediment is usually described as diffusive for many practical interpretations. The amplitude decays with propagation and the phase shifts relative to the transmitter. This differs from interpreting a seismic arrival as a sharply defined travel-time ray reflecting at an interface. An electromagnetic anomaly involves a broad sensitivity pattern rather than one uniquely traced bounce point. Frequency controls how the experiment trades penetration, resolution and detectability.
If a report claims that a transmitter emits 0.25 Hz, one period is four seconds. A signal at 1 Hz has a one-second period. These values should not be confused with survey shot interval or receiver sampling frequency. Frequencies that are too high may suffer strong attenuation at desired offsets; very low frequencies may have weak resolving power for some shallow contrasts and face different noise. The useful band depends on target scale, water depth, resistivity and geometry.
Skin depth gives a scale, not a hard cutoff
For a homogeneous nonmagnetic good conductor under a simplified plane-wave approximation, electromagnetic skin depth is approximately 503 times the square root of resistivity in Ω·m divided by frequency in Hz, with result in metres. One skin depth is the scale at which an ideal field amplitude decreases to about 1/e, roughly 36.8 percent, of its original amplitude for the relevant idealized propagation. Marine electric dipole fields are more complicated than plane waves, so skin depth is a guide to frequency sensitivity rather than a guaranteed maximum survey depth.
At resistivity 1 Ω·m and 1 Hz, the approximate skin depth is 503 m. At 4 Ω·m and 1 Hz, it is about 1,006 m. At 1 Ω·m and 0.25 Hz, it is also about 1,006 m. The last two examples have the same nominal skin-depth scale for different reasons. A reader should not infer identical measured responses: geometry, layer structure and source frequency can still make fields differ. The formula is useful for identifying orders of magnitude and checking units, not for locating a particular boundary.
Amplitude curves require normalization and units
Amplitude plotted against offset may drop dramatically with distance. Some displays use a logarithmic vertical axis because fields span many orders of magnitude. A small vertical distance between plotted lines can represent a large multiplicative difference. Raw observed amplitude and anomaly relative to a reference model are not the same quantity. Differences can arise from source moment, receiver calibration, electrode spacing, orientation, noise and earth structure. Comparing amplitude at unequal frequencies or offsets without adjustment risks a false resistive anomaly.
Suppose a measured field amplitude is 1.8 times a particular reference model at a matched source, receiver, frequency and component. The ratio indicates an 80 percent increase relative to that model, not an 80 percent increase in resistivity. Expressed as an amplitude decibel ratio it would be 20 log10(1.8), about 5.1 dB. A positive amplitude anomaly can have many causes, including geometry mismatch or incorrect background model. Every ratio requires the denominator and definition in the caption.
Phase is a relative angle, not an independent time stamp
A frequency-domain CSEM response has amplitude and phase relative to the source reference. Phase may be displayed in degrees or radians, and the sign convention depends on the Fourier transform and data processing. A 35-degree phase shift at 0.5 Hz corresponds to a fraction 35/360 of a two-second period, approximately 0.194 s as a periodic phase-equivalent shift. That arithmetic does not make it a unique travel time through one layer. In diffusive media the field is a superposition of responses, not a single ray arrival.
Phase can wrap by full cycles, so values near +180 degrees and −180 degrees may be close after unwrapping rather than opposites. Comparing two phase curves requires consistent component, orientation, time convention and reference signal. A phase anomaly may be meaningful even when amplitude is weak, but very low signal-to-noise ratios can make phase unstable. The translator should not write ‘delayed 0.194 seconds at depth’ without explaining that it is a phase-equivalent calculation for a specified frequency and convention.
Real and imaginary components carry the same complex response
An electromagnetic frequency-domain response can be expressed as a complex quantity, with real and imaginary components or equivalent amplitude and phase. The representations contain equivalent information when the definitions are consistent. Mixing the sign conventions, swapping quadrature channels or dropping calibration phase can change an inversion result. A plotted imaginary field below zero is not necessarily a physically negative field strength; it is a signed component relative to the chosen complex reference. Figures should make the convention explicit.
A toy complex field of 3 + 4i in arbitrary consistent units has amplitude 5 and phase arctangent(4/3), about 53.13 degrees. If a translator reproduces amplitude five but calls the phase 4 degrees, the meaning is corrupted. Real observations require calibrated component units and source reference. For teaching, make clear that the toy values are mathematical rather than actual CSEM measurements. Conversions are legitimate only if the source and receiver use the same phase and temporal reference system.
Offset controls which volumes influence a response
Longer transmitter–receiver offsets can provide sensitivity to different subsurface regions, but do not simply increase investigation depth in proportion to distance. At large offsets the signal weakens, noise becomes important, and airwave or other pathways may dominate. Short offsets can be sensitive to shallower conductivity and source geometry. A useful acquisition design includes a range of offsets and frequencies rather than one heroic distance. An inversion may show a deep feature even when the longest-offset data are too noisy to constrain its shape.
If two surveys have different receiver spreads, comparing their final resistivity sections without looking at offset coverage is risky. A high-resistivity region near the model bottom might be poorly illuminated in one survey. Ask which data actually change when that cell is modified in forward modeling. A report should include offset ranges, detection thresholds and sensitivity plots. Describing a feature as ‘unresolved beyond the available offsets’ is more honest than declaring its physical termination at the edge of the model grid.
The airwave problem in shallow marine surveys
Electromagnetic energy can couple through the air and return to seafloor receivers, producing a response whose dependence on water depth and offset may complicate interpretation. Under some shallow-water conditions, that pathway can dominate fields at ranges where an analyst hoped to sense deep sediments. The specific behaviour depends on source geometry, frequency, water properties and receiver setup. A strong far-offset amplitude is therefore not automatically evidence for a resistive deep target. Correct modelling needs the air–sea boundary.
Compare forward models with and without the air boundary under the actual water depth, rather than assuming it can be ignored because the receivers rest on the seabed. If an anomaly vanishes when realistic water depth is included, the earlier interpretation was incomplete. Do not generalize that airwave always makes marine CSEM impossible or always makes it useful; survey design and frequency choice can manage some limitations. The meaningful question is whether data sensitivity to the intended target survives the competing propagation path.
Noise floors and weak-signal phase instability
Receivers record natural electromagnetic disturbances, instrument noise, motion effects, electrode drift and possible cultural interference. A published amplitude curve may approach the noise floor at long offsets. Below that floor, plotted phase can fluctuate wildly and a tiny apparent anomaly may be numerically meaningless. Stacking repeated waveform cycles can improve detection under appropriate assumptions, but coherent noise does not disappear just because more time is averaged. Quality control must address frequency-domain uncertainty and instrument calibration.
If a measured field amplitude is 0.2 units while the noise amplitude is of similar size, the inferred phase and amplitude ratio may be unreliable. Printing six decimal places does not improve signal-to-noise ratio. A transparent report shows data errors, rejected points, noise estimates and how they affect inversion weights. A model that fits weak observations perfectly while neglecting reliable ones may be overfitting. Describe non-detections as below the stated sensitivity, not as proof that a geological conductor or resistor is absent.
A resistive anomaly has non-unique geological causes
Hydrocarbon-filled reservoirs can be electrically resistive relative to some saline sedimentary backgrounds, which is one motivation for marine CSEM. But rock texture, pore-fluid salinity, carbonate cementation, gas, fresh pore water, dense lithologies and anisotropy can also affect effective resistivity. The same field anomaly may have more than one geological explanation. A high-resistivity patch is not a direct chemical assay or an automatic commercial discovery. Domain knowledge and independent seismic or well evidence are required for classification.
A responsible translation could say ‘the recorded response is consistent with a localized resistive structure under the tested conductivity models; its cause and geometry are not uniquely established.’ That sentence gives a useful finding without making an unsupported commodity claim. If a report instead writes ‘oil detected’ because an inversion cell is orange, it skips the entire material-identification problem. The more high-stakes the interpretation, the more clearly alternative explanations, acquisition sensitivity and validation requirements should be presented.
Thin-layer equivalence and the depth–resistivity tradeoff
For thin resistive layers, measured CSEM response may depend on combinations of layer thickness, resistivity and depth rather than fixing each individually. A thinner strongly resistive layer can sometimes produce an anomaly resembling that of a thicker less resistive layer within the measured frequency and offset range. Model parametrization and regularization further shape the final geometry. This is an equivalence problem, not an instrument fault. The final figure’s sharp layer thickness may be imposed by choices rather than uniquely recovered.
To test robustness, forward-model several thickness–resistivity combinations and compare their predicted fields with the data uncertainties. If a broad family fits, report that family or a constrained combination instead of one exact thickness. The method may detect an elevated resistive response without resolving its vertical distribution. Complementary reflection data or borehole information can reduce ambiguity. When translating a section, retain a phrase such as ‘equivalent resistivity structures remain possible’ rather than converting a preferred numerical model into direct observation.
Anisotropy changes the meaning of a single resistivity colour
Layered sediments can conduct differently parallel and perpendicular to bedding. An isotropic inversion assigns one scalar resistivity to a cell; an anisotropic model may need direction-dependent values. A survey’s sensitivity to those directions depends on source–receiver orientation and geometry. Ignoring anisotropy can map a directional material response into spurious layer thickness or depth. The word resistivity alone can therefore hide a modelling convention with major geological consequences.
Imagine a finely laminated sediment whose horizontal conductivity differs from vertical conductivity. An inline electric field and a broadside field might respond differently even without a distinct localized reservoir. Before treating a mismatch as an exotic target, test whether reasonable anisotropy explains the data. Report whether the inversion assumes isotropy and how that assumption affects the key result. A single colour scale cannot faithfully express every direction-dependent property of the physical formation.
One-dimensional, two-dimensional and three-dimensional earth models
A one-dimensional model assumes electrical properties vary only with depth, useful for studying layered backgrounds and some idealized targets. Two-dimensional models allow lateral variation in one horizontal direction but remain invariant along another; the source itself remains three-dimensional. Full three-dimensional modelling allows variation in all directions. More complex models can represent more geology but also require more computational resources and greater data support. Choosing 3D does not automatically remove ambiguity; it can introduce additional poorly constrained parameters.
A survey crossing a narrow channel or complex seabed feature may violate 1D assumptions even if individual stations look smooth. If the anomaly changes markedly with azimuth, test 2D and 3D possibilities. Conversely, if the acquisition has only one line with little crossline information, a detailed 3D image may be heavily model determined. State what dimensionality means and what evidence supports it. Terms like 2.5D describe specific source and earth dimensional assumptions, not half a measurement or half the accuracy.
Forward modelling is the basic test of a proposed interpretation
A forward model predicts the electromagnetic response of a specified conductivity structure, including transmitter, receiver, seawater and boundary geometry. Inversion works backward, seeking models that match observed fields under chosen criteria. The two directions should be kept separate. A geological sketch may look sensible but must still predict the actual amplitudes and phases within realistic uncertainty. Conversely, a numerically good fit can arise from a geologically implausible construction if enough parameters are available.
For example, suppose a preferred model reproduces an observed amplitude ratio but misses a phase trend by much more than the reported uncertainty. Calling it fully validated would be premature. Examine source calibration, target conductivity, model geometry and overlooked pathways. Simulate plausible alternatives rather than tuning one parameter until one curve looks attractive. A good report describes which observations are explained, which are not, and whether the residual pattern is systematic.
Smoothness regularization makes a particular kind of answer
Many inversion methods balance data misfit against a model roughness or complexity penalty. Occam-style inversion often seeks comparatively smooth models that fit data to an acceptable level. The smooth result is not proof that geological conductivity changes are smooth in reality. A sharp true boundary can be spread across several cells, while poorly sampled regions may inherit values from regularization. An interpreted boundary line drawn afterward should identify that extra decision.
Compare inversions using different reasonable regularization strengths, initial resistivities and bounds. A target that persists may be more robust than one that vanishes with a small parameter change. Yet robustness under a narrow algorithm family does not guarantee unique geology, particularly if source calibration or dimensionality is wrong in every run. Report model sensitivity and the data points that constrain the feature, not just a colour map with an appealing smooth gradient.
Misfit, weighting and correlated errors
A CSEM inversion commonly fits multiple complex observations at many frequencies, offsets and receiver components. If uncertainties differ, data weighting prevents tiny noisy measurements from dominating simply because their raw residuals are expressed in convenient units. Amplitude and phase errors can be correlated, and systematic source or geometry biases may affect groups of observations. A small scalar misfit number cannot by itself certify geological correctness. Look for residual plots and evidence that the chosen error floor is realistic.
Suppose an inversion has a normalized misfit close to one under a particular error model. That suggests consistency with the assigned errors if the assumptions are valid, not an abstract 100-percent accuracy score. Increasing assumed errors can reduce normalized misfit without changing measurements. Conversely, insisting on unrealistically small errors can force the model to explain noise. A rigorous translation states the error definition, data selection and whether residuals retain coherent structure with frequency or offset.
Survey sensitivity and resolution tests
Sensitivity describes how predicted fields change when conductivity or geometry changes in a particular region. A model cell can have a definite resistivity value while the data barely respond to it. Resolution tests compare responses for alternative targets or attempt to recover synthetic anomalies with the actual acquisition layout and realistic noise. The result should be interpreted as a property of the experiment and inversion choices, not a universal detector guarantee. The lower edge of a coloured model is often a computational boundary rather than the depth of investigation.
If a synthetic resistive body placed in the model is not recoverable with available offsets and noise, the real dataset cannot uniquely exclude a similar body simply because the final inversion lacks it. Likewise, if the recovery spreads a sharp anomaly over hundreds of metres, the final displayed thickness should not be quoted to the nearest metre. Ask for sensitivity sections or credible bounds on geometry. A transparent explanation distinguishes detection, localization and classification as different achievements.
Combining CSEM with seismic data without circular reasoning
Seismic reflection estimates acoustic contrasts and structural geometry; CSEM constrains electrical responses under electromagnetic models. They can complement each other because their physical sensitivities differ. But if the seismic interpretation is built into the CSEM model as a fixed boundary and the resulting resistivity anomaly is later called an independent proof of that boundary, reasoning has become circular. Document which priors were introduced and which observations are genuinely independent.
A defensible integrated report might say that seismic data constrain a candidate layer geometry and CSEM responses are compatible with higher resistivity within that geometry. It should then test alternatives without the fixed layer or with different resistivities. A consistent pair of models is valuable, yet classification still needs additional evidence. Teaching this distinction helps readers understand why agreement between two processed figures is stronger only when their assumptions have not been silently made identical.
Source-receiver geometry can be mistaken for geology
An unexpected change in transmitter heading, dipole altitude, receiver tilt or distance can alter amplitude and phase. If the inversion uses incorrect geometry, the solver may fit those changes by introducing artificial conductive or resistive bodies. Before describing a lateral anomaly, examine navigation and deployment metadata at the same positions. Repeated measurements under changed geometry, or complementary components, can help distinguish a physical feature from a setup error.
Consider two adjacent tow segments where the source altitude rises because the seabed drops away. A receiver may see an amplitude change even if subsurface electrical properties remain similar. A model that holds altitude constant could create a false lateral contrast. The repair is not to smooth the anomaly away by eye but to input the best supported geometry, rerun forward predictions and compare the remaining residual. Describe only the contrast that survives plausible acquisition corrections.
Amplitude uncertainty should be expressed relative to scale
Field measurements may differ by orders of magnitude between near and far receivers. Reporting one fixed absolute error for every amplitude may be inappropriate; a mixture of relative error and noise floor is common in some analyses. Phase uncertainty can also grow where signals weaken. Every anomaly plot should communicate its reference model and uncertainty definition. A statistically significant small change at strong signal is not equivalent to the same percentage difference at the detection limit.
For a synthetic comparison, measured amplitude 1.00 × 10⁻⁹ V/m versus model 0.90 × 10⁻⁹ V/m gives a ratio of approximately 1.111, or 11.1 percent above the model. That number has no inferential value until measurement error and geometry are evaluated. If uncertainty is ±0.20 × 10⁻⁹ V/m, the apparent difference may be uninformative. The translation should not call the contrast a confirmed geological anomaly merely because the ratio is greater than one.
Phase wrapping and reference signs need reproducibility
Phase plots may express angles from −180 to +180 degrees or from 0 to 360 degrees. A displayed jump from +179° to −179° can represent a continuous two-degree change across the wrap, not a sudden 358-degree physical event. Referencing phase to the source waveform requires consistent timing and sign convention. Component rotations can introduce apparent sign changes. If these choices differ between figures, the same field can appear contradictory.
An appropriate QC step recreates phase from the underlying complex data and compares unwrapped curves under one stated convention. Do not draw geological boundaries at a phase-wrapping discontinuity without checking the representation. A reader-oriented caption explains whether phase is wrapped and what zero means. As in language translation, preserving the sign of a scientific term is more important than making the graph visually smooth.
Survey design must match the decision
A small shallow conductivity contrast and a deep resistive target can require very different frequencies, offsets, transmitter moments and acquisition layouts. Choosing parameters by copying a different survey’s brochure may produce measurements but not sufficient sensitivity for the current question. A meaningful survey plan begins with expected water depth, geological alternatives, conductivity ranges, target scale and available validation. Forward studies then test whether plausible cases generate distinguishable predicted responses above noise.
If two competing models predict fields differing by less than the instrument uncertainty at all planned receivers, the proposed experiment cannot resolve the decision as designed. That result is valuable before spending money on acquisition. It may motivate new frequencies, positions, components or complementary methods. In a report, state the target and discrimination objective rather than calling a survey ‘high resolution’ with no defined resolvable difference.
Worked case: skin depth under two conditions
For the ideal nonmagnetic homogeneous approximation, δ ≈ 503 sqrt(ρ/f) metres. At resistivity 0.5 Ω·m and 0.5 Hz, ρ/f = 1 and δ is about 503 m. At resistivity 2 Ω·m and the same frequency, ρ/f = 4 and δ is about 1,006 m. The second skin depth is twice the first, not four times. The square-root relationship matters when comparing frequency choices.
Now double frequency in the 2 Ω·m case from 0.5 to 1 Hz. Then ρ/f falls from 4 to 2 and δ becomes about 711 m. These are ideal characteristic scales, not promised detection depths for a dipole survey. A sound interpretation adds the actual source–receiver geometry, conductivity layering and noise. If someone writes that a skin depth of 711 m means all information below 711 m is absent, they have converted exponential attenuation into a false hard boundary.
Worked case: checking source moment and receiver field
A transmitter with effective dipole length 150 m and current 400 A has nominal moment 60,000 A·m. At one receiver, a projected 6 microvolt difference across a 30 m baseline gives 0.2 microvolt per metre, or 2 × 10⁻⁷ V/m. The moment and field have different units and cannot be compared by division as if they were the same property. A model must relate them through source geometry and earth conductivity.
If the transmitter current changes to 300 A while length stays fixed, nominal moment falls to 45,000 A·m, a 25 percent decline. Comparing field amplitudes before and after this change without source normalization may create an artificial interpretation. This exercise illustrates the difference between measured electrical activity and inferred geology. Record actual current waveform and receiver calibration rather than trusting nominal settings alone.
Worked case: amplitude and phase together
A synthetic receiver has observed amplitude 1.8 units and reference-model amplitude 1.0 in the same units and component. The anomaly ratio is 1.8; the amplitude is 80 percent higher than the reference. Suppose measured phase is 35° and reference phase is 25° under the same convention. The phase anomaly is 10°, not a claim about ten degrees of seabed dip. Both numbers constrain the model; satisfying one while neglecting the other is not full consistency.
At 0.5 Hz, 10 degrees corresponds to about 0.0556 s as a periodic phase-equivalent shift, but not to one unique propagation time. A different conductivity structure or source calibration could affect both response features. Comparing model alternatives requires their predicted complex fields and associated errors. The scientifically accurate conclusion is that the specified response departs from the chosen reference in amplitude and phase, with geological meaning still dependent on modelling.
Worked case: nonunique layer interpretations
Imagine a synthetic CSEM dataset that is compatible with a thin strongly resistive layer or a thicker moderately resistive one because the measured frequencies and offsets have limited resolving power. The first model may place the layer top slightly deeper and make it thinner; the second may distribute a contrast over a larger vertical region. If both predicted complex responses fall inside defensible uncertainties, choosing the more vivid figure does not establish the true thickness or resistivity.
The next useful action is to identify which new measurement would separate the alternatives. Perhaps a different receiver orientation, a better frequency range or an independent seismic boundary could help; in other cases the problem remains inherently ambiguous. The report should present the unresolved tradeoff and avoid quoting each parameter as if independently measured. This form of translation is useful for technical and public readers because it reports both the result and its limits.
Transfer problem 1: reciprocal quantities
A model lists conductivity 0.25 S/m. What is resistivity? Answer: 1/0.25 = 4 Ω·m, assuming the scalar isotropic relation. Do not confuse the two units or invert a number already given as resistivity.
Transfer problem 2: source strength
A 100 m effective dipole carries 600 A. What is the ideal dipole moment? Answer: 60,000 A·m. It is a source quantity, not the electric field at a receiver.
Transfer problem 3: electrode geometry
A receiver registers 5 microvolts over 25 m along one axis. What projected field is inferred? Answer: 0.2 microvolt/m, equivalently 2 × 10⁻⁷ V/m, before broader calibration considerations.
Transfer problem 4: skin depth scale
For ρ = 1 Ω·m and f = 4 Hz, the ideal skin depth is about 503 sqrt(1/4) = 251.5 m. It is a scale of attenuation, not a hard survey range.
Transfer problem 5: source period
At f = 0.2 Hz, what is one oscillation period? Answer: 1/0.2 = 5 s. Do not call it the instrument’s recording sample interval.
Transfer problem 6: amplitude ratio
Observed amplitude is twice the reference response. What is the correct relative statement? Answer: the measured field component is 100 percent higher than that model under comparable conditions; resistivity has not thereby doubled.
Transfer problem 7: phase wrap
A phase curve moves from +179° to −179°. Must it contain a 358-degree physical jump? Answer: no. Across a wrap convention it can be a continuous two-degree change; inspect unwrapped phase.
Transfer problem 8: low signal-to-noise ratio
An apparent far-offset phase anomaly comes from a signal below a stated noise floor. Should you identify a deep resistor? Answer: no. The phase estimate may be unreliable and sensitivity must be demonstrated.
Transfer problem 9: model dimension
A one-dimensional model gives a good fit along one line crossing variable bathymetry. Is it certainly correct? Answer: no. Evaluate whether lateral structure or the geometry violates 1D assumptions.
Transfer problem 10: material identity
An inversion contains a high-resistivity lens. Does the image prove petroleum? Answer: no. It establishes a model-dependent electrical contrast with several possible physical causes.
Transfer problem 11: independent corroboration
Seismic picks were fixed as boundaries in a CSEM inversion. Can the final CSEM section be presented as entirely independent proof of those boundaries? Answer: no. Disclose the introduced structural priors.
Transfer problem 12: robust reporting
Write a sentence about a CSEM anomaly compatible with several resistivity structures. Model answer: ‘Measured frequency-domain fields depart from the stated reference response and are consistent with a resistive subsurface contrast; layer dimensions and material cause remain non-unique within the tested model family.’
FAQ: Does marine CSEM measure resistivity directly?
It measures electromagnetic responses such as voltages and field components. Resistivity is inferred using source, receiver and earth models. The final map carries assumptions and uncertainty.
FAQ: Does a lower frequency always give a better image?
No. Lower frequency may increase the ideal skin-depth scale but can reduce sensitivity to some shallow features and faces noise and acquisition tradeoffs. Target size and actual conductivity matter.
FAQ: Does higher resistivity always mean hydrocarbons?
No. Effective resistivity reflects pore fluids, rock matrix, temperature, salinity, anisotropy and many other properties. A geological identification requires independent information.
FAQ: What is the airwave?
It is an electromagnetic pathway involving the air boundary that can influence offshore CSEM responses, notably under some shallow-water and offset configurations. It must be considered in the forward model.
FAQ: Why use both amplitude and phase?
They are complementary parts of the complex frequency-domain response. A model that matches amplitude but not phase may be incomplete; both must be evaluated with appropriate uncertainty.
FAQ: Is a coloured inversion section unique?
No. Different conductivity structures and regularization assumptions can produce similar fields. The colours express a preferred model under stated constraints, not directly sampled rock.
FAQ: What does below detection limit mean?
It describes data sensitivity under specified noise and acquisition conditions. It does not prove that every geological contrast is absent beyond the reported range.
FAQ: What is the most useful caption for a resistive anomaly?
Name the field component and frequency range, source–receiver geometry, reference comparison, inversion assumptions and unresolved alternatives. State whether the material identity is verified or proposed.
Final checklist for an accurate marine CSEM translation
Check voltage and field units, effective electrode separation, source current and dipole length, seawater properties, receiver orientation, offset, frequency and phase convention. Determine whether curves are raw, normalized or model anomalies. Examine noise floors, sensitivity, residuals, regularization, anisotropy and alternative conductivity models. Only then translate a resistive patch into language about the subsurface. An interpretation is strongest when a reader can reconstruct the evidence chain and see what further data would settle the remaining alternatives.
Primary sources and verified further reading
Scripps Marine Electromagnetics Laboratory: Marine CSEM and MT concepts
Scripps: Web Hosted Active-source Modeling (Dipole1D)
Scripps: OCCAM1DCSEM inversion methodology and tools
UC San Diego: CSEM method and skin-depth interpretation
Google Search Central: People-first content guidance
eduKateSG: How to Translate Easily index