The image is a model of electrical response between boreholes
Crosshole electrical resistivity tomography, often abbreviated crosshole ERT, uses electrical measurements made with electrodes installed in or near separate boreholes to infer the distribution of subsurface resistivity between them. The reader’s practical task is to separate measured currents and voltages from the inverted resistivity image, then decide which parts of that image are actually constrained by survey geometry and data. If you need to understand crosshole ERT electrode arrays, inversion resolution and time-lapse changes, start by checking how electrodes were positioned and how current and voltage measurements produced the displayed panel.
A low-resistivity feature between two wells is not automatically a contamination plume, a wet fracture or clay. Electrical conduction can be influenced by pore-fluid salinity, connected porosity, saturation, temperature, clay surface effects and some conductive minerals. A colourful panel does not identify those causes uniquely. The interpretation becomes stronger when the geometry, measurement quality, background geology and independent sampling are available. The useful question is what combination of models and direct evidence supports the feature, not whether the colour looks convincing.
Before drawing a conclusion, read the survey header: borehole separation, electrode depths and number, depth datum, deviations from vertical, current injection patterns, measured potential differences, estimated errors, contact resistance, inversion dimensionality, starting model, regularization and any time-lapse reference survey. Also ask which parts of the crosshole panel were tested for sensitivity or depth of investigation. These details govern the reliability of individual image features. Two surveys with the same display palette can convey very different measurement strength.
What electricity does in the ground
An electrical current injected through two current electrodes produces a potential field influenced by the conductivity structure of the surrounding ground. Two potential electrodes sample a voltage difference produced by that field. The simplest four-electrode relation is apparent resistivity rho_a = K × delta V / I, where I is injected current, delta V is the measured potential difference and K is a geometry factor with units of length. The factor accounts for the idealized electrode arrangement and medium assumed in the calculation.
This is an apparent property because the current travels through a volume of earth, not a single pixel at the position of one electrode. The same voltage measurement can be affected by many subsurface regions simultaneously. Crosshole ERT inversion solves a different question from direct logging: it looks for a spatial conductivity or resistivity distribution whose forward-modelled responses match a large collection of measurements. That distribution is an inference constrained by data and model assumptions. The distinction between direct voltage and reconstructed pixel is foundational.
Four electrodes and a concrete numerical check
Take a constructed data example in which the instrument injects 0.10 ampere and records a potential difference of 0.020 volt. The measured resistance for that configuration is delta V/I = 0.20 ohm. If the properly computed geometric factor for its idealized configuration is 50 metres, the apparent resistivity is 10 ohm-metres. Doubling both current and voltage would retain the same ratio and the same apparent resistivity, provided the system remains linear and measurement conditions are otherwise comparable.
If the analyst forgets that 20 millivolts equals 0.020 volt, the computed value can be wrong by a factor of one thousand. If the geometry factor is copied from a surface array that does not match the borehole electrodes, the result can be physically misleading even with correct voltage arithmetic. Crosshole geometry, finite borehole spacing and boundaries require an appropriate forward model; a surface-array geometric factor should not be transplanted by name alone. Always verify units and electrode configuration before reading the processed resistivity.
Why crosshole is not a single borehole resistivity log
A conventional single borehole resistivity log investigates electrical response near a moving probe and plots a value against depth. Crosshole ERT uses many current and voltage configurations spanning a region between electrodes to reconstruct a two- or three-dimensional distribution. The resulting image is sensitive to broad patterns in the inter-borehole space and to its survey design. The methods may use related electrical physics, but the measurement geometry and inference process are not interchangeable.
A single normal-resistivity log might show a narrow conductive interval where a tool passes through clay or saline groundwater. Crosshole ERT may show whether a conductive region extends between two wells, but the apparent continuity depends on resolution and regularization. Smooth inversion can bridge separate conductors into one band; an under-sampled interval may not be resolved at all. The existing borehole normal-resistivity guide covers the near-well technique. Read a tomography panel with a separate geometric and statistical mindset.
Crosshole electrode arrays change sensitivity
Different combinations of electrodes create different current paths and voltage sensitivities. Some configurations place current electrodes in different boreholes and measure a voltage difference using another crosshole arrangement. Other configurations may have current or potential pairs within a single hole. Their relative response to changes in subsurface structure, geometric error and electrode contact can differ. A dense collection of measurements does not automatically provide uniform sensitivity across the entire image.
Research on crosshole ERT has shown that some electrode configurations are unusually sensitive to errors in electrode positions, particularly particular in-hole geometries. Small vertical offsets can create large apparent-resistivity anomalies in unfortunate configurations. Consequently, field-data filtering and electrode layout design matter before inversion. Do not assume that because four electrodes were involved the measured point belongs to their geometric midpoint. Electrode sensitivities extend through the domain in nontrivial patterns determined by electrical boundary conditions.
Depth in two wells must share one coordinate system
One hole may lean away from the other as depth increases, while another leans toward it. Their spacing at the collar therefore may differ from spacing at the depths of the electrode array. When the forward model assumes perfectly vertical boreholes and the true geometry differs, predicted electrical potentials can be wrong. The inversion may compensate by inventing resistivity contrasts that are not present in the ground. This is a geometry-to-geology translation error.
Use surveyed coordinates for electrodes when accuracy requires them. A departure of one metre at depth might be negligible in a widely separated sparse survey or crucial in a narrow high-sensitivity panel. Compare electrode positions in east, north and elevation rather than only measured-depth labels. The companion deviation and true vertical depth guide explains that conversion. Incorrect geometry can survive even a numerically well-converged inversion, because the solver may fit wrong positions with wrong material properties.
What the inversion actually optimizes
A resistivity inversion typically starts with a parameterized model of subsurface conductivity or resistivity. A forward solver predicts voltages for the measured electrode configurations. The difference between observed and predicted voltages is a data misfit. The inversion changes model values to reduce that misfit, often while adding a regularization penalty to discourage implausibly rough or unstable solutions. This combination turns a physically underdetermined problem into a tractable estimate.
Regularization is not an evil manipulation of evidence; it is an explicit assumption needed because a limited set of measurements cannot specify every subsurface property independently. However, it can smooth sharp contacts, blur small conductive zones or spread an anomaly toward areas with weak sensitivity. Therefore, report regularization choices and sensitivity tests. A low final misfit does not prove the spatial distribution is unique. Multiple models can fit the same observations within measurement uncertainty, particularly where ray or current-path coverage is limited.
An image cell is not a discrete measurement
A coloured square or polygon in an ERT panel represents a parameter of the chosen inversion mesh. The instrument never placed a miniature resistivity meter into each cell. A coarse mesh can hide local variations; a very fine mesh can suggest detail that data cannot resolve. Cell dimensions should be interpreted with reference to survey spacing, electrode coverage, regularization and resolution tests. Increasing display resolution by interpolation does not create new physical information.
Compare a panel displayed as broad blocky cells with an artistically smoothed version. Both may derive from the same inverted values. The second can appear to show a continuous pathway crossing several metres, but that continuity may come from plotting style or regularization. Annotate actual electrode locations and model boundaries. If a feature is smaller than the effective resolution, describe it as a localized response rather than a sharply bounded physical object. The honest figure makes the gap between measurement and interpretation visible.
High and low resistivity are relational clues
Resistivity is the inverse of conductivity for the same material in an appropriate scalar isotropic description. A value of 20 ohm-metres corresponds to 0.05 siemens per metre, while 10 ohm-metres corresponds to 0.10 siemens per metre. Resistivity halved, conductivity doubled. These reciprocal relationships matter when comparing reports that use different colour scales or physical quantities. But neither number directly identifies salinity without other information about pore geometry and conductive minerals.
In saturated sandy materials with limited clay contribution, increased dissolved-ion concentration can reduce electrical resistivity. Yet clay-rich sediments may also be conductive even where the pore water is not especially saline, and dry rock may be highly resistive for other reasons. Geological interpretation therefore requires context. A saline tracer injection might produce a time-lapse conductive anomaly, but a baseline conductive clay layer could look similar in a single survey. Separating static properties from controlled change improves causal interpretation.
The meaning of an apparent resistivity pseudosection
Some reports display measured apparent resistivity using a plotting convention called a pseudosection. It places measurements according to electrode geometry in a convenient graphical layout. That plotted vertical coordinate is not necessarily the physical depth of a measured object. A pseudosection is a data visualization, not an inverted subsurface section. Confusing it with a material-property image is among the simplest ways to overinterpret ERT output.
An inverted section, by contrast, uses a physical forward model and parameterized subsurface geometry to estimate a resistivity distribution. Even then, the apparent sharpness of a feature depends on inversion regularization and coverage. A report should label which view is shown. Compare raw observed versus calculated values, the residual distribution and the final model. If the plotted data are not identified, do not assign an exact geological depth to a colour anomaly. The first diagnosis is to determine what was actually plotted.
Contact resistance and acquisition errors
Poor electrical contact at electrodes can increase noise or limit attainable current. Cable failures, channel switching errors, power-line interference, unstable electrode polarization and changing borehole fluid chemistry can also affect measurements. A quality-controlled survey records injected current, potential response, contact resistance or equivalent diagnostics, acquisition stacking and appropriate error estimates. The source data require inspection before inversion, not only after an unexpected image appears.
An instrument might inject 0.001 ampere in a configuration where intended current was 0.1 ampere because an electrode connection is poor. If measured voltage noise remains similar, the signal-to-noise ratio can deteriorate dramatically. Using a nominal current instead of the measured current also corrupts apparent resistivity. Some acquisition systems automatically adjust voltage or current, so read the actual logged values and instrument flags. A smooth inversion obtained after filtering an unreported fraction of noisy measurements should be interpreted alongside a transparent data-rejection record.
Reciprocal readings as a quality check
Electrical reciprocity provides a practical diagnostic under suitable linear and stationary conditions. In a four-electrode configuration, exchanging the current and potential electrode pairs should yield equivalent transfer resistance when sign conventions and geometry are handled consistently. The difference between reciprocal measurements can reveal unstable contacts, electrode problems or noise. It is not a magical proof of correctness: correlated systematic errors or changing environmental conditions may survive a reciprocal test.
For example, one configuration returns a transfer resistance magnitude of 0.200 ohm and its appropriately signed reciprocal returns 0.212 ohm. The absolute difference is 0.012 ohm, around 6 percent of the first measurement. Whether that is acceptable depends on the survey error model and planned interpretation. A third pair with a 40 percent mismatch deserves closer scrutiny. Retain both readings and the sign convention rather than averaging before quality review. Inversion weighting should reflect credible measurement errors, not an arbitrary demand for every point to fit perfectly.
The difference between fitting data and validating geology
An inversion algorithm may report a low root-mean-square residual, but that statistic says only how well the model reproduces data under the chosen assumptions and weights. It does not establish that a conductive body has the depicted size, shape or material identity. A model can sometimes fit noise, especially with flexible parameterization, or fit inaccurate electrode geometry by distorting conductivity. Residual maps, sensitivity analysis, alternative regularization and independent observations remain necessary.
Imagine two models with similar residuals: one contains a narrow high-conductivity channel, while another contains a broader moderate anomaly. If electrode coverage cannot discriminate them, the claim of a narrow fracture is unsupported. A direct borehole sample or independent image might help choose between candidates. Report the plausible family of interpretations rather than granting the visually preferred one exclusive status. This is the central inversion lesson: a good data fit is necessary for many conclusions but not sufficient to establish uniqueness.
A useful sequence for an initial reading
Begin with the actual field question: locating a preferential pathway, tracking an injected amendment, estimating background layering or checking an engineering structure. Confirm borehole and electrode coordinates, depth datums and installation conditions. Review measurement counts, current and voltage ranges, reciprocity and contact diagnostics. Identify whether the displayed plot is raw pseudosection, inverted resistivity or time-lapse ratio. Inspect model domain, regularization and coverage or resolution indicators.
Only after these checks should the reader translate colour regions into geological possibilities. State alternative causes of conductivity variation, then compare direct sampling, lithologic logs, tracer history or hydraulic evidence. Highlight areas with reliable sensitivity and avoid exact boundaries in unsupported zones. The result should be a decision-support interpretation with known limits, not a decorative image. If the method’s resolution cannot answer the original question, the appropriate recommendation is additional evidence, not stronger adjectives in the caption.
Read an ERT image through its spatial resolution
An inverted crosshole ERT image is composed of model cells, not little direct conductivity measurements made at each cell. The electrical instrument recorded currents and voltages for particular electrode combinations. An inversion estimated a spatial distribution that could reproduce those measurements while respecting a set of mathematical constraints. The apparent width and sharpness of a feature therefore depend on electrode coverage, measurement error, mesh geometry and regularization. Treating each coloured cell as independently measured geological fact would confuse a model representation with an observation.
To test which features are actually resolvable, make synthetic conductivity bodies with known locations, predict measurements using the same electrode geometry, and invert those simulated measurements without telling the solver the true pattern. If it merges two nearby conductors into one diffuse band, the real survey may not discriminate those shapes. Such tests depend on chosen contrast, size and noise. A test that resolves a broad block does not establish that it could resolve an arbitrarily thin fracture. Match the resolution experiment to the intended geological question.
Sensitivity is not the same as the ability to separate objects
Sensitivity describes how strongly a particular measurement responds when a property’s value changes in a particular region. Resolution also depends on whether measurements contain enough independent information to separate neighbouring variations. A region can influence multiple voltages and still lack the independent coverage necessary to distinguish two nearby conductive bodies. In the absence of discrimination, a smoothness-constrained inversion may combine two distinct objects into one attractive shape. That visual continuity is then partly an assumption rather than a direct finding.
Suppose two narrow fractured zones are spaced a metre apart midway between widely separated boreholes. A model may depict a single conductive stripe, while another plausible conductivity distribution with two separate stripes produces nearly identical calculated voltages. If both fit the observations within stated measurement uncertainty, a report should avoid claiming a continuous fracture. It can defensibly describe an electrically conductive region with uncertain internal structure. Additional electrodes, different configurations or independent geological data could determine whether the more precise interpretation is justified.
A two-dimensional section may conceal three-dimensional current paths
Current flows through a three-dimensional ground volume even when a report displays one vertical plane between two boreholes. A two-dimensional inversion generally assumes that variation outside its displayed plane is sufficiently restricted to justify the model. A conductive object beside the plane can influence voltages and may be projected into the two-dimensional picture as if it occupied the gap between the holes. The problem is physical, not a defective screen-rendering choice: electricity follows the available conductive pathways rather than the boundaries of a figure.
Check whether the forward calculations are two-dimensional, two-and-a-half-dimensional or fully three-dimensional, and what assumptions each makes about the surrounding ground. A narrow interwell colour feature is not proof of a connected pathway lying precisely in that section. Additional electrode configurations, neighbouring sections, a third borehole or groundwater samples can test that hypothesis. When such evidence is unavailable, state the geometric ambiguity explicitly. Avoid describing the physical volume as if it were fully mapped by a single slice.
A conductivity percentage cannot be copied from resistivity
Resistivity and conductivity are reciprocals under an appropriate scalar material model. Suppose an inverted cell has a baseline resistivity of 20 ohm-metres and a later value of 12 ohm-metres. Its resistivity fell by (20 minus 12)/20, or 40 percent. Conductivity instead rose from 1/20 = 0.050 siemens per metre to 1/12 = approximately 0.0833 siemens per metre. The conductivity increase relative to its original value is about 66.7 percent, not 40 percent.
An independent check uses multiplication: 20 ohm-metres times 0.050 siemens per metre equals one, and 12 ohm-metres times approximately 0.0833 siemens per metre is approximately one. If the report shows change in logarithmic resistivity or another transformed quantity, use that definition rather than interpreting the legend as a straightforward difference. A more conductive region is not automatically more saline groundwater. Saturation, clay contributions, geometry and other factors can alter electrical response, so chemistry requires independent support.
Time-lapse observations require a stable reference
A time-lapse ERT comparison may follow an injected tracer, recharge event or remediation treatment. The baseline is a specific earlier measurement state, not an abstract natural condition guaranteed to remain fixed. Electrode contacts, borehole fluid conductivity, temperature, water level, cable configuration and noise can change between surveys. Consequently, a difference between inverted pictures can arise partly from changing measurement conditions even when formation conductivity is stable. A time-lapse interpretation starts by comparing matched acquisition records wherever possible.
Identify whether each survey was inverted separately or whether a joint inversion constrained temporal changes. The latter may make maps easier to compare but imposes assumptions about what can change and how smoothly. The former may allow differences caused by changing model choices. Neither procedure eliminates the need for field quality control. Record baseline date, configurations, current and voltage ranges, rejected measurements and reference calibration. If a dramatic feature appears only after data quality changes, test that explanation before attributing it to fluid movement.
A second numerical check for the time-lapse legend
Consider a model value that changes from 50 ohm-metres to 40 ohm-metres. Its resistivity decrease is 10 ohm-metres, or 20 percent of the baseline. Conductivity changes from 0.020 siemens per metre to 0.025 siemens per metre, an increase of 0.005 siemens per metre, or 25 percent. These deliberately simple numbers show how reciprocal quantities use different relative denominators. A legend displaying twenty percent change must identify the variable and the baseline used.
If a presentation instead displays the logarithm of conductivity, the same change is ln(0.025/0.020), approximately 0.2231 in natural-log units. It is not 0.2231 siemens per metre and not directly a percent. The expression is dimensionless because it uses a ratio. Keep absolute properties, ratios, log changes and percentages distinct. A technically correct inversion can still be misrepresented by a careless caption, and a reader can catch the error using these independent arithmetic checks.
Deviation errors become inversion errors
The forward electrical model needs electrode positions. If a borehole bends, the distance between electrodes at depth can differ from the separation measured at the ground surface. A survey based on straight vertical holes can therefore mispredict potential responses. Because inversion seeks to reduce those differences, it may alter the resistivity model to compensate for inaccurate geometry. The resulting anomaly can look geological even though its source is positional. This is a crucial connection between trajectory surveys and electrical tomography.
Imagine two boreholes separated by ten metres at the surface while a lower electrode pair is actually nine metres apart because one hole drifts toward the other. A model that still assumes ten metres uses a wrong spatial geometry for its current paths. One cannot fix that simply by multiplying every inferred resistivity by nine-tenths; the three-dimensional field and other electrode configurations may respond differently. Use surveyed station coordinates and test plausible geometric errors. The borehole deviation guide explains the necessary coordinate translation.
Borehole fluid and casing form part of the electrical system
Electrodes do not float in an abstract geometric diagram. They connect electrically through the borehole environment, which may include water with elevated ionic conductivity, casing or screen materials, grout and irregular rock walls. A fluid-filled column can form a conductive pathway unlike the surrounding formation. Metallic casing can substantially change how current is distributed. If these structures are neglected in a forward model, the inversion may redistribute their effects into neighbouring earth cells and produce a misleading near-hole pattern.
Read the well-completion record and the electrode contact details. Check whether the survey method was designed for the casing and fluid conditions present, and whether a water-column correction or explicit representation was used. A near-borehole conductor should not be called a formation fracture merely because it is drawn on the formation side of a model cell boundary. Equally, do not overfit speculative borehole structures without evidence. The right approach is to include known important effects and disclose material uncertainty for unresolved ones.
Reciprocal measurements help assess quality
In an appropriately linear, stationary medium, electrical reciprocity relates a four-electrode measurement to another configuration in which the current and potential electrode roles are exchanged. When signs and wiring are handled consistently, the transfer resistances should correspond. Disagreement can reveal noisy contacts, unstable electrodes, switching problems or changing environmental conditions. Reciprocity is a diagnostic rather than a guarantee: a shared systematic bias may survive both configurations. Still, the comparison is one of the most useful independent checks in an ERT dataset.
Suppose the original transfer resistance magnitude is 0.200 ohm and the reciprocal yields 0.212 ohm. Their difference is 0.012 ohm, about six percent of the first reading. Whether that is acceptable depends on the prescribed error model. If another pair differs by 35 percent, inspect it before inversion. Do not silently replace both values with their average and erase the discrepancy. Preserve raw measurements, signs, injection current, electrode references and the chosen criterion for excluding suspect readings.
Misfit is evidence about a fit, not proof of geology
An inversion algorithm often minimizes weighted discrepancies between measured and predicted data plus a penalty on model complexity or roughness. The reported misfit is interpreted relative to the assumed measurement errors. If those error estimates are inflated, a model can appear statistically acceptable while failing to reproduce meaningful observations. If errors are underestimated, an algorithm may chase noise and create unstable features. A sensible workflow examines the distribution of residuals, not just a single summary statistic.
Plot observed versus predicted transfer resistances or voltages, and inspect residuals by electrode pair, depth and signal strength. Systematic errors associated with one borehole may indicate geometry or contact issues. If a feature changes substantially when noisy configurations are removed, its reliability deserves caution. Data rejection should follow recorded acquisition defects or a transparent statistical rule rather than the desire for a preferred image. A low number displayed beside a tomogram cannot substitute for this evidence audit.
A repeatable four-electrode arithmetic example
For an illustrative valid electrode configuration, suppose the injected current is 0.050 ampere and the measured potential difference is 0.015 volt. The transfer resistance is 0.015 divided by 0.050, or 0.30 ohm. If the appropriate geometry factor for the stated idealized arrangement is 80 metres, the apparent resistivity is 24 ohm-metres. Doubled voltage and doubled current would preserve the resistance and apparent resistivity under linear conditions and unchanged geometry. That is a simple but powerful check.
Now imagine that the current was incorrectly transcribed as 0.005 ampere instead. The calculated resistance becomes 3 ohms and apparent resistivity 240 ohm-metres, exactly ten times larger. The error is arithmetic, not a geological discovery. Separately, a correct voltage and current cannot rescue an inappropriate geometry factor copied from an unrelated surface-array configuration. Verify recorded units and electrode positions before trusting the computed apparent resistivity. These calculations are hypothetical teaching cases, not field observations.
Regularization changes how anomalies look
Many inverse problems are underdetermined: measurements do not uniquely specify every model-cell resistivity. Smoothness regularization favours gradual transitions, while other constraints may favour sharper or more compact features. Each choice embodies assumptions about plausible geology. A smoothly connected conductive region in one model might appear as separated patches in another model that fits observations comparably well. Reporting one visual outcome as an unqualified physical fact hides this dependence.
Run a small set of justified alternative regularization values or starting models, then ask which important conclusions remain stable. If a broad conductive region persists but its thin connecting neck moves or disappears, confidence in the region is greater than confidence in the neck. Explain that distinction in the caption. A stable colour pattern does not establish salinity or contamination; it supports an electrical contrast. The next interpretive step should use independent geology or samples rather than merely increasing the contrast of the image.
The limits of interpolated pictures
A model mesh may contain many small cells while the measurement geometry effectively resolves only larger structures. Rendering or smoothing can create apparently continuous edges between cells. The visual resolution of a digital image is therefore not the same as the physical resolution of the survey. A high-resolution exported PNG does not grant additional subsurface information. Readers should look for electrode positions, sensitivity zones, measurement coverage and resolution tests rather than judging precision by pixel size.
If a map depicts an anomaly narrower than the electrode spacing in a poorly constrained region, ask what synthetic recovery test or independent observation supports that narrow geometry. It may still represent a real small object, but the inversion alone might not locate or size it robustly. A better report describes the feature at a scale that survives reasonable alternative models. When a numerical estimate of location is needed for drilling, state its uncertainty and how it was determined instead of converting line thickness on the picture directly into metres.
Boundary conditions and off-panel influences
Forward calculations require a finite numerical region with chosen boundary conditions, even though the physical electrical field extends through surrounding ground. A computational boundary placed too close to important current paths can distort predicted voltages. Nearby conductive infrastructure or water bodies may influence measurements outside the displayed survey panel. A solver can misattribute such effects to resistivity within the model if its assumptions are inadequate. Edge anomalies therefore need careful interpretation, particularly where electrode coverage is weak.
One useful sensitivity check increases the model domain while keeping important acquisition assumptions comparable. If the inferred central feature changes radically simply because distant numerical boundaries moved, the model is not robust. Check whether the survey is truly suitable for a 2D representation and whether any conductive installation close to the holes was accounted for. A dramatic anomaly at the edge should not receive the same confidence as a feature with independent sampling support and good measurement coverage.
Compare ERT with crosshole radar without equating them
Crosshole radar tomography uses electromagnetic wave travel times, attenuation and sometimes reflections to investigate interwell structure. Crosshole ERT uses electric currents and potentials. Both can respond to changes in fluid or formation properties, but they are sensitive to different mechanisms. A saline tracer may create a strong conductivity response while producing a different radar velocity or attenuation response. Agreement can be useful, yet disagreement does not automatically prove one instrument wrong.
Read the associated crosshole radar interpretation guide as a companion rather than a substitute. If two surveys show overlapping anomalies, inspect their spatial resolution and measurement timing before claiming a connected fracture or plume. Radar and ERT models may use different regularization and have different coverage patterns. A joint interpretation becomes strongest when the properties are connected by a defensible physical explanation and supported by direct geological or chemical evidence.
Samples and hydraulic tests distinguish proposed causes
Electrical response alone rarely identifies one material or chemical mechanism. Clay minerals, saline porewater, saturation, conductive minerals and temperature can change conductivity in overlapping ways. Groundwater samples, lithologic logs, temperature measurements and hydraulic tests can help distinguish them. A local chloride measurement is chemically informative but samples one place and time; the ERT image is spatially extensive but inversion-dependent. Combining the two can reduce ambiguity without pretending they have identical measurement support.
Imagine ERT shows declining resistivity after an introduced saline tracer while a nearby well independently records a rise in chloride concentration. The two observations support a tracer-related interpretation more than either alone if timing and hydraulic geometry align. Yet they may not establish the plume’s exact width or path. Another plausible conductive process could contribute. State which portions of the conclusion come from direct chemical samples and which come from the inversion. This clear division strengthens a technical report even when the conclusion remains provisional.
Transfer exercises with checked answers
Question one: resistivity decreases from 30 to 15 ohm-metres. What happens to conductivity? Question two: a plotted narrow conductive bridge cannot be recovered in the survey’s synthetic resolution test. Can it be called a confirmed continuous fracture? Question three: two measurements have identical electrode geometry, but voltage and injected current both double. What happens to their apparent resistivity under the ideal linear relationship?
Answers: conductivity changes from 1/30 to 1/15 siemens per metre, so it doubles, a 100 percent increase. The narrow bridge remains an interpretation hypothesis, not a resolved physical connection; a wider electrically anomalous region may be supported. The apparent resistivity remains unchanged because delta V divided by current does not change. These answers require distinguishing reciprocal variables, physical resolution and current-voltage ratios. They can be checked without relying on the aesthetics of the tomographic panel.
A second diagnostic test for readers
Question four: one configuration has a reciprocal mismatch of 40 percent while most others agree within four percent. What should be checked? Question five: an apparent time-lapse plume coincides with changed electrode depths. What is the next analytical step? Question six: a low-resistivity area appears only near the outside edge of the displayed inversion where sensitivity is weak. Is it safe to name a precise contaminant source from that colour alone?
Answers: inspect contact, wiring, measured current, noise and repeatability, then apply a documented quality criterion. Reconstruct actual electrode coordinates and compare predicted responses under the altered geometry before attributing the difference to groundwater movement. The edge feature cannot establish contaminant identity or precise location on its own. It needs sensitivity assessment and independent sampling or other supporting observations. Each question asks the reader to separate what an instrument recorded from what a model is allowed to claim.
Questions frequently asked about crosshole ERT
Can ERT directly measure salinity? No. It estimates electrical response, with salinity being one among several influences. Does every low-resistivity zone contain water? No; clay and other conductive materials can contribute. Can a good fit locate fractures to centimetre precision? Not by itself; resolution depends on electrode coverage, model and noise. Are 2D sections always faithful maps of the exact interwell plane? No; out-of-plane structure and modelling assumptions can affect them. Is more data always better? Additional independent, high-quality configurations may help, but repeated noisy or redundant measurements do not guarantee resolution.
What does a reciprocal check establish? It helps detect certain acquisition inconsistencies under suitable conditions, but it does not remove all systematic error. Should baseline and repeat images be inverted independently? That depends on the investigation and the assumptions behind joint or separate inversions. Can radar confirm ERT? It provides another physical constraint, not a simple one-for-one confirmation. Why preserve measured voltage tables? They allow an auditor to recheck current normalization, geometry, rejection decisions and the modelling chain.
A useful final interpretation method
Write a conclusion that begins with measured geometry and acquisition quality, then identifies what was calculated, what was inverted and which features survive reasonable sensitivity tests. State the model type, assumptions about out-of-plane variation, error estimates and resolution limitations. Describe broad electrical features before proposing specific rock types or fluid conditions. Where an interpretation is based on another sample or hydraulic test, cite that observation separately. If an apparently continuous anomaly is unresolved, disclose that rather than implying that a smooth colour field maps a continuous fracture.
The general reading principle is the same as the eduKate entailment lesson: a conclusion must follow from the evidence and assumptions, not simply resemble a convincing story. An apparent resistivity is not a directly sampled cell; a model fit is not a geological identification; conductivity is not resistivity; a two-dimensional picture is not a three-dimensional measurement. These distinctions let readers transfer technical understanding to unfamiliar instruments and topics.
Primary research and related interpretation
Primary methodological sources: US EPA, Electrical Resistivity; NERC, electrode configuration and positional sensitivity in crosshole ERT; USGS, comparison of measured and inverted ERT responses; USGS, time-lapse three-dimensional conductivity inversion. Worked figures are independently constructed educational examples, not actual field measurements.
Connected guides: borehole deviation and true vertical depth; normal resistivity logging; crosshole radar interpretation; vocabulary and differences in meaning.