Start with the decision, not the colour
A borehole optical televiewer image is a colour photograph of the well wall unrolled into a rectangular strip. It is not a geological map and not a photograph of all the rock surrounding the hole. To read an optical televiewer, first identify its depth reference, azimuth convention, image quality and borehole diameter. Then identify a feature on the image, ask whether it represents intact rock, a natural discontinuity, a drilling mark or an artefact, and only then calculate orientation. This guide explains how to turn colour, brightness, shapes and sinusoids into testable descriptions rather than guesses.
The practical question is often whether a dark curved line represents a fracture that could matter to groundwater movement, engineering stability or geological interpretation. A photograph can establish that a line is present at the borehole wall, but cannot by itself prove that the line conducts water, extends far into the formation or is connected to another fracture. A reliable description therefore keeps separate what is visible, what follows geometrically and what still requires a hydraulic measurement or comparison with core.
The word translate matters here. The instrument measures reflected light and assigns recorded pixels to positions around a cylinder. An interpreter translates those pixels into words such as bedding, fracture, alteration band and tool shadow. Each translation adds a hypothesis. A useful description retains the chain between the measurement and the claim, much as careful language translation preserves context and register instead of replacing each word mechanically. Readers who understand that distinction make stronger decisions even when the final image appears visually obvious.
How the instrument makes a 360-degree picture
Most optical televiewers illuminate the borehole wall and image it with a camera and a conical mirror. As the sonde moves vertically, the tool records successive circular views and assembles them into a continuous image. Each circumference is then unrolled into a horizontal row, and rows are arranged by measured depth. In the resulting panel, the vertical direction is distance down the hole and the horizontal direction is compass position around the wall. The image is a cylindrical surface projection, not a cross-section through the entire rock mass. A visually impressive panel cannot overcome poor illumination or an incorrectly measured tool position.
Imagine cutting a paper tube along one longitudinal seam and flattening it on a table. A vertical scratch on the tube becomes almost vertical on the flat sheet; a ring around the tube becomes a horizontal line. A plane cutting through the tube obliquely traces a curved intersection, because its height varies as you walk around the circumference. Those elementary geometric observations explain why interpreters search for sinusoidal features in oriented televiewer panels. They also explain why a single line that appears to slope diagonally across one small part of the image should not be assigned a dip until its continuation around the full circumference is examined.
Orientation comes from a separate instrument system, normally magnetic heading, tool inclination and sometimes gyroscopic data. The camera records relative positions around the tool; navigation data connect them to a geographic reference. A display may begin at north and move clockwise through east, south and west, but software can place its split elsewhere or reverse its displayed axis. The legend, not intuition, decides. A north-arrow annotation is especially important near magnetic rocks, steel casing, power cables or other sources of magnetic interference, because a plausible colour image may coexist with an unreliable azimuth.
The five questions to ask before naming a feature
First establish the borehole type: open hole or cased, air-filled or water-filled, recently drilled or long standing. A transparent fluid may reveal detail that opaque fluid conceals. Second confirm the reference for depth: ground level, casing collar, measuring point or another datum. A fracture recorded at 42 metres below collar cannot safely be matched to core labelled 42 metres below ground if the collar stands two metres above it. Third inspect the pixel quality and the degree of wall coverage. A feature that disappears behind a dark shadow is not necessarily absent on that side of the hole.
Fourth check how the tool is centred. A probe that sits closer to one side can illuminate that side differently, enlarge it geometrically and leave a poorly exposed region opposite. Oval, enlarged or rough holes can distort apparent patterns even when camera optics are behaving correctly. Compare the image against an independent diameter or caliper log when available. Fifth confirm the image-display processing: contrast stretching, colour balancing, stitching, depth registration and seam placement. An enhanced picture can be excellent for inspection but dangerous when readers confuse software display choices with an observed physical difference in rock.
A well-written interpretation records these conditions before making claims. For example: ‘Continuous optical coverage from 25.4 to 46.0 m below top of casing; azimuth clockwise from true north after correction; minor image shadowing on the east-facing wall; local borehole enlargement near 38 m.’ That statement does not sound spectacular, but it tells another reader which observations are reproducible and which intervals deserve caution. Good scientific vocabulary works similarly: words such as observed, inferred, possible and unconfirmed are not stylistic decorations; they communicate the strength of the evidence.
Translate sinusoidal shapes into fracture geometry
A planar feature crossing a circular borehole wall produces a sinusoid on the unrolled picture. The distance between its highest and lowest points on the vertical display is related to plane dip and borehole diameter. For an ideal vertical cylindrical borehole of diameter D, the peak-to-trough depth excursion H is approximately D times tan of the dip angle measured from horizontal. Consequently dip equals arctan of H divided by D. This is a geometric model for a planar intersection, not a universal formula to apply blindly to deviated wells, arbitrary scaling or incomplete traces.
Worked example: suppose the measured hole diameter is 0.100 m and a confidently traced bedding sinusoid spans 0.060 m vertically from highest to lowest crossing. The ratio H/D is 0.060/0.100 = 0.60, dimensionless. Its geometric dip is arctan(0.60), approximately 31.0 degrees from horizontal. A common mistake is to divide 0.060 m by the borehole radius of 0.050 m and obtain an inflated dip; the full peak-to-trough excursion corresponds to the full diameter in this simplified projection. Always state which feature of the sinusoid was measured and in what units.
The direction of greatest downward intersection on an oriented image gives the apparent down-dip azimuth after accounting for the display convention and borehole deviation. If the deepest part lies beneath the panel’s southeast direction, the plane dips toward approximately southeast, provided the panel is truly azimuth-corrected. But if the borehole is inclined, the observed sinusoid reflects the relative angle between plane and borehole axis. Structural software should correct into geographic coordinates using measured borehole trajectory. That correction is not optional when borehole inclination is substantial or structural orientations will guide engineering decisions.
A second geometric check rather than trusting the first read
Every useful worked example needs an independent check. In the 0.100 m diameter example, a 31-degree plane predicts D times tan(31 degrees), or approximately 0.100 times 0.601 = 0.0601 m of peak-to-trough depth excursion. The predicted 60.1 mm matches the observed 60 mm to rounding. This reverse calculation checks the algebra, unit consistency and dip convention. It does not establish whether the visible feature really is planar, because that question depends on image morphology and corroborating evidence.
Now consider a second hypothetical sinusoid with the same hole diameter and a 0.150 m excursion. The ratio is 1.50 and arctan(1.50) is about 56.3 degrees. The steeper structure generates a larger vertical swing on the unrolled cylinder. If an illustrator stretches the depth axis twofold while leaving the azimuth width unchanged, the feature looks even steeper to the eye without changing its measured dip. This is why visual steepness should never substitute for explicit use of depth coordinates. Capture the endpoints in physical units rather than pixel counts unless the pixel calibration is known.
A third self-check asks whether the inferred feature closes smoothly as it crosses the unrolled seam. A genuine continuous planar intersection should meet itself after 360 degrees, within image resolution and geological irregularity. If it abruptly ends at the seam, first test whether the display has been split at a different azimuth or whether separate logging passes are stitched out of register. Only after processing artefacts are excluded should the discontinuity itself be assigned geological meaning. More generally, the shape predicts the geometry, but the geometry does not prove the mechanism by which the feature formed.
Read actual colour without turning it into mineral identification
Colour can carry useful geological information. Variations may indicate staining, weathering, veins, oxidation, lithological changes or borehole-wall coatings. Yet camera white balance, illumination distance, turbidity and image processing can all change apparent colour. A dark line might be an open shadowed aperture; it could equally be a narrow dark mineral band or a strip made dark by insufficient illumination. The safe description starts with optical appearance: ‘dark continuous band approximately 4 mm wide on the wall image’, not ‘highly conductive open fracture’, unless independent measurements support the latter.
Brightness differences should be interpreted relatively within the same acquisition conditions. A pale band in a darker rock may correspond to a vein, a different mineral assemblage or a polished drill mark. Ask whether its edges remain consistent all around the cylinder, whether it crosses other structures, and whether the core or cuttings show a corresponding material. A colour photograph cannot yield a laboratory mineral analysis from its pixels alone. Cross-checking a physical sample is particularly important where iron staining makes several geologically distinct surfaces look similar.
Consider a hypothetical interval where a red-brown band appears from 18.20 to 18.32 m, while nearby core is logged as oxidised fracture fill. The agreement supports a plausible staining interpretation. It does not show the timing of oxidation, hydraulic transmissivity or the persistence of the band beyond the borehole. Another sample could reveal clay-rich gouge rather than an open fracture. A defensible report would say ‘reddish band consistent with oxidised or clay-bearing material; composition unconfirmed without sampling’ and identify the next observation that could discriminate between those possibilities.
Distinguish joints, bedding, veins and drilling artefacts
Bedding contacts often form laterally persistent surfaces in layered rock and can produce repeated families of similarly oriented sinusoids. Joint sets may cut across bedding at a different orientation, and veins may show colour or reflectance contrast with the host rock. Fault-related surfaces may exhibit roughness, infill or multiple close traces. These are possibilities, not labels to apply solely on the basis of orientation. Structural classification improves when the same intervals can be matched to core, formation descriptions, fracture aperture observations and adjacent boreholes.
Tool artefacts frequently mimic geology. Drill-bit scratches, axial gouges and vertical scoring may form stripes almost parallel to the depth axis. Suspended particles in water can create irregular transient marks that change between repeated passes. A moving cable can cast shadows, while reflective surfaces can be overexposed and apparently featureless. The first diagnostic test is repeatability: does the mark remain at the same depth and azimuth during a second logging pass, after accounting for calibration? If not, investigate the imaging chain before assigning a geological cause.
A convenient diagnostic table has three columns: observation, alternative explanation and discriminating check. A dark curved trace might be a fracture, mineral seam or lighting defect; check repeat imaging, core and acoustic televiewer data. Parallel vertical marks might be drilling scratches or near-vertical fractures; check their continuity, wall geometry and formation structure. Colour patches might be oxidation or camera white-balance drift; check samples and nearby illumination consistency. Explicit alternatives reduce the risk that the first compelling visual pattern becomes an unfalsifiable story.
Fracture counts need a denominator and a visibility rule
Reports often include a fracture frequency, but the number is meaningless without the inspected interval length, detection criteria and image completeness. Suppose an idealised continuous OTV record from 20 to 35 m contains twelve accepted natural discontinuities, excluding obvious drilling marks and duplicate traces. The reported frequency is 12 divided by 15 m, or 0.80 interpreted features per metre. This is not the rock’s universal fracture density: sub-resolution apertures, image shadows and orientation sampling biases can make actual discontinuities invisible.
Suppose 3 m of that interval has opaque fluid or severe shadows. Dividing twelve by the full 15 m yields 0.80 m⁻¹, but an image-coverage-adjusted exploratory figure would be twelve divided by 12 m of readable interval, or 1.0 m⁻¹, with an explicit warning that the missing 3 m might have a different fracture population. Neither number is automatically preferable; the report should state what denominator corresponds to the question being asked. A rigorous calculation never silently changes the denominator between wells just to obtain an apparently comparable statistic.
Feature-selection rules matter as much as arithmetic. Does the count include lithological boundaries? Are closely spaced parallel traces counted independently? Are partly obscured sinusoids counted only if at least half their circumference is visible? Can an interpreter distinguish fractures induced by drilling from natural structures? Write the inclusion rules before comparing depth intervals or wells. If two reviewers identify different numbers of candidate fractures, a disagreement log is more informative than an unexplained average. Agreement should be assessed on common depth coordinates and common definitions, not merely a similar total count.
Understand the difference between orientation and permeability
An optical image can suggest which discontinuities may be open at the borehole wall. It cannot measure hydraulic conductivity, pressure gradient or connected transmissivity. A striking fracture with visible aperture may terminate a short distance away, be clay-sealed beyond the wall or sit outside the main flow path. Conversely a narrow fracture almost invisible in poor imaging conditions may carry significant water where it connects to a larger network. Good hydrogeological interpretation therefore distinguishes wall-scale appearance, geometrical continuity, hydraulic connection and measured flow. These are different propositions.
A useful integration combines OTV with a heat-pulse flowmeter, fluid-temperature and conductivity logs, packer tests, and, when available, borehole or surface geophysics. Suppose a conspicuous sinusoid appears at 44.1 m but an independently measured vertical flowmeter profile shows a change at 45.0 m instead. Do not shift the feature’s depth to make the datasets agree. First check depth datums, cable stretch and logging direction; then consider separate fractures or flow caused by other intervals. Matching a beautiful image to an unverified hydraulic signal can conceal the real flow network.
The companion guide to heat-pulse flowmeter logs explains why vertical flow measurements and zone transmissivity require their own logic. The fluid temperature and conductivity guide shows how inflow clues can support but not settle an interpretation. The word consistent is useful here: a flow anomaly can be consistent with a fracture image without being proof that the visible surface is the sole conduit.
Compare optical and acoustic televiewers without assuming they are interchangeable
An optical televiewer depicts reflected visible light and usually gives colour and texture when the borehole fluid is sufficiently clear. An acoustic televiewer sends ultrasonic pulses and displays return amplitudes and travel times; it commonly works in water- or mud-filled holes whose turbidity would defeat a camera. Because the physical measurements differ, corresponding patterns can strengthen an interpretation. An acoustic travel-time enlargement and a photograph of a broken wall may together support a borehole-geometry observation, while contrasting amplitude and colour patterns may reveal separate material and lighting effects.
The optical instrument may make a vein or weathered film conspicuous through colour even where the acoustic contrast is subtle. The acoustic instrument may reveal geometry across a turbid interval that optical imaging cannot see. Neither instrument alone establishes mineralogy or groundwater flow. An aligned comparison should use matching depth datums, compatible azimuth references, corrected borehole trajectory and explicit depth uncertainty. Reviewers should expect some features to be visible in one modality and weak in another. That can be diagnostic rather than evidence that one log is faulty.
Read the published acoustic televiewer interpretation guide as the measurement counterpart to this optical guide. The two articles have a different reader task: one interprets camera colour, lighting and exposed wall textures; the other interprets acoustic travel-time geometry and ultrasound amplitude. Comparing them teaches a general scientific lesson: different instruments can draw apparently similar pictures while measuring genuinely different physical quantities. Strong interpretation identifies the quantity before decoding the image.
Worked borehole interval: build an evidence log
Consider a fictional 120-mm-diameter borehole with good optical coverage from 30.0 to 33.0 m, oriented clockwise from true north and referenced to the top of casing. A dark feature crosses the circumference near 31.42 m, and its peak-to-trough depth excursion is 90 mm. The ratio H/D is 90/120 = 0.75, with the same length units cancelling. The ideal vertical-hole dip estimate is arctan(0.75), approximately 36.9 degrees. Assume the low point occurs near a corrected azimuth of 125 degrees: the plane would be described as dipping approximately 37 degrees toward 125 degrees, subject to the stated geometry assumptions.
Check the calculation in reverse: tan(36.9 degrees) is about 0.750, and 0.120 m times 0.750 gives 0.090 m, or 90 mm. The numerical geometry is internally consistent. Next ask whether the mark really closes smoothly around 360 degrees, whether the shadowed section is small enough to interpolate responsibly, and whether another plausible trace overlaps it. If the deepest point is uncertain by 15 degrees of azimuth because of partial obscuration, the report should retain that directional uncertainty instead of announcing 125.0 degrees. False precision is a communication error as well as a scientific one.
Now introduce conflicting evidence: nearby core contains a mineralised seam at 31.45 m, and no clear inflow anomaly is present during ambient flow logging. The refined statement is ‘a probable mineralised planar discontinuity with image-derived orientation near 37 degrees toward 125 degrees; measured aperture and hydraulic activity unconfirmed’. It would be misleading to call it a major open groundwater fracture. If subsequent packer testing detects transmissivity in the interval, that evidence can strengthen the hydraulic interpretation, but still requires care because the test integrates the packed section rather than a single visible plane.
A second worked case: when two images disagree
Suppose a logging company supplies two optical passes through the same 10-m interval. The first shows a sharp brown band at 56.8 m; the second shows it at 57.1 m. A 0.3-m difference is too large to dismiss if a construction design depends on an exact fracture elevation. Begin with the measured-depth chain: logging speed, encoder calibration, tool stretch, water level, surface reference and whether the images were sampled or interpolated onto an identical depth grid. Ask whether the feature is identifiable by several accompanying landmarks instead of forcing its best-looking pixel to match.
A practical registration can use multiple stable features along the hole. If recognizable lithological contacts near 52, 57 and 61 m all appear systematically 0.30 m lower in one pass, a consistent datum or encoder offset is plausible. If the offset varies with depth, cable stretch or processing resampling is a stronger possibility. Avoid correcting one trace locally until the pattern of discrepancy is understood. Record both original coordinates and the basis of any correction. Quietly modifying a depth can make later core matching appear convincing while hiding the original uncertainty.
Suppose the same 0.3-m offset also appears in a caliper log taken with the second optical pass. That observation supports an acquisition-depth explanation but not the precise cause. Before converting to construction coordinates, an independently surveyed top-of-casing elevation and a verified logging datum are needed. The final report should provide feature coordinates in a clearly named system, preserve the raw pass identifiers and describe residual alignment uncertainty. This demonstrates why scientific translation must retain provenance: moving from instrument numbers to an engineering instruction is a sequence of justified transformations.
Image processing can help, and can also manufacture confidence
Contrast enhancement may reveal details in a dark interval, but it also amplifies sensor noise, reflected particles and seam artefacts. Local histogram equalisation can give neighbouring rock patches dramatically different apparent brightness even when their material properties are similar. Colour balancing may improve continuity across illumination changes while weakening evidence for genuine staining. Therefore preserve or provide access to original acquisition images and note substantial processing operations in the method description. An attractive enhanced image is not a substitute for traceable raw measurements.
Orientation correction likewise requires care. If tool heading rotates while the tool is winched upward, software repositions pixels to a fixed north reference. This is appropriate if heading measurements are accurate, but magnetic disturbance or incomplete motion correction can smear features across azimuth. An artificial sinusoid may appear where rows were distorted by changing relative orientation. Check stable wall marks, trajectory information and repeated passes. Where correction quality is poor, it is better to report a confidently observed feature depth with uncertain dip direction than to produce a spuriously precise orientation table.
Stitching requires interpretation across a discontinuity created by unfolding the cylinder. A feature intersecting the north-facing seam appears divided between the left and right edges of the panel, yet the physical trace is continuous. Students sometimes count the two halves as separate joints. Another error is averaging two separate sinusoids because their fragments happen to meet at the seam. The solution is to imagine wrapping the panel back around a cylinder and to check physical continuity at the same depth. This simple spatial exercise improves both interpretation and communication.
From visible evidence to defensible report language
The report should have a hierarchy of statements. At the first level, describe reproducible observation: ‘light grey rock with a thin dark arcuate trace between 63.14 and 63.19 m’. At the second, describe geometry: ‘one apparently planar feature with an estimated corrected dip’. At the third, cautiously interpret geology: ‘possibly a natural joint or mineralised fracture’. At the fourth, describe implications only where independent evidence supports them: ‘may coincide with the flow response recorded during packer testing’. A reader should never have to guess whether a sentence reports a pixel, a calculation or an engineering judgement.
Vocabulary strengthens that distinction. Terms such as apparent aperture, open fracture, stained interface, fracture trace, hydraulic connection and transmissivity are not interchangeable. An apparent aperture is an image-based width or gap requiring calibration and viewing assumptions. Hydraulic aperture is a flow-equivalent parameter derived from a hydraulic model. Transmissivity describes integrated ability to transmit water under a hydraulic gradient. These concepts cannot be inferred from one another by substituting a synonym. In the semantic feature guide, the same principle appears in language: a small change in meaning can change the truth of an entire sentence.
An effective summary table records feature identifier, measured depth, image-quality grade, observed morphology, interpretation confidence, dip and direction where valid, supporting logs, conflicting evidence and recommended follow-up. Do not hide uncertainty in a footnote: show it next to the conclusion it qualifies. A decision maker may need only three sentences, but those sentences should be backed by an auditable record. That is why careful interpretation is often more concise than speculation; separating facts from assumptions reduces the amount of corrective explanation later.
Practical limitations and questions readers ask
Can optical televiewers see through opaque water? Usually not well enough for detailed wall imaging. Optical imaging depends on adequate light transmission and return; muddy or turbid fluid limits useful contrast. In suitable fluid an acoustic televiewer may be a better choice, but it produces different measurements. Can an optical televiewer see behind steel casing? No: it photographs the accessible internal surface, not the rock through an opaque casing. Does a dark fracture always carry groundwater? No, since optical appearance does not establish connectivity, hydraulic gradient or transmissivity.
Can dip be calculated by measuring the apparent slope with a ruler on a screenshot? Not reliably. The screenshot may have rescaled axes, the hole may deviate, and orientation may not be corrected. Use real depth excursion, measured diameter, azimuth convention and borehole trajectory. Can colour determine mineral composition? Colour can suggest hypotheses, but controlled sampling and appropriate laboratory methods are needed for a defensible mineral identification. Can two boreholes be combined to infer a fracture plane? Potentially, but correlation across holes requires geometry, lithology and uncertainty checks; similar-looking traces do not prove they are the same surface.
What if a feature appears only on one imaging pass? Investigate repeatability, lighting, water clarity, acquisition settings and instrument position before declaring that the rock changed. What does missing image coverage mean? It means the corresponding geology is unobserved, not intact. Why retain a quality flag in the final log? Because downstream interpreters otherwise convert absence of visible features into evidence of absence. The safest routine is to report where the image is reliable, what can be calculated from it and which scientific or practical conclusion remains unresolved.
Independent transfer exercises and answer key
Exercise one: an ideal vertical borehole is 150 mm wide. A smooth feature traces a peak-to-trough depth excursion of 75 mm. Find the simplified dip from horizontal. Exercise two: an interval from 12 to 32 m contains 18 accepted discontinuities, but 5 m of image is unreadable. Calculate both a full-interval frequency and a readable-interval frequency, explaining the denominator. Exercise three: an optical image shows a bright vertical stripe only during one logging pass. List at least two plausible non-geological explanations and a useful check.
Answers: for exercise one, H/D = 75/150 = 0.50 and dip is arctan(0.50), about 26.6 degrees. Reverse check: 150 mm times tan(26.6 degrees) is approximately 75 mm. For exercise two, the full 20-m interval yields 18/20 = 0.90 features per metre. The 15-m readable portion yields 18/15 = 1.20 features per readable metre. Neither figure should be quoted without its denominator and the note that unreadable sections may differ. For exercise three, possible causes include tool reflection, a lighting artefact or suspended matter; compare repeat passes, stable wall landmarks and camera conditions.
Exercise four: an image-derived dip estimate is 40 degrees toward 080 degrees, but magnetic interference is suspected and the borehole inclination log is missing. How should the feature be reported? Answer: retain the measured image trace and depth, describe the dip as provisional under the uncorrected reference model, and withhold reliable geographic orientation until the navigation data are checked. This answer requires no sophisticated computation. It requires resisting an unjustified inference, which is often the hardest part of technical interpretation. In every exercise, the distinction between measurable, inferred and unknown is the central transferable skill.
An integrated reading workflow
Begin with a problem statement: identify bedding orientations, map candidate fractures or inspect borehole wall condition. Then check acquisition geometry, top-of-casing datum, diameter, fluid transparency, camera exposure, azimuth accuracy and depth continuity. Establish readable intervals before searching for interesting shapes. In those intervals, mark repeatable features without prematurely classifying them. For each planar candidate trace, follow the full circumference, measure depth excursion and direction using corrected coordinates, and note whether an ideal-cylinder approximation is sufficient or full trajectory correction is required.
Next compare suspected discontinuities with core, caliper, acoustic televiewer, fluid and hydraulic logs. Keep visual aperture separate from hydraulic transmissivity. Label contradictions explicitly and revisit depth registration before forcing a match. Retain raw and processed image references, uncertainties and feature-selection rules. Produce a concise interpretation table alongside representative images with scale bars and clearly labelled north and depth axes. Where a conclusion would trigger drilling, grouting or other costly action, define the independent measurement or field check that would change the decision. The goal is not to extract as many labels as possible from a colourful image; it is to make a few useful claims with traceable support.
For broader reading, see the borehole caliper guide on wall geometry and the normal-resistivity guide on a distinct electrical measurement. The context stack guide and word-for-word translation guide develop the wider skill of preserving evidence, setting and intended meaning. In science and in language, a reliable translation is an accountable transformation, not a visual or verbal resemblance.
Turn competing explanations into a measurable decision
A useful decision matrix separates three explanations for every high-consequence line. In a hypothetical borehole a dark sinusoid near 70.8 m could indicate an open fracture, a narrow mineralised seam, or a deep tool shadow. The fracture hypothesis predicts a continuous trace that follows a plane, plausible local aperture and perhaps a corresponding hydraulic signal. The mineral seam hypothesis predicts a persistent coloured or reflective band, often supported by core mineral observations, but not necessarily flow. The shadow hypothesis predicts sensitivity to lighting, camera centring or repeat-pass conditions. The image alone supports the existence of dark pixels, not their preferred explanation. Record what independent test each hypothesis would fail, and choose that test before changing drilling or grouting plans.
Suppose that the original photograph shows a 6-mm apparent dark band, while repeat-pass exposure adjustment makes the dark region appear only 3 mm wide. If an interpreter reports a 50 percent reduction in ‘fracture aperture’, they are confusing colour thresholding with physical width. The difference of 3 mm divided by the original displayed 6 mm is indeed 0.50, but the denominator is an image-derived width, not a measured flow aperture. The appropriate response is to state that the apparent visual width changed under processing and that aperture remains uncertain. In this setting a good numerical calculation can still underpin a bad physical conclusion when the defined quantity is wrong.
Now suppose a hydraulic test isolates a 2.0-m packed interval encompassing that trace and yields an estimated transmissivity of 3.0 × 10⁻⁵ square metres per second. The test supports water movement somewhere in the isolated interval under its test assumptions. It does not allocate all transmissivity to the 6-mm dark feature. A second unobserved thin discontinuity might also carry flow. To infer contribution by individual fractures, additional isolated measurements, flowmeter interpretation or appropriately constrained models are needed. The word linked should be reserved for evidence stronger than mere proximity. Accurate translation from imaging to hydrogeology requires that measurement support and spatial scale travel with every derived claim.
A final independent check follows the entire chain from measured depth to recommendation. Confirm the casing datum, audit the image-quality log, verify that the feature is repeatable, inspect the geometry against borehole diameter, and compare the independently acquired flow and core records. If any link is missing, report the result as a bounded hypothesis rather than as a borehole construction instruction. It is better to identify two candidate conductive zones for focused investigation than to announce one ‘proven water-bearing fault’ from a colour band. The same discipline applies when translating between languages: a word that seems equivalent in isolation may change its meaning once the surrounding evidence, audience and purpose are restored.
Primary methodological references
US EPA, Borehole Optical Televiewer (OTV).
US EPA, Physical and Imaging Borehole Methods.