A survey aircraft passes over covered ground without touching it. It records position, clearance, attitude, time and a changing physical field. The rock stays below the trees, the till and the sandstone.
No instrument in the aircraft has seen it. The map produced afterward makes that separation easy to forget. A compact coloured high acquires edges. A conductor takes on a trend.
A response is given a name. In the Athabasca Basin of northern Saskatchewan, the name can begin to sound like a buried body waiting for a drill rig.
It is not.
An anomaly is a measurement of contrast.
That is already valuable.
It is not yet a discovery.
The instrument records a property
It is tempting to treat the sequence as a pipeline.
Survey, anomaly, target, hole, discovery.
The nouns become progressively more confident, as though the instrument had handed the rig an address.
It has not.
Geophysics begins with a physical question that can fail. Is there a conductive sheet beneath cover? Does a dense body interrupt the background?
Is a magnetic contact displaced?
Is there a resistivity break, a shallow radioelement boundary, a reflector with the expected geometry? The field procedure in the corpus insists that forward models include plausible barren alternatives.
A model that tests only the preferred orebody has not tested whether the survey is useful. The distinction runs through every method.
A magnetometer measures field variation.
A gravimeter contributes to a corrected gravity product. Electrical and induced-polarization work record current, voltage, contact condition and decay before they become apparent-property sections.
Radiometrics begin as spectra.
Seismic starts as traces in time. The geological body is inferred later.
So is its value.
An anomaly can be real, repeatable and correctly processed while its preferred explanation is wrong. A conductive overburden layer can fit a response that a thin bedrock plate also fits.
A radiometric response can be influenced by moisture, cover or clearance as well as by ground composition. A magnetic contrast says that magnetic properties change across space.
It does not say which rock, alteration pattern, depth or geometry produced the change. The ambiguity is not a defect introduced after acquisition. It is the condition of remote measurement.
Location is part of the observation
For a compact anomaly, a map coordinate is not administrative detail. It is part of the datum. Planning guidance treats navigation and altitude accuracy as an explicit aeromagnetic specification category, and the report has to say whether the stated tolerance is horizontal position, absolute elevation or sensor clearance.
Those are different quantities with different effects. Move a compact magnetic, electromagnetic or radiometric response a few metres and the planned collar moves with it. Change elevation and a gravity correction changes.
Misstate a seismic source or receiver position and the offset and common-midpoint binning change too. A row in a database is not just a number. It is a physical observation tied to a place and a geometry.
This is why quality control begins before gridding. Raw routes are inspected for gaps, turns, duplicate positions, altitude excursions, missing samples and inconsistent line names. Traverse-to-tie intersections are checked in profile form.
So are repeated passes.
An attractive shaded grid can hide a line-level error that is obvious before interpolation. A USGS aeromagnetic processing example reduced profile and tie-line differences to less than 4 nT root-mean-square. That is not an acceptance threshold.
It is an achieved check on agreement between independently acquired lines. For a broad regional question it may not matter. For a weak local response it may decide whether the colour on the map belongs to the ground or to the survey.
The unit is small.
The inference is not.
One Saskatchewan helicopter magnetic survey planned 100 m line spacing over its block. That specification describes coverage, not discovery power.
The same spacing can oversample a deep regional problem and miss a narrow response whose wavelength is shorter. Line density has no meaning without a physical question, a target geometry and an anticipated noise scale.
Processing makes assumptions consequential
The route from raw reading to target map contains legitimate work, and none of it is neutral to interpretation. Magnetic data may be diurnally corrected, have a reference field removed, be levelled, then be transformed into derivatives or analytic-signal maps.
Each version should stay identifiable.
A derivative can make a compact feature visible. It can also amplify residual line noise. Gravity proceeds through base closures, drift estimation, elevations, and free-air, latitude, terrain and Bouguer corrections.
The Bouguer anomaly is a calculated product, not a gravimeter reading. Electrical inversions need observed-against-calculated plots, residuals, model roughness, sensitivity, and recorded starting models and bounds. Seismic processing applies geometry, filtering, deconvolution, statics, velocity analysis, stacking and migration.
A reflector in time is still not a surface in depth. Each step is a claim about what should be removed, retained or made spatially coherent.
That claim can be tested.
It cannot be made invisible.
An airborne system should preserve raw and compensated electromagnetic responses, because compensation can remove a genuine weak anomaly along with aircraft noise. Radiometric products should retain individual channels, count uncertainty, altitude correction and the flight path, not only a ternary image.
Calibration pads establish an external reference for instrument response, with dedicated potassium, thorium and uranium surfaces. They do not make the field ground simple. The phrase equivalent uranium is doing necessary work.
Gamma-derived uranium is a spectral inference from decay products. It is not a laboratory concentration in a sampled rock. A map that suppresses that distinction has begun to manufacture geology before the drillhole arrives.
Depth comes from geometry
Depth is where an anomaly most readily becomes a picture of something it has not resolved. A deeper colour on an inversion is assumed to mean a deeper body.
The field geometry has to earn that extension. Ground electromagnetics makes the point plainly. Published tables give approximate investigation depths for stated coil spacings and dipole orientations. Those are depth-support ranges under stated configurations.
They are not a promise that every inversion can distinguish every body within them. Coil spacing and orientation alter the volume of ground contributing to the response.
A new line direction or a rotated loop is a new experiment, not a second look at the same answer. Electrical arrays behave the same way. A pseudosection places readings according to array geometry.
It is not a vertical log. In one square-array study, observed anisotropy changed as larger electrode spacings sampled below a shallow resistive overburden. The apparent property changed because the sampled volume changed.
Seismic carries a related warning.
Trace density is observed geometry.
Vertical resolution still depends on bandwidth, velocity and wavelet. A time section and a depth-converted section are not rival illustrations of one direct measurement. Resolution is not a decorative feature of the final image. It is the limit that tells the next test what it can honestly resolve.
A drillhole changes the support
Drilling does not make the inverse problem disappear. It changes the support of the evidence. A survey constrains a property over a volume and through a model. A core interval is material recovered from a path.
So the hole has to begin with control over a mundane and absolute fact: where the path starts. One exploration programme surveyed surface collars to the centimetre and verified initial dip and azimuth by surveying the top and bottom of in-hole drill steel.
The tolerance is project-specific.
The principle is not.
Without a known collar and orientation, no later depth has a stable spatial reference. Then the hole has to preserve what it encounters. PQ core is about 85 mm across, HQ 63.5 mm, NQ 47.6 mm.
Larger diameter retains more rock and a more legible surface for structures, alteration and fracture mapping, while demanding more torque, water, lifting capacity and consumables. A reduction in diameter can be a planned response to depth and deviation. It is not automatically evidence that the drilling failed.
Recovery has its own geometry.
In difficult ground, one reported practice kept core runs to 1.5 m or less and used shorter increments where necessary. A long core column can jam, wash away or abrade before retrieval.
Broken pieces recovered later cannot be quietly assigned to a missing interval. The run block and the field record have to keep the gap.
This is more than custody.
It is epistemology in a core tray. A core can establish lithology, mineralogy, alteration and structure at the part of the hole it actually samples. It cannot prove that the same material continues beyond the core, reaches the anomaly centre, or produces the response by itself.
It can also miss the part of a dipping, offset or narrow body that matters. A negative hole therefore has a question attached to it. Did it test the model, or did it test the ground beside the model?
The hole leaves the drawing
The target is normally drawn in three dimensions. The hole is imagined as a line that meets it. In the ground, bit, barrel, rods, feed regime, reaming-shell condition and formation steering can build angle or turn the hole.
No drawing protects against that.
The procedure is to survey inclination and azimuth at planned stations, retain measured depth and tool status, reconcile the survey with the driller’s rod tally, and acquire a final survey before the rig leaves.
One technical report recorded gyroscopic deviation measurements at 3 m intervals. That spacing is not a general requirement. It shows the necessary logic: measure a departure before it consumes the positional tolerance at the intended depth.
A longer hole, a tighter target tolerance, anisotropic rock or a known deviation history can justify closer measurement or active control. What cannot be justified is replacing the measured trajectory in a database with the planned trace.
A late survey cannot recover the path of a blocked, abandoned or collapsed hole. A smooth final offset can conceal a sharp local bend. A final depth can exceed the geological end of the useful sample.
These are not clerical complications.
They decide whether the recovered core belongs on the section where an interpreter has put it. The drill rig does not bore through uncertainty in a straight line. It has to measure its way through it.
What don’t we know?
At a covered Athabasca target, the geophysical record may support a conductor, a magnetic contrast, a density contrast, an apparent resistivity structure, an equivalent-uranium response, or some combination of them.
It may do so at a stated resolution and a stated geometry. It does not select a unique body from the possible causes. The unknown is not merely whether there is ore.
Which body produced the response?
At what depth?
With what thickness, dip, strike, conductivity, susceptibility, density or alteration pattern? Is the apparent anomaly one object or several superposed contrasts? Does the planned hole intersect the volume the data are actually sensitive to, or only the volume that looked persuasive after gridding and inversion?
Independent constraints narrow these questions.
A repeat line checks whether the response repeats. A tie line checks line-to-line agreement. A calibration pad checks instrument response. A ground traverse with changed coupling can test strike. A core hole tests material along its measured path.
A downhole survey says where that path actually went. None of them converts surviving effects into a unique history. Several bodies can fit one anomaly. Several survey models can fit one set of observations.
Several reasons can explain an empty interval: no body, the wrong depth, the wrong geometry, inadequate resolution, poor recovery, trajectory error, or a target physically present but not intersected.
The correct response is not to call every alternative equally likely. It is to state which alternative the next measurement can discriminate.
Discovery begins after contact
Geophysics is sometimes described as a way of seeing underground. The phrase is useful only while it stays qualified. It sees physical contrast through a measurement system, at finite resolution, after corrections, under model assumptions. Drilling is sometimes described as the proof.
That also needs qualification.
A hole samples a narrow measured path and can preserve, lose, or miss the evidence a model requires. The two methods are not rivals.
They are a chain of constraints with a vulnerable junction between them: the geometry of the proposed body and the geometry of the actual hole. An anomaly earns a drillhole because it makes a physical question testable. It does not earn the word discovery in advance.