What can you really see in seismic?

Module 02  ·  Vertical resolution

The wedge model and tuning thickness

Take the middle layer from module 01 and thin it sideways to nothing. Watch what happens as the top and base reflections close in on each other. The amplitude behavior over that interval is the origin of many misread amplitude maps.

The question this answers

You are handed a section with a sand in it. How thin can that sand get before you can no longer see its top and base as two separate things — and once you cannot, what does the seismic show you instead?

The setup is called a wedge model: one bed, thick at one end, thinning to zero at the other. Read left to right, the same rock gets thinner. Nothing else changes — same rocks, same wavelet, same everything.

Two things happen as the bed thins.

What happens as the bed thins

Merging. Below some thickness the two reflections combine into a single event, and the bed can no longer be measured from the section.

Tuning. Just before they merge, the bed becomes brighter than it was at any greater thickness, because the top and base reflections line up and reinforce each other. That peak is called tuning. A bright event can therefore be a thin bed rather than a thicker or higher-contrast one.

The lower two plots show both effects: one the brightening, the other the thickness picked from the section departing from the thickness in the model.

Start from

The wavelet

Display

The three rocks

The middle layer is the wedge. Its velocity also sets how thickness converts to time.

The wedge

Click or drag anywhere on a panel to move the marker

1 The wedge itself impedance  ·  pale = low, dark = high

2 The seismic response reflectivity convolved with the wavelet

What the seismic tells you, versus what is there

Both plots share the x-axis of the wedge above
Measured from the synthetic Truth (1:1) λ/4
Dominant frequency30 Hz
Wavelength in the wedge— m
λ/4 — the rule of thumb— m
λ/8 — Widess detection limit— m
Measured tuning thickness— m
Top / base reflection coefficient— / —
Marker: true thickness— m
Marker: thickness the seismic reports— m
Marker: amplitude vs a thick bed—
Verdict at the marker—

Try this — and what to watch for

  1. Find the brightest part of the wedge. The point: this one fact is what makes bright spots ambiguous. Drag the marker to the brightest column. Is it at the thick end? Compare the measured tuning thickness with λ/4. The peak sits near λ/5 rather than λ/4: for a Ricker wavelet the measured tuning thickness is about 20% thinner than the rule of thumb. Below that thickness the bed keeps thinning but the seismic gets dimmer, not thinner.
  2. Watch a thickness map break. The point: if you ever turn an isochron into a thickness map, this plot is why it can be wrong. Watch the lower plot as you drag left from the thick end. Where does the red curve peel away from the dashed 1:1 line? Right around tuning. Left of that point the reported thickness is essentially constant — it has become a property of your wavelet, not of the rock. Isochron thickness maps are meaningless below this value.
  3. Buy resolution with frequency. The point: this is what you are paying for when somebody proposes broadband reprocessing. Push peak frequency from 20 Hz to 60 Hz and watch the tuning thickness. It scales as 1/f. Tripling frequency thins the resolvable bed threefold, which is why broadband processing is worth paying for and why deep targets are so much harder.
  4. Change the rock, not the geometry. The point: it separates two things beginners merge — how loud an event is, and how well resolved it is. Switch from the encased sand to the gas sand and watch the reflection coefficients in the table. The wedge is the same shape and the tuning thickness barely moves, because tuning depends on wavelength, not on contrast. What changes is how loud the whole thing is. Brightness and resolution are separate questions, and confusing them is how a strong reflector gets mistaken for a well-resolved one.
  5. Reverse the amplitude behavior. The point: the shape of the tuning curve depends on the rocks either side of the wedge, not on the wedge alone. Edit the middle layer until its impedance sits between the layers above and below — try the "hardening downward" preset. Now both reflection coefficients have the same sign, and the wedge behaves in reverse: the two reflections cancel near λ/5 and reinforce as the bed vanishes. Same geometry, opposite amplitude behavior. Nothing about the wedge changed except the rocks on either side, which is why you cannot read a tuning curve without knowing your impedance profile.
  6. Add noise. The point: published resolution limits assume noise-free data. Set signal-to-noise to about 4 and look for the thin end. The λ/8 detection limit assumes noiseless data. With real noise you lose the thin end well before λ/8. Detectability depends on data quality as much as on physics.

If you remember nothing else

  • Resolving and detecting are different questions. Around λ/4 you stop being able to separate top from base. Down to roughly λ/8 you can still tell that something is there, if the data are clean.
  • Below tuning, amplitude works as a thickness proxy, but only just. Amplitude falls roughly linearly with thickness there, so bright can mean thick. Above tuning that relationship inverts and then breaks down.
  • A bright spot at tuning does not mean more rock. It is the same rock interfering with itself. Channels routinely look brightest along their thin edges.
  • Tuning thickness is set by the wavelet; loudness is set by the rocks. The two are independent, and only one of them is under the processor's control.
  • Because λ = V/f, resolution is poorest at depth, where velocity is high and attenuation has already removed the high frequencies.
Method, assumptions, and where to read more

Three layers, each with a velocity and a density. Acoustic impedance is their product, and the two reflection coefficients follow from R = (I₂ − I₁) / (I₂ + I₁) at the top and base of the wedge. The middle layer's velocity converts thickness to two-way time as 2h/V. Which way the wedge behaves is therefore decided by the rocks, not by a setting. Opposite-sign coefficients reach maximum amplitude near λ/5 for a Ricker wavelet. Same-sign coefficients do the reverse: they are loudest as the bed vanishes, where the two reflections coincide and add to roughly twice a single coefficient, and they are quietest near that same λ/5 separation, where the peak of one wavelet sits on the trough of the other. A same-sign wedge therefore has no tuning thickness in the usual sense, and the measured value is reported as none.

Each trace is convolved analytically, as a sum of shifted and scaled wavelets rather than on a sample grid, so reflector timing is exact and the tuning curves come out smooth instead of stair-stepped. The top of the wedge sits at 50 ms, clear of the start of the display.

Amplitude and apparent thickness are picked the way you would pick them in an interpretation package: the extremum matching the sign of the top reflection, and the time between that extremum and the one matching the base, refined by parabolic interpolation. Apparent thickness converts back to length with the same interval velocity.

Deliberately left out, because each deserves its own module: wavelet phase other than zero, offset and NMO stretch, anything 3D, transmission losses and multiples inside the bed, and lateral velocity change. This is the textbook case; real data is generally worse rather than better.

Next up

03 · Can you see the fault?

Fault throw runs into the same λ/4 limit, with vertical offset in place of bed thickness. The next module presents the same physics as a test: it hides the answer, randomizes the case, and asks whether a fault is visible at all. About a third of the cases contain no fault.

Open module 03 → All modules
← Where a trace comes from Next: can you see the fault? →