From 2D lines to 3D volumes

Part 1, Foundations of Seismic

Learning objectives

Understand why the shift from 2D to 3D was a transformation in interpretation, not just an increase in data
Recognize the problems that 2D data cannot solve (sideswipe, misties, out-of-plane reflections)
Use inlines, crosslines, time slices, and arbitrary lines as complementary views of a volume
Adopt a volume-centric interpretation mindset

Through the 1970s, seismic interpretation was done almost exclusively on 2D lines, individual cross-sections recorded along a survey track, displayed side-by-side on a light table, and correlated manually. 2D lines still have uses (reconnaissance, frontier areas, long regional profiles), but for any serious interpretation of a prospect or field, they have been replaced by 3D volumes since the 1980s. The shift was not just a numerical upgrade; it changed what was even possible to interpret.

Why 2D is not enough

Three problems that plague 2D interpretation and that 3D largely solves:

Misties. Two intersecting 2D lines should show the same reflector at the same time at their intersection. In practice they rarely do: processing differences, small navigation errors, and 3D geology that projects differently onto the two line directions all cause discrepancies in time or depth. Reconciling misties used to take a large share of an interpreter's time.
Sideswipe. A 2D line records energy not only from directly below the line but also from geology off to the side. A fault plane a few hundred metres off the line that dips toward it returns a coherent reflection, and the section places it directly beneath the line. 3D acquisition samples the surface in both directions, so migration can put that energy back where it came from.
Continuity assessment. A 2D line tells you what is under the line. It tells you nothing about whether a feature extends 100 m or 10 km perpendicular to the line. Interpreters extrapolated, and got it wrong often.

A 3D survey samples the surface in both directions, typically on a grid of 12.5 to 25 m bins, densely enough that each subsurface point is recorded from many directions and 3D migration can place each reflection where it belongs. Structural geometry and stratigraphic continuity can then be read from the volume directly, in a way a grid of 2D lines cannot match.

Here is sideswipe at work. Figure 1.6 puts a fault beside a 2D line, dipping toward it. Find where its echo is born, slide the fault until its echo lands on a real bed, steepen it until only its top edge answers, and then record the same earth in 3D.

With the fault's base 200 m off the line and a dip of 60°, the ray that returns to the line leaves it 60° from vertical and meets the fault 670 m to the side, at 387 m depth. The line records only the time, 0.62 s, and because the fault strikes along the line the event is flat and continuous, so the 2D section puts it 773 m straight down, where a bed would be. Move the fault's base to 690 m and its echo lands on the bed at 0.96 s: along the fault's length the bed's peak falls to 38 % of its amplitude, a change on the section with no cause beneath the line. A line crossing the fault would record it with its dip in the section, 2D migration there would move the echo toward where it belongs, and the two lines would then disagree where they cross: one way misties arise. Recorded in 3D, the surface is sampled across the line as well as along it, migration returns the energy to where it was born, and the inline beneath the old line keeps only the bed.

Once you have a 3D volume, the interpretation interface changes. You no longer flip through 2D cross-sections imagining what lies between them. You slice the volume interactively, picking the cross-section or map view that best illuminates the feature you are studying. Every interpreter uses four standard views.

The four interpretation views

Inline, a vertical cross-section parallel to one horizontal axis. Usually the original acquisition line direction. Good for structural interpretation along the primary axis.
Crossline, a vertical cross-section parallel to the other horizontal axis. Perpendicular to inlines. Shows a different "face" of the same geology: a fault that strikes parallel to the inlines runs along an inline and is hard to see there, while a crossline cuts it at right angles and shows its true dip.
Time slice (or depth slice in depth-migrated data), a horizontal cut at a fixed time/depth. Map-view of the subsurface at that level. Invaluable for pattern recognition: channels, faults, and other geologic features often stand out on time slices in ways that are invisible on vertical sections.
Arbitrary line, a cross-section along a user-drawn polyline that does not need to follow inline or crossline directions. Essential for aligning a section perpendicular to a dipping structure or along the strike of a fault.

Section 1.0's cube viewer exposes the first three of these directly. An interpreter flips between them constantly, building a 3D mental model from the three orthogonal 2D views. Arbitrary lines and horizon-slicing tools extend that basic set and make specific workflows (fault interpretation, channel mapping) much faster.

The volume-centric mindset

Working in 3D is not about viewing more cross-sections. It is about treating the entire volume as a single object and asking questions of it:

Is this channel continuous? (Slice at the channel's time; the answer is a picture.)
Does this fault cut higher up the section? (Follow it through a sequence of time slices.)
How does this reservoir thickness vary across the field? (Extract the time difference between top and base picks and map it.)
Where in the volume does a particular seismic signature appear? (Compute a volume attribute; threshold it; the result is a 3D object.)

None of these questions is naturally answered by a grid of 2D lines. All of them are naturally answered by a 3D volume with the right tools. The rest of this textbook assumes the volume-centric mindset: every question starts with "given this volume, what do I slice / compute / threshold to see X?"

One practical consequence. 3D data is much larger than 2D data: a post-stack volume commonly runs from gigabytes to hundreds of gigabytes, and the prestack data behind it to terabytes. The software, hardware, and workflow of 3D interpretation all reflect this scale. It is also why this book works with teaching-sized subsets and synthetic volumes: a full survey would not fit in a browser. The cube in Section 1.0 is one such subset, a window from the real F3 Netherlands survey.

References

Brown, A. R. (2011). Interpretation of Three-Dimensional Seismic Data (7th ed.). AAPG Memoir 42 / SEG IG13.
Bacon, M., Simm, R., & Redshaw, T. (2003). 3-D Seismic Interpretation. Cambridge University Press.
Yilmaz, Ö. (2001). Seismic Data Analysis (2 vols.). Society of Exploration Geophysicists.
Sheriff, R. E., & Geldart, L. P. (1995). Exploration Seismology (2nd ed.). Cambridge University Press.