From Raw Traces to an Exploration Well: Seismic Processing and Interpretation
Key takeaways
- Raw data cannot be read directly: processing removes noise, corrects times and moves reflections to their true positions.
- NMO correction removes the delay of a reflection with offset: t(x) = √(t0² + x² ÷ V²). In our example the delay is 281 ms.
- Migration moves dipping reflections to their true location and sharpens fault edges.
- The interpreter ties seismic to wells with a synthetic seismogram, then maps horizons and faults and converts maps from time to depth.
- Geological chance of success is the product of the trap, reservoir, seal and charge probabilities. Example: 0.7 × 0.8 × 0.6 × 0.5 ≈ 17%.
Educational content only: procedures and operating values vary between companies and sites. Always follow your company’s approved procedures, your supervisor’s instructions and the permits in force at your site.
Two stages: processing, then interpretation
When the survey ends, a huge volume of traces arrives at the processing centre, millions of them for a single 3D survey. The processing team turns them into a clean image of the layers. Then the interpretation team, geophysicists and geologists, reads the image, looks for traps and proposes where to drill.
Main processing steps
- Geometry: linking every trace to the true coordinates of its source and receiver. Any mistake here spoils everything after.
- Noise attenuation: reducing ground roll, wind noise and bad traces.
- Static corrections: removing delays in the loose near-surface layer and elevation differences between points.
- Deconvolution: compressing the wavelet so it is shorter and sharper, improving resolution.
- Velocity analysis and NMO correction: finding the velocity that flattens each reflection in every CMP.
- Stacking: summing the corrected traces of each CMP into one strong trace.
- Migration: moving dipping reflections to their true position, sharpening fault edges and collapsing diffractions.
NMO correction in numbers
Within one CMP, the trace with the larger offset records the reflection later, because its path is longer. The delay follows a hyperbola:
t(x) = √(t0² + x² ÷ V²)
where t0 is the zero-offset time, x the offset and V the NMO velocity.
A reflection at t0 = 1.0 s, velocity V = 2,500 m/s, offset x = 2,000 m:
x ÷ V = 2,000 ÷ 2,500 = 0.8 s
t = √(1.0² + 0.8²) = √1.64 ≈ 1.281 s
NMO = 1.281 − 1.000 = 0.281 s = 281 ms
With the right velocity the reflection becomes flat on all traces and stacks strongly. With the wrong velocity it stays curved and weakens after stacking. That is why the "best stacking velocity" also tells us about velocities in the layers.
Tying seismic to the well
Seismic is in time and wells are in depth. To tie them we build a synthetic seismogram from the well logs: the sonic log gives velocity and the density log gives ρ, so impedance and the reflection coefficient are computed at every depth and convolved with a wavelet like the survey's. This synthetic is placed beside the real seismic at the well location, showing which reflection is the reservoir top and which is the seal. This links directly to the lesson on log interpretation and well correlation.
Interpretation: from image to trap
- Horizon picking: following the key reflection, such as the reservoir top, across every section to build a time map.
- Fault mapping: marking breaks in the reflections and drawing fault surfaces. A fault can be a barrier that traps oil, or a pathway that lets it escape.
- Depth conversion: using a velocity model calibrated with wells to turn the time map into a depth map.
- Trap definition: looking for an anticline or an uplifted fault block with closure, an area higher than its surroundings where oil can collect. There are also stratigraphic traps where the rock type changes.
- Attributes: reflection strength and its change with offset (AVO) may indicate gas, with the caution seen in the first lesson.
Estimating the chance of success
Before an exploration well, the team estimates the geological chance of success. For an accumulation to exist, every element must be present together: trap, reservoir, seal, and source rock with a migration path (charge). So the independent probability of each element is multiplied:
Pg = P(trap) × P(reservoir) × P(seal) × P(charge)
Pg = 0.7 × 0.8 × 0.6 × 0.5 = 0.168 ≈ 17%
Each element looks good alone, yet the overall chance is about 1 in 6. That is why exploration wells are studied carefully and often drilled within a programme of several targets. The better the seismic data, the better the estimate of the trap probability.
Takeaway for the engineer
Processing builds the image; interpretation reads it. Velocity is the thread that links them: it corrects NMO, converts time to depth and ties seismic to wells. And the final decision to drill is not a certainty but a studied probability.
Test your understanding
Related lessons
- Land seismic acquisition: vibroseis and fold
- Seismic reflection basics
- Log interpretation & correlation
Sources
A question about this article? Ask the library