ML-assisted first-break picking

Part 9 — Machine Learning in Processing

Learning objectives

State the first-break picking problem and why it matters for refraction statics
Compare STA/LTA, AIC, and CNN pickers and their failure modes
Describe the U-Net training regime for first-break picking
Recognise why ML picking has become the production standard

The first break — the onset of the direct or refracted arrival on a shot record — is the single most-picked quantity in seismic processing. Every shot on every survey gets first-break picks used for: refraction static corrections (land), direct-wave velocity QC, near-surface tomography, overburden velocity building. Classical pickers were heuristic and required constant human oversight. Modern CNN-based pickers have taken over this role in production.

1. The three classical pickers

STA/LTA (short-term / long-term average). Compute the ratio of a short-window (10–40 ms) energy to a long-window (50–200 ms) energy as a function of time; trigger when the ratio crosses a threshold (typically 2–5). Simple, fast, but sensitive to pre-arrival noise and threshold choice. Systematic late-pick bias at low SNR.
AIC (Akaike Information Criterion). Model the trace as two segments split at index $k$ : pre-arrival (white noise) + post-arrival (white noise with different variance). Compute AIC as a function of $k$ ; the minimum marks the split. More robust than STA/LTA to noise level but assumes noise is white and stationary.
Correlation. Cross-correlate each trace with a template (an assumed wavelet). The correlation peak marks the first break. Good when the wavelet is known; poor when it varies shot-to-shot.

2. Why ML wins at this task

First-break picking is a pattern-recognition problem with a tight labelled training set available (any historical survey has hand-picks). A CNN sees:

The full waveform context of each trace (not just a moving window).
Adjacent traces' arrivals (moveout consistency).
Wavelet variations learned from training examples.
Noise patterns learned across many surveys.

A trained U-Net pick is ~1 ms/trace vs 10–100 ms/trace for a careful human. Accuracy typically 2–5 ms RMS vs hand picks (which are themselves uncertain at this level).

24-trace shot gather with first arrivals on a linear direct-wave moveout (V = 1800 m/s, 50 m spacing). The user-controlled pre-arrival noise level affects classical pickers more than the CNN. Three pick lines overlaid on the gather: yellow STA/LTA, orange AIC, pink CNN. Ground truth (green dashed) lies between them for reference.

At low noise (~0.02) all three pickers are within a few ms of truth. At moderate noise (~0.10) STA/LTA and AIC start to show systematic late-picking and scatter; the CNN stays within ~3 ms. At high noise (~0.25) STA/LTA and AIC produce wild outliers; the CNN still roughly tracks truth because it was trained on noisy examples and learned to use trace-to-trace consistency.

4. U-Net for first-break picking

Input: a 2D patch of the shot gather (traces × time). Output: a binary mask where 1 marks the first-break arrival. The network is trained with a cross-entropy loss on the mask. At inference, the argmax of the predicted mask per trace gives the pick time. Training data: tens of thousands of shots with hand-picks, usually from multiple surveys to capture wavelet variety.

Hybrid variants use the mask output as an initial pick, then refine with AIC over a narrow window around it; this combines ML robustness with AIC's sub-sample accuracy.

5. Production deployment considerations

Cross-survey generalisation. A network trained on one survey (one wavelet, one noise regime) may systematically miss arrivals on another. Standard practice: pre-train on a diverse set of surveys, fine-tune per project.
Human QC. Every production workflow retains interpreter QC on a subset of picks. The CNN replaces the bulk labour but not the final review.
Uncertainty flagging. CNN outputs a confidence map; low-confidence picks are flagged for manual review. Saves time by focusing QC on ambiguous cases.
Outlier rejection. Post-CNN, apply a moveout-consistency filter: picks that deviate by more than some threshold from a smooth fit are flagged or rejected.

6. Failure modes

Refraction static errors. If the first breaks are systematically mis-picked by a few ms, near-surface velocity models become biased, and refraction statics are wrong — the downstream imaging shows residual statics artefacts.
Cycle-skipping. At very low SNR the CNN may pick the wrong cycle (one period late or early). Harder to catch than STA/LTA cycle-skipping because CNN outputs look confident. Mitigation: compare predicted picks to a smoothed moveout model, flag outliers.
Domain shift under novel noise. Electrical hum, drone noise, unexpected infrastructure — the network has not seen these in training and may produce plausible but wrong picks. Mitigation: add diverse noise to training; include "synthetic noise augmentation" during training.

**The one sentence to remember**

CNN first-break picking (typically U-Net on 2D gather patches) has replaced STA/LTA and AIC in production because it is ~100× faster at comparable accuracy, with the main caveat that cross-survey domain shift requires fine-tuning per project.

Where this goes next

§9.5 closes Part 9 with the most recent ML frontier in seismic processing: accelerating FWI. Instead of replacing the physics, ML initialises FWI with a learned prior, replaces expensive iterations with network inference, or provides a learned regulariser that keeps the solution geologically plausible.

References

Yilmaz, Ö. (2001). Seismic Data Analysis (2 vols.). SEG.
Sheriff, R. E., Geldart, L. P. (1995). Exploration Seismology (2nd ed.). Cambridge UP.
Claerbout, J. F. (1976). Fundamentals of Geophysical Data Processing. McGraw-Hill.