|

Horizon Tracking with Physics Guided AI

Byadmin
 

Horizon Tracking with Physics Guided AI

Introduction

Horizon interpretation is fundamental to seismic analysis. It defines stratigraphy, reservoir boundaries, and structural frameworks. Manual horizon picking is slow and subjective, while traditional AI models often struggle with discontinuities or noisy data.

Physics Guided AI (PG AI) provides a powerful solution by combining machine learning with geological rules.

1. Why Horizon Tracking Is Difficult

Challenges include:

  • Variable reflector strength

  • Fault offsets

  • Noise and multiples

  • Stratigraphic complexity

  • Lateral facies changes

AI models must understand both local and regional geological context.

2. How Physics Guided AI Enhances Horizon Tracking

PG AI incorporates:

  • Dip limits

  • Continuity rules

  • Stratigraphic relationships

  • Amplitude behavior

This ensures that horizons follow realistic geological patterns rather than noise or artifacts.

3. Horizon Tracking Workflow

Stage 1: Data Preparation

Seismic volumes are conditioned and normalized to ensure consistent amplitude and phase behavior.

Stage 2: Feature Engineering

Attributes used for horizon tracking include:

  • Dip

  • Instantaneous phase

  • Envelope

  • Spectral decomposition

These features help the model understand reflector geometry and continuity.

Stage 3: Physics Constraints

Rules ensure:

  • Continuous reflectors

  • Realistic dip changes

  • Proper fault offsets

  • Stratigraphic consistency

These constraints prevent unrealistic horizon shapes.

Stage 4: Model Training

The model learns horizon patterns while respecting physics‑based rules and geological context.

Stage 5: Inference

The model outputs:

  • Horizon probability volumes

  • Auto‑tracked surfaces

These results dramatically accelerate interpretation.

Stage 6: QC

Interpreters validate and adjust the results to ensure geological accuracy.

Stage 7: Refinement

Models are iteratively improved based on QC feedback.

Stage 8: Deliverables

Outputs include:

  • Horizon surfaces

  • Probability volumes

  • Interpretation layers

Conclusion

Physics Guided AI dramatically accelerates horizon interpretation while maintaining geological realism. It reduces manual effort, improves consistency, and enhances reservoir understanding.

Similar Posts