Surface-Related Multiple Elimination (SRME)
SRME stands for Surface-Related Multiple Elimination. It is a widely used method in seismic data processing to remove surface-related multiples unwanted seismic reflections that bounce between the surface and subsurface layers before being recorded. These multiples can interfere with the interpretation of primary reflections and reduce the clarity of seismic images.
Key Concepts:
- Multiples are seismic events that have undergone more than one reflection.
- Surface-related multiples specifically involve at least one reflection off the sea surface (or land surface).
- SRME does not require subsurface information (like a velocity model), making it a data-driven approach.
Advantages:
- No need for subsurface velocity models.
- Effective for complex geology where multiples obscure primary events.
Limitations:
- Sensitive to data quality (e.g., missing near offsets, irregular sampling).
- Assumes free-surface is the main cause of multiples.
Incomplete Shot/Receiver Coverage
SRME theoretically assumes every shot gathers every receiver (full coverage). But in real surveys:
- Marine acquisition has sparse receiver spacing (streamer cable has limited channels).
- Land acquisition can be irregular or missing receivers/shots.
Solution:
- Interpolation or regularization is performed first to create a dense and regular grid before SRME.
- Techniques like anti-alias interpolation, minimum-curvature interpolation, or Fourier-based methods are used.
Higher-order Multiples
SRME naturally predicts:
- First-order multiples (one surface bounce).
- Second-order multiples (two surface bounces).
- Etc., all orders are inherently included.
But:
Higher-order multiples are weaker and harder to predict accurately because they depend on many surface interactions.
Solution:
- Some implementations prioritize first-order multiples.
- Special techniques or iterative SRME can enhance higher-order multiple removal.
How SRME Works
1. It predicts multiples by cross-convolving the recorded data with itself.
2. Imagine the first reflection comes up, hits the surface, and then reflects back down. SRME simulates this by using the recorded upgoing wave to model how it would reflect at the surface and create a multiple.
3. These predicted multiples are then subtracted from the original data, ideally leaving only the primaries.
Practical SRME Processing Flow (Typical):
- Input data regularization (interpolation to fine grid).
- Multiple prediction (via convolution of traces with each other).
- Adaptive subtraction (matching amplitude/phase).
- Post-cleaning (mute, filtering, residual demultiple).
