Short-Path Multiple

Key Takeaways

• The geological settings most prone to generating problematic short-path multiples share the characteristic of having two or more high-impedance contrasts in close vertical proximity: shallow marine environments where a hard seafloor (high-velocity carbonate or cemented sand) overlies a soft sub-bottom reflector create a perfect short-path multiple generator; coal-bearing sequences in shallow continental settings generate interbed multiples because coal has very high acoustic impedance contrast with surrounding shales; carbonate reef environments with tight porous carbonates alternating with tight non-porous carbonates generate interbed multiples that contaminate the seismic image of the reservoir below; in each case, the wavefield energy that should be a single downgoing and upgoing reflection becomes trapped in a resonating bounce between the two reflectors, generating a series of multiple reflections at integer multiples of the two-way travel time through the thin layer, each one potentially interfering with the primary reflections from deeper targets.
• The predictive deconvolution algorithm applied in standard seismic processing was originally developed specifically to address short-path interbed multiples: if the earth behaves as a collection of horizontally layered reflectors and the wavelet is minimum-phase, predictive deconvolution uses the autocorrelation of the seismic trace at a chosen lag (the prediction distance, set equal to the two-way travel time through the multiple-generating layer) to predict and subtract the multiple arrivals from the trace; the method works well when the multiples are truly periodic (constant two-way travel time in the generating layer) and when the subsurface geology is sufficiently simple that the statistical assumptions of the algorithm are met; in practice, lateral velocity variation, dipping layers, and complex multiple ray paths violate these assumptions, making predictive deconvolution a partial but never complete solution that is typically followed by additional surface-related multiple elimination or model-driven subtraction methods to address the residual multiple energy.
• Surface-related multiple elimination (SRME) is highly effective for long-path water-bottom multiples but struggles with short-path interbed multiples because SRME constructs the multiple model by autocorrelating the seismic data along the surface receiver line, which captures reflections involving the surface as one of the bounce points; short-path interbed multiples involve bounces between subsurface reflectors only and do not use the surface as a reflection point, placing them outside the theoretical scope of SRME; internal multiple attenuation (IMA) algorithms — which use the data itself to predict and subtract internal multiple energy — are the appropriate tool for short-path interbed multiples, but require accurate knowledge of which reflector pair is generating the multiple, and the prediction becomes computationally intensive in three-dimensional datasets; the combination of SRME for surface-related multiples and IMA for interbed multiples represents the current best practice for comprehensive multiple attenuation in geologically complex areas, but neither method is perfectly effective in isolation.
• The interpretation risk posed by unattenuated short-path multiples is significant in frontier basin exploration where the primary reflection from a potential reservoir may be of similar amplitude to a short-path multiple from a shallow high-impedance layer: amplitude-versus-offset (AVO) analysis intended to identify hydrocarbon-bearing sands relies on the assumption that the analyzed event is a primary reflection, and an AVO anomaly generated by a short-path multiple rather than a hydrocarbon contact is a direct cause of exploration dry holes; the identification of multiples versus primaries in the interpretation stage uses several techniques including forward modeling of expected multiples from known reflectors, comparison of the event's move-out with the theoretically predicted multiple velocity, and scrutiny of the event's coherence across multiple vintages of seismic data or across different acquisition geometries; a bright-spot anomaly that disappears when a different source vessel acquires the same line (different source ghost characteristics affecting the multiple generation) is strong evidence that the anomaly is a multiple rather than a lithological or fluid effect.
• The increasing use of broadband seismic acquisition (which extends the recorded frequency range to both lower and higher frequencies than conventional acquisition) has changed the character of short-path multiples and their interaction with primary reflections: at low frequencies, short-path interbed multiples have wavelengths comparable to the distance between the generating reflectors, causing the multiple to interfere constructively or destructively with the primary in a frequency-dependent pattern that manifests as spectral notches in the data; at high frequencies, the multiples are more resolvable from the primaries but also more numerous (higher-order multiples are above the noise floor at high frequencies), requiring more sophisticated attenuation; broadband processing workflows that address multiples in the extended frequency band rather than the traditional 8-80 Hz window have been developed specifically for areas where the increased frequency content of broadband acquisition makes interbed multiples more prominent and more damaging to interpretation quality.