Long-Path Multiple

Key Takeaways

• The geometry of long-path multiples determines their traveltime and their velocity discrimination from primaries: a first-order water-bottom multiple (energy that reflects from the water bottom, travels up to the sea surface, reflects back down, reflects from the water bottom again, and then travels back to the receiver) has a traveltime of 3*t_wb (where t_wb is the two-way traveltime to the water bottom), and a stacking velocity approximately equal to the water velocity divided by sqrt(3) -- lower than the primary velocity at the same traveltime; a peg-leg multiple (energy that reflects from the water bottom, travels up to the sea surface, reflects down, reflects from a deeper reflector, and returns to the receiver) has a traveltime of t_wb + t_deep (where t_deep is the two-way traveltime to the deeper reflector), and an intermediate stacking velocity between the water velocity and the formation velocity at depth; in practice, the velocity difference between long-path multiples and primaries at the same traveltime is the primary basis for multiple discrimination in velocity analysis (the Dix-equation velocity extracted by semblance analysis will be lower for the multiple than for the primary at the same traveltime), but the discrimination becomes increasingly difficult in shallow water where the water-bottom multiple traveltime approaches the primary traveltime of the target, and in areas of high formation velocities where the multiple velocity closely approaches the primary velocity from a different depth.
• Interbed multiples (long-path multiples generated entirely between subsurface interfaces, without involving the sea surface) are among the most challenging to attenuate because SRME (which predicts and removes surface-related multiples) does not apply to them: an interbed multiple between two carbonate reflectors at 2,000 and 2,500 ms TWT generates a first-order interbed multiple at 3,000 ms TWT (one bounce off the shallow carbonate, one bounce off the deep carbonate, returning to the receiver without involving the sea surface); this interbed multiple at 3,000 ms may overlap with the primary reflection from a target at 3,000 ms TWT, with similar moveout because both the interbed multiple and the primary at 3,000 ms have traveled through formation velocities at depth; the inverse scattering series (ISS) method for internal multiple attenuation (developed by A.B. Weglein and colleagues) provides a data-driven prediction of interbed multiples without requiring a subsurface model, using the fact that internal multiples can be expressed as a convolution of shorter-path events; however, ISS internal multiple attenuation is computationally intensive and requires good data quality and spatial sampling; in practice, many interbed multiples survive standard multiple attenuation processing and reach the interpretation workstation where they must be identified and discounted by the interpreter using velocity discrimination, wave equation modeling, or well-tie comparison (a well penetrating the implied depth would fail to find the reflector corresponding to the multiple).
• Multiple identification on seismic data using wave equation modeling involves computing the synthetic seismogram for a proposed multiple-generating horizon and comparing the synthetic multiple to the observed event on the seismic data: given the reflectivity model from wells and the seismic velocities, the forward seismic modeling tool (ray tracing or finite difference) computes the traveltime, amplitude, and moveout of the proposed multiple event; if the modeled multiple matches the observed event in traveltime, amplitude variation with offset, and character (wavelet shape), the observed event is likely a multiple; if the modeled multiple does not match the observed event (wrong traveltime or amplitude), the observed event may be a primary from a real reflector; this process requires that the velocity model and the reflectivity model are sufficiently accurate to make the modeled multiple credible, which is a strong assumption in frontier exploration areas where neither the velocity structure nor the reflectivity sequence is well constrained below the first reliable well penetration; multiple modeling is therefore most reliable in mature areas with multiple wells penetrating the generating horizons, and least reliable in frontier basins where the multiple identification must rely primarily on velocity discrimination and periodicity analysis.
• Long-path multiples in the context of land seismic acquisition arise from different geometries than marine multiples but can be equally damaging to interpretation: on land, long-path multiples are generated by the free surface (the ground surface, which acts as a near-perfect reflector for downward-traveling seismic energy), by near-surface reverberation layers (low-velocity, high-impedance interfaces in the weathering layer), and by interbedded high-impedance reflectors (coal seams, basalt flows, evaporites) in the overburden; the coal seam multiple problem in the Cooper Basin of Australia, the Carboniferous coal basin of Europe, and similar geological settings produces strong long-path multiples from the top and base of coal seams that interfere with the primary reflections from the target Permian or Triassic sands, requiring aggressive pre-processing demultiple to produce usable images of the subsalt or sub-coal structures; evaporite sequences (salt, anhydrite, halite) with very high acoustic impedance generate strong interbedded multiples between the top and base of the salt and between salt intervals and overlying or underlying carbonates, creating complex multiple patterns that can mimic primary reflections from both intra-salt and sub-salt targets.
• Economic consequences of misidentifying long-path multiples as primary reflections in exploration are severe and well-documented: the history of seismic exploration includes numerous examples of exploration wells drilled on apparent closures or anomalies that turned out to be multiple reflections rather than primary reflections from real structural traps or stratigraphic features; in the Gulf of Mexico shelf, water-layer multiples at 50 to 200 meters water depth create apparent anticlines at the traveltime of the target that are mirror images of the real bathymetric and structural features above, causing explorers in the 1960s and 1970s to drill prospects that were multiples of the water bottom rather than real subsalt structures; the development of SRME in the 1990s and its widespread adoption in the 2000s substantially reduced (but did not eliminate) this category of exploration error in marine seismic; in land seismic, coal and basalt multiple problems continue to cause significant exploration failures in basins where the near-surface geology generates strong, consistent multiples that look like deep reflectors in the stacked section; the cost of an exploration well drilled on a multiple is the full well cost ($5 to $100 million depending on depth and location) with zero probability of a discovery, making multiple identification the highest-value quality control step in seismic interpretation for exploration decisions.