Multiples
Learning objectives
- Define a multiple as a reflection that bounces more than once
- Recognise the surface-multiple train at multiples of the primary time
- Explain why a free surface (a mirror) creates multiples
- Know that convolution cannot make multiples and processing removes them
Echoes of Echoes
A primary reflection bounces off an interface once and returns. A multiple bounces more than once, and the most troublesome is the surface multiple. The reflected wave arrives at the sea surface, which is a near-perfect mirror between water and air, reflects back down, hits the reflector a second time, and returns again at roughly twice the primary's traveltime. It repeats: a third arrival near three times, a fourth near four, a fading train of echoes.
The Mirror Makes Them
Switch the free surface off, replacing the mirror with an absorbing top, and the record shows a clean primary with almost nothing after it: the reflected wave leaves through the top and never comes back. Switch the free surface on and the multiple train appears, marked at two and three times the primary time. The mirror is the whole cause; take it away and the multiples vanish.
Two consequences follow. First, convolution can never make a multiple. Its reflectivity series reflects each interface exactly once by construction, so a convolutional synthetic has primaries and nothing else, one more item on the list of what it leaves out. Only a wave-equation engine with a reflecting boundary produces the train. Second, multiples are a genuine menace in real data: a strong surface multiple can mimic a deeper primary and lead an interpreter to a reflector that is not there. Demultiple, the removal of multiples, is one of the largest efforts in seismic processing, and a synthetic that faithfully contains multiples is exactly how those removal algorithms are built and tested. Making the problem on purpose is how you learn to solve it.