diffraction hyperbola

[Tingpingbing]

Hello,
Is anyone here familiar with how to understand a diffraction hyperbola?
In the attached figure the first picture is a velocity model and the second picture is a NMO stack and also a corresponding semblance plot.
Why does the diffraction apex and the reflection equal to the water layer around 1500 to 2000m/s?
And what does it mean when the diffraction has low velocity on the apex and increase its velocity further in the flanks
Why does it look like the apex is large? Like a leakage? What indicates this?

For the NMO stack:
What does it actually indicate?
Why is the amplitude high on the apex?

For semblance plot:
Why is the semblance at the highest on the apex? What does actually a semblance plot of a diffraction show?

Hope someone can help!

[GuyM]

Hmm - that doesn't look tight to me.

A diffraction is caused by a discontinuity in the sub-surface geology (and hence velocities), so the model you have on the left would not produce a diffraction like you see on the right; it might produce a complex reflection with a hyperbolic shape, but that's not a diffraction.

In general if you perform stacking velocity analysis on CMP gathers than include a diffraction, the diffraction tails "stack up" (ie are flattened) with an anomalously *high* stacking velocity. However, this is *not* what you should be picking. You should be flattening reflections, not diffractions, at any position away from the apex. I

The diffractions are (from a stacking velocity analysis perspective) just noise, to be ignored. Only if you have applied some form of pre-stack time migration will the diffractions have been collapsed to their correct sub-surface position.

I'd expect an interval velocity model in two-way-time to have an small, higher velocity "blob" at the apex of the diffraction hyperbola on the stack, with a constant background (or gradient) behind it.

So - not really sure what you are showing. Are you starting with the seismic data, or the velocity model?

[Tingpingbing]

the first to start with is the velocity model, then from this model the stack is created.
Then after a semblance search we get the semblance plot

What does it mean that "reflections and diffractions have different kinematic properties"?
And "the kinematic part of data sorted in CMP gather can be a normal moveout"?

[GuyM]

the first to start with is the velocity model, then from this model the stack is created.
Then after a semblance search we get the semblance plot

Okay. Then what you are seeing is *not* a diffraction, just a reflection from that anticline in the velocity model. You din't need to build the shape of the diffraction into the model, its a natural consequence of wave propagation as described by Huygen's principle - assuming this is actually running some kind of wave equation simulation rather than just a 1D convolution model down each trace, of course.

Not sure what you are aiming at with the semblance search/plot on the stacked data, however?

Try making some different but very simple models that have simple geological structures in them like a normal fault with a large displacement, and generating a response for those.

What does it mean that "reflections and diffractions have different kinematic properties"?

How the travel time of the event varies with offset is different; a diffraction and reflection at similar two-way-times will stack (ie flatten on an NMO gather) with different NMO velocities.

And "the kinematic part of data sorted in CMP gather can be a normal moveout"?

When you take a CMP gather with a reflected event in it, the shape of that event (if it comes from a more-or-less flat interface) will obey the NMO equation. Dipping events (and structure) will modify the shape of the reflected event.

[Tingpingbing]

thank you