Anisotropic Moveout

Part 7, Part 7: Anisotropy, the Physics

Learning objectives

  • Recall hyperbolic NMO in an isotropic earth
  • See that the short-offset NMO velocity is set by delta
  • Watch far-offset moveout go non-hyperbolic with eta
  • Read the residual hockey stick as the anisotropy signature

The Gather That Will Not Flatten

On a common-midpoint gather, a reflection arrives later at longer offset. In an isotropic earth that delay is a clean hyperbola, t(x)2=t02+x2/V2t(x)^2 = t_0^2 + x^2/V^202+x2/V2, and choosing the correct NMO velocity VV subtracts the moveout and lays the event flat at t_0t_0. One velocity, one flat gather. This is the engine of stacking.

VTI anisotropy breaks it in two places. First, the velocity that flattens the near offsets is not the true vertical velocity but Vnmo=Vp0sqrt1+2deltaV_{nmo} = V_{p0}\sqrt{1+2\delta}p0sqrt1+2delta. The near-offset moveout is governed by delta\delta, so a gather can look perfectly flat at short offset and still be built on the wrong velocity. Second, at long offset the moveout stops being a hyperbola at all.

Anisotropic moveoutt0offset 0 to 3200 m (best hyperbola dashed; true moveout curls away far out)Delta sets the short-offset NMO velocity; eta makes far-offset moveout non-hyperbolic, leaving a residual hockey stick after correction.

Eta, and the Hockey Stick

The departure from a hyperbola is controlled by the anellipticity eta=(epsilondelta)/(1+2delta)\eta = (\epsilon-\delta)/(1+2\delta), and the industry-standard description is the Alkhalifah-Tsvankin non-hyperbolic moveout equation. When eta\eta is zero (the elliptical case, epsilon=delta\epsilon=\delta) the moveout is exactly hyperbolic and one velocity flattens everything. When eta\eta is not zero, correct the gather with the best single hyperbola and the near offsets snap flat while the far offsets curl up into a residual hockey stick. The size of that curl is a direct measurement of eta\eta.

This is why long-offset velocity analysis needs two numbers, VnmoV_{nmo}nmo and eta\eta, not one. It is also the fit-for-purpose boundary for modelling: if you only care about a short-offset stack, an isotropic velocity that absorbs delta\delta is enough, but the moment you use far offsets, for AVO or for imaging, ignoring eta\eta leaves that hockey stick in your data and mis-positions your reflectors. The next section takes the same idea sideways, into the azimuth, where fractures live.

This page is prerendered for SEO and accessibility. The interactive widgets above hydrate on JavaScript load.