Purdue University

EAS 557

Introduction to Seismology

Robert L. Nowack

Lecture 13A

I.  Reflection and Transmission of Waves at Boundaries

 

 

            There are several different kinds of boundary conditions that we will consider.

 

a)   Welded boundaries

 

            This boundary condition requires continuous particle displacements and tractions.  The normal to the boundary at a given point is noted by  pointing from the negative side to the positive side of the boundary.

 

 

 

 

1)   Displacements are continuous.  This can be written

 

 

      where  and  are on the plus and minus side of the boundary.

 

2)   Tractions on the surface are continuous.  This requires

 

 

 

b)   Solid-fluid boundaries (i.e., the ocean bottom or the core-mantle boundary)

 

            This is the same as for the welded case, except now lateral slip along the boundary can occur.  Thus, only the component of displacement normal to the boundary will be continuous.  But, tractions will still be continuous.  Thus,

 

 

and

 

 

 

c)   Free surface (i.e., the Earth’s surface)

 

            An example of a free surface is the surface of the Earth.  (In fact, this could be modeled as a fluid-solid boundary except that the properties of the air are so much different than the solid that it is sometimes modeled as a free surface.)  Thus, the boundary conditions are zero tractions on the boundary.

 

 

There is no restriction on displacements.

 

These general boundary conditions must be satisfied at all discontinuities in the medium.

 

 

 

 

No matter what computer program one uses (finite differences, finite elements, etc.), one must use these boundary conditions.  A welded boundary has six total boundary conditions, a fluid solid boundary has four total boundary conditions and a free surface has three boundary conditions.

 

            Examples in the Earth of different boundaries are:

 

The Earth’s surface (solid, fluid) – free surface

Ocean-seafloor – fluid-solid

Basement sediments – wielded

Depositional discontinuities – welded

Conrad discontinuity in continental crust – welded

Moho discontinuity between crust and mantle – welded

“410” and “660” upper mantle discontinuities – welded

Core mantle boundary – fluid-solid

Outercore-innercore – fluid-solid

 

 

            Earlier, we found the wavefields for a point force, a double couple shear source, and an explosion source in a whole space.

 

 

 

 

For the source sufficiently far away from a boundary, the wavefront will be effectively planar.  Thus, as an approximation, planar waves can be assumed for a very distant source.

 

            Earlier, we found that a plane wave could be written in complex form as

 

 

where it was found that  is the wavenumber vector and points in the direction of propagation.  The magnitude of  where c is the wave velocity.

 

 

 

 

            In 2D, .  From the diagram above,

 

 

Sometimes we will also use the slowness vector

 

 

where c is the wave speed.

 

            Both P-waves and S-waves can propagate in the far fiield

 

1)   P waves.  The speed is   and the particle motion  is parallel to .

 

 

 

 

Thus, for a  vector in the x1 direction,  and

 

 

 

2)   S waves.  The speed is  and the particle motion  is perpendicular to .

 

 

 

 

Thus, for a  vector in the x1 direction,  and

 

 

The flux rate of energy transmission in a plane wave (the amount of energy transmitted per unit time across unit area normal to the direction of propagation) is proportional to  for P waves and  for S waves.  This is a local property depending on

 

a)   Material parameters

b)   Local planar nature of wavefront

c)   E is related squared amplitude through .

d)   Depends on , which is the seismic impedance, where c is the particular wave speed.  This is also related to the ratio of stress to particle velocity.

 

            The energy flux across a tilted plane is proporational to , where c is the wave speed and i is the angle of the tilt.  We will later find that reflection coefficients will depend on mis-matches in the seismic impedance across layer boundaries.

 

 

            We will now assume a plane wave incident on a plane boundary.  In order to simplify the problem, a local coordinate system will be centered at the boundary with the (x1,x3) plane determined by the incident  vector and the normal to the boundary .  This is called the plane of incidence.  An incident P-wave is first shown.

 

 

 

 

The orientation of the interface will “polarize” the S waves into “SV” waves with particle motions in the plane of the incidence “SH” waves with particle motions perpendicular to the plane of incidence. 

 

 

 

 

 

 

Positive P, SV, and SH will be chosen with regards to (x1,x2,x3) local coordinates of the interface for the incident, transmitted, and reflected waves.  Note that other sign conventions can be used, but if one knows what they are, one can change them as required.

 

        For an incident “SH” wave, the particle motions of the incident wave and all scattered waves are perpendicular to the plane of incidence.

 

        In order to solve for the reflected and transmitted wave, we first apply the boundary conditions for a welded interface.

 

 

 

 

a)   Apply continuity of displacement at x3 = 0.  Then,

 

 

      where we will assume the incident wave to be an SH wave, then  or

 

 

      where

 

 

with

 

 

 

 

 

      Also,

 

 

 

      for the reflected and transmitted waves.

 

b)   Apply continuity of tractions at x3 = 0.  In terms of the stress matrix, this can be written

 

   where   

 

and the stress matrix for isotropic media can be written

 

 

            The tractions are then .  For

 

 

then,

 

 ,

 

 

For a plane wave of the form

 

 

then,

 

 ,  ,

 

Now,

 

 

 

 

 

then, for continuity of traction

 

 

Note for the incident “SH” case, there is no coupling of A2 with A1, A3 for continuity of displacements or tractions.  Then,

 

 

and

 

 

For continuity of displacement and traction, then

 

 

 

 

where both are evaluated at x3=0.

 

            First, these equations will only be true if the phase is the same for all terms.  This will occur if

 

where

 

 

Thus, .  This is commonly called Snell’s law.  In French, this is sometimes called Descartes law

 

 

 

 

Another way of saying this is that the horizontal component of the wavenumber is unimpeded by a horizontal interface. 

 

            In seismology, the slowness vector is also used where .  Thus,

 

 

Snell’s law also states that the horizontal component of the slowness vector travels without being impeded by a horizontal interface.  Thus, upon reflection, assuming no phase conversion from P to S (as there would be for the in-plane P-SV case), then

 

1)   The first simple law of geometric optics is

 

 

or this says that the angle of reflection is equal to the angle of incidence, regardless of the material properties across the boundary.  This is also called Euclid’s law of reflection.

 

2)   For transmission, then

 

 

 

 

 

      Thus, for , .

 

      Comments

 

a)   The phase relations don’t depend on density, only on velocity.  Thus, phase information or travel times cannot give a complete picture of the Earth’s interior.

 

b)   For an earth with increasing velocity with depth, then downgoing “rays” represented by the wavenumber vector  will flatten with depth and then turn back up to the surface.

 

 

 

 

c)   Upward traveling “rays” will steepen for lower velocity layers

 

 

 

 

d)   In horizontally layered structures, the horizontal slowness, , does not change along the path, where c is the wave speed.  This is given a special name and symbol

 

 = ray parameter

 

      since in a vertically varying medium, p uniquely identifies the “ray” going through the structure.