Sismik Prospeksiyon ( ingilizce ) Introduction to Seismology - 4 12.510 Introduction to Seismology 02/27/2008 – 02/29/2008 12.510 Introduction to Seismology Feb. 29 th 2008: We have introduced the equations: and With and These equations can be solved using 3 different methods: 1. D’Alembert’s solution (most ‘physical’ approach) 2. Separation of variables 3. Fourier Transforms (mathematically, the most powerful method) There is a whole class of theoretical development in applied maths that uses Fourier Integral Operators (FIOs) 1.D’Alembert’s Solution: Take as an example, the wave equation: And the function: (45) The first term: f(x – ct) represents propogation in the positive x-direction The second term: g(x+ct) represents propogation in the negative x-direction c = the wave speed or phase speed/velocity Consider the profile of the wave at a time and at some later time Figure 4: 12.510 Introduction to Seismology 02/27/2008 – 02/29/2008 (x-ct) is known as the phase of the wave. The phase speed is given by: (46) Figure 5: Diagram to illustrate the concept of wavefronts: Wavefront = surface connecting points of equal phase A wavefront is a line in 2d (or surface in 3d) connecting points of equal phase. In reality, the wavefronts are circular, but locally they behave as a plane wave. All points along the wave-front have the same travel-time from the origin. The relationship between the wavenumber (k) and angular frequency is given by: (47) The relationship between wavenumber (k) and wavelength is given by: (48) So we can re-write the phase in terms of wavenumber: (in one dimension) In 3 dimensions, this becomes: or The harmonic function is a solution to the wave equation: () ? () ? () xyz kx ky kz t ? ++-() . kx t ? - k c ? = 10 10 x x c tt - - - 2 k ? ? = () x tk x c t ? ?? -›- ?? ?? 12.510 Introduction to Seismology 02/27/2008 – 02/29/2008 9) Where we have used the identity: exp i (?? )=+ cos( ) i sin(? ) (50) 2. Separation of variables: Using the method of separation of variables, we trial a solution of the wave equation of the form: (51) Substituting this into the equation: ?? =? c 22 gives: (52) To satisfy this equation, each term must be equal to a constant and the constants must sum to zero. We choose the constants: respectively (53) Applying the condition that these constants must sum to zero gives us the dispersion relation: (54) Substituting the solutions for X,Y,Z,T back into the original trial solution: Gives our final solution of the form: (55) Note: Wavenumbers and the wavevector: ) ?? =- cos () kx t+ i sin ( kx- ? t ) = expi ( kx- t (4 ? .. 2 222 ''' 2 xyz kkk c ? ?? --- ?? ??12.510 Introduction to Seismology 02/27/2008 – 02/29/2008 A lot of imaging is done in the frequency domain by finding solutions to the helmholtz equation. 3. Fourier Transforms: Fourier transforms allow us to understand the relationship between the space-time (x,t) and wavenumber-frequency (k, domains. In one dimension, the forward and reverse fourier transforms between the space- frequency and space-time domains are given by: (Note: in seismology, we normally take the exponential as having a positive sign when we are transforming into the space-frequency domain, however this is simply a convention) Similarly, we can use fourier transforms to convert between the space-time and wavenumber-time domains. In 3d the fourier transforms between the space-time and wavenumber-time domains are: Combining these gives the double-fourier transform: Note that does not appear in this equation. This is because, are related via the dispersion relation. Hence, if we have specified the angular frequency, and it follows that has already been determined. Numerically, this double fourier transform is very difficult to work with. Note: Synthetic seismograms: (69) (68) (67) (66)12.510 Introduction to Seismology 02/27/2008 – 02/29/2008 A synthetic seismogram is given by a plane-wave superposition. Suppose we want to make a synthetic seismogram that looks similar to the wave. We do not need to integrate over the full range and because this implies we do not know anything about the direction of the wave. A synthetic seismogram can be produced by limiting the integration over directions and frequency . (70) The integrand ?(, kk xy , ?z ) is the amplitude or weight. Slowness: We have see already that the modulus of the wave-vector gives the wavenumber: In 2d, we have: (71) kk 222 w =+ xy k + k z = c kk 22 w =+ xz k = c Figure 6: In figure 6, the arrow is used for a ray and the dashed line is used for a wavefront. The wavenumber k indicates the direction of the ray. The angle i is both the ‘take-off’ angle and the ‘angle of incidence’ For a P-wave, the wavenumber is given by: and for the S-wave, the wavenumber is given by Since, in general: it follows directly, that: (72) k ? ? = k ß ? ß = ß ? < w kk ß? ? ß ? =>= ? 0 d ?? ± 0 kd k ± x k ? ? -< < , y k ? ? -< <12.510 Introduction to Seismology 02/27/2008 – 02/29/2008 Since gives the length of the vector representing the P-wave and represents the length of the vector representing the S-wave, it follows directly that P-waves ‘dive’ less steeply into the medium than S-waves. Figure 7: S P x z The phase ‘speed’ c, is given by: (73) and is a vector in the direction of propogation At the surface, we measure: (74) which is the ‘apparent’ velocity/speed Horizontal slowness: eter (76) Vertical slowness: We know that: (77) The vertical slowness is given by: Combining the vertical and horizontal slowness: (78) Rearranging this gives: (79) k ß = ? ß k ? = ? ? ds c dt = x dx c dt = c p (75) = horizontal slowness = ray param This follows from Snell’s law ds dt ?? 1 sin(tc )== =c ?? = dx dx ?? c x 1s i n ( t) ?== cc x (,) xxz ccc = 22 22 22 sin cos 1 tt cc ?? += + = 2 c 12.510 Introduction to Seismology 02/27/2008 – 02/29/2008 So, the vertical slowness does change with depth, because c is a function of depth. The vertical slowness is zero if (which represents a horizontally propagating wave) ? is imaginary for evanescent waves. (This is important for understanding the behaviour of surface waves). There is a direct relationship between the wave-vector and the slowness components: (80) (81) (82) Figures from notes of Patricia M Gregg (Feb 2005) Kang Hyeun Ji (Feb 2005) Notes: Katie Atkinson, Feb 2008 2 2 1 p c = () ? x x kp c ? ? == k c ? = z z k c ? ?? == (,)(,) (,) xz kkk p p ? ?? ? ? ===