Using Seismic P And S Arrivals To Determine Shallow Velocity Structure

[0001] This application claims the benefit of U.S. Provisional Patent Applications 61/374,888, filed 18 Aug. 2010, entitled USING SEISMIC P AND S ARRIVALS TO DETERMINE SHALLOW VELOCITY STRUCTURE, the entirety of which is incorporated by reference herein.

[0002] The invention relates generally to the field of geophysical prospecting and, more particularly to seismic data processing. Specifically the invention is a method for using P and S arrivals to determine shallow velocity structure.

[0003] Active source seismic reflection data are commonly used in hydrocarbon exploration to remotely infer subsurface geologic structure and rock properties. A major advantage of these data are that the source and receiver locations can be strictly controlled, thereby allowing very detailed imaging of the subsurface. However, reflected signals from strata are recorded as time series, and must be migrated to depth in order to infer geologic structure. This process requires a reliable seismic velocity reference model, whose accuracy directly impacts the accuracy of the inferred structure.

[0004] A common approach to building migration models is to perform Normal Move-out Analysis or travel-time analysis on the seismic reflection data [e.g., Sheriff and Geldart, Exploration Seismology, Cambridge University Press, 134-135 (1982)]. In complex regions such as fold-thrust belts, regions with basalt sills and dykes, or intruding salt bodies, the quality of these models is often poor, leading to poor migration of seismic reflections and therefore inaccurately inferred structure.

[0005] The present invention is a new method for using seismic waves to obtain a subsurface velocity model for use in hydrocarbon exploration. Seismic energy may be generated either actively (e.g. by explosions or vibrations) or passively (e.g. by earthquakes or landslides). Seismic energy in the Earth travels in the form of either compressional waves (P waves) or shear waves (S waves). P waves can be identified by the presence of particle motion in the direction of wave propagation; whereas S wave particle motion is perpendicular to the direction of wave propagation (FIG. 1). P and S waves travel at different speeds through the Earth, and these speeds are important indicators of subsurface properties such as lithology, porosity, or fluid content.

[0006] FIG. 2 is a schematic diagram of the process of seismic exploration. Seismic waves can be observed at any geographic location using a purpose-built recording instrument (for example, a geophone or seismometer). These instruments are designed to record particle displacements or acceleration, and require three independent (typically orthogonal) components to fully describe particle motion at the Earth's surface.

[0007] Seismic energy observed at a deployed instrument (commonly referred to as a station) is recorded in the “station reference frame.” For consistency, stations are usually carefully positioned and leveled in the field such that the station reference frame corresponds to a standard geographic reference frame, with components pointing in the North (N), East (E), and vertical (Z) directions.

[0008] In one embodiment, the invention is a method for estimating a model of seismic velocity in a subsurface region from seismic data recorded on 3-component instruments, comprising using a computer to: (a) determine a wave front incidence angle for a selected seismic event in the seismic data at a plurality of frequencies or frequency bands; (b) determine a frequency dependent velocity from the wave front incidence angle for each frequency or frequency band; and (c) infer a depth-dependent velocity by inverting the frequency-dependent velocity.

[0018] The invention will be described in connection with example embodiments. However, to the extent that the following detailed description is specific to a particular embodiment or a particular use of the invention, this is intended to be illustrative only, and is not to be construed as limiting the scope of the invention. On the contrary, it is intended to cover all alternatives, modifications and equivalents that may be included within the scope of the invention, as defined by the appended claims.

[0019] The present inventive method relies on two fundamental properties of seismic waves. The first property is that waves propagate at speeds representing the average seismic velocity of the material the wave front passes through during one wave period. This means that high-frequency waves with shorter periods are sensitive to smaller-scale structure than low-frequency waves with long periods. The second property is that the inclination of the ray θ and velocity of the material (V) are related to a constant (p) called the ray parameter. The ray parameter represents the apparent slowness of the wavefront in a horizontal direction, which is why p is sometimes called the horizontal slowness of the ray. Together, these two properties imply that, for a given wave front with ray parameter p traveling through a heterogeneous Earth, waves of different frequencies are sensitive to different effective velocities α_eff(f), and will therefore have different apparent ray inclination angles (or incidence angles) θ(f).

[0020] The method disclosed herein is a technique for measuring the apparent incidence angle of a seismic wave observed at a surface as a function of wave frequency. This apparent angle is converted to an effective velocity as a function of frequency, which is then inverted to obtain a velocity model. In at least some of its embodiments, the inventive method involves the following steps, as shown in the flowchart of FIG. 3:

[0021] At step 31, obtain seismic data recorded at instruments in the desired location.

[0022] At step 32, measure apparent incidence angle for different frequency bands.

[0023] At step 33, convert incidence angle to velocity for each frequency band.

[0024] At step 34, convert frequency-dependent velocity to depth-dependent velocity.

[0025] These four steps will next be discussed in more detail. It will be obvious to those who work in this technical field that steps 32, 33 and 34 are performed using a computer in all practical applications of the invention.

[0026] Step 31: Obtain Seismic Data

[0027] Generally, any seismic event recorded at a station (active or passive) may be used by this technique, providing the event has a known source location and the wave-front slowness can be obtained. Use of P wave arrivals yields a compressional wave velocity model, and use of S wave arrivals yields a shear wave velocity model. Final depth resolution depends strongly on the frequency content of the data.

[0028] Step 32: Measure Apparent Incidence Angle

[0029] Incidence angle θ may be obtained by determining the angle necessary to rotate the station reference frame seismogram, D(N,E,Z), into the wave front reference frame seismogram D(P,S_V,S_H). Two steps may be taken: first, using the known source and receiver locations rotate D(N,E,Z) by an angle φ around the Z axis to D(R,Z,S_H) where the radial component R points horizontally along the great circle path between source and receiver and the transverse component S_H points horizontally perpendicular to R (FIG. 4). The angle φ may be obtained from known source/receiver pair locations. Second, rotate D(R,Z,S_H) by an angle θ around the T/S_H axis to D(P,S_V,S_H), where P points in the direction of P wave propagation, and (S_V,S_H) defines the plane of S wave propagation. This suggests how θ may be determined from the seismic data. For P waves, θ is the angle that places all seismic energy in the P direction, whereas for S waves, θ is the angle that places all seismic energy in the (S_V,S_H) plane. Examples of rotated seismograms are given in FIG. 5.

[0030] Due to the presence of random Earth and instrument noise recorded contemporaneously with the seismic event, it is impossible, in practice, to obtain a single rotation angle that results in all arriving energy E being correctly partitioned to the appropriate component. The method therefore determines the optimum arrival angle: for P waves, this is the angle that maximizes energy in the P direction; for S waves, this is the angle that minimizes energy in the P direction.

[0031] One method to quantify the energy partitioning E is simple integration (or summation) of recorded particle motion P(t) during the wave front arrival (Park et al., 1987):

E=∫|P(t)|·dt  (1)

The rotation angle that maximizes E is the incidence angle. Another method is to compute the zero-lag cross-correlation between the P and S_V components (Abt, 2010):

E=∫P*(t)·S_V(t)·dt  (2)

In this equation, the rotation angle that minimizes E is the incidence angle. Accuracy of these measurements can be improved by the use of standard time-series techniques, for example windowing or tapering the seismic data in the vicinity of the event arrival.

[0032] Frequency dependence may be achieved by performing these calculations for seismograms that have been filtered to specific frequency bands. While individual frequencies may be used, wave form stability is enhanced by using a filter that decays smoothly from the desired central frequency. Thus, θ is obtained as a function of frequency.

[0033] Step 33: Convert Incidence Atmle (θ) to Velocity (α)

[0034] A physical model must be used to convert incidence angle to effective seismic velocity. For example, Snell's Law (see FIG. 6) states that the horizontal slowness p is a conserved quantity between source and receiver, i.e.:

$\begin{array}{cc}p=\frac{\sin \left(\theta \right)}{\alpha }& \left(3\right)\end{array}$

where α is the material wave speed and θ is the incident angle of the wave front upon a material interface (as illustrated in FIG. 4). The quantity p will be known. For example, it can be obtained from freely available global catalogues and analyses of earthquakes but can also be measured between two stations and inferred in that way. Other conversion methods, i.e. other physical models that include higher order physical effects, may be used, for example the free-surface transfer matrix (Kennett, 1991):

$\begin{array}{cc}\left[\begin{array}{c}P\\ S_V\\ S_H\end{array}\right]=\left[\begin{array}{ccc}\frac{{\beta }^2p^2-0.5}{\alpha q_{\alpha }}& \frac{p{\beta }^2}{\alpha }& 0\\ p\beta & \frac{0.5-{\beta }^2p^2}{\beta q_{\beta }}& 0\\ 0& 0& 0.5\end{array}\right]·\left[\begin{array}{c}Z\\ R\\ T\end{array}\right]& \left(4\right)\end{array}$

In equation 4, q_α=√{square root over (α⁻²−p²)} and q_β=√{square root over (β⁻²−p²)}, with α and β the P and S wave speeds respectively.

[0035] Step 34: Convert Frequency-Dependent Velocity to Depth Dependent Velocity

[0036] In order to convert frequency-dependent velocity to depth-dependent velocity, assumptions must be made regarding how waves of different frequencies sample the Earth:

α_eff(f)=∫_λ(f)α(z)·G(z)·dz  (5)

Where G(z) is the sensitivity kernel, α(z) is the true Earth model wave speed, and the integral is performed over a single wavelength λ(f) (See FIG. 7). The sensitivity kernels can vary depending on the order of approximation desired: for example a constant value (G=1) or decaying exponential (G(z)=exp{z/λ(f)}) may be sufficient for low-order solutions. An inversion must be performed to solve equation (5) for α(z). The inversion is done for the velocity model α(z) that best satisfies all α_eff(f). Each event can contribute information at several different frequencies. This inversion can be performed numerically by discretizing G(f,z) and α_eff(f), and inverting the following matrix equation for α(z):

G(f_i,z_j)·α(z_j)=α_eff(f_i)  (6)

[0037] In general, the inversion is nonlinear, as λ(f)=α/f and therefore G(z) will be a function of α.

[0038] The inversion can be generalized to obtain a model for the structure sampled by any number of seismic events, with each independent measurement of α_eff(f) providing an additional constraint on the inversion solution. (As described previously, a “seismic event” can refer either to a seismic disturbance caused by a human controlled seismic source or by earthquake activity.) Further, by assigning effective velocities, i.e. measurements of α_eff(f) identified as being associated with a particular seismic event, to region sampled (or azimuth of incoming seismic energy), two and three-dimensional models can be obtained. Most inversion techniques that are capable of handling nonlinear inversions are suitable for this purpose, for example relaxation or gradient-descent methods.

[0039] The inference of subsurface properties via the method disclosed herein can be further constrained by the addition of additional geophysical constraints, for example gravity measurements via a relationship between velocity and density. Additionally, the inventive method can be extended to time-lapse applications by windowing the data by time such that a velocity model is inferred from data recorded over one time-period and then a separate model is inferred from data recorded over a later time-period.

[0040] Naturally occurring teleseismic (greater than 3000 km away) earthquakes may be used for this analysis. These seismic sources may be preferable as these events have dominant energy in the 0.1 to 5 Hz frequency band, providing resolution complementary to the information contained in reflection data. Years of study by the academic community allow for accurate source location and predicted event arrival times, though small errors in these parameters will contribute to errors in this analysis.

[0041] To identify a particular earthquake in the recorded seismic data, the seismic data may be widowed around the arrival time predicted from a global reference model (for example, IASP91 [Kennett & Engdahl, 1991] or PREM [Dziewonski & Anderson, 1981]) with a Gaussian filter of width 2λ. The reference model also provides the slowness p used in the calculations. A grid-search technique may be used to determine the effective incidence angle, where correlation between the P and S_V components of the seismogram is minimized. A real Gaussian filter of width 0.5 Hz maybe used to bandpass the signal.

[0042] In FIG. 8, results of using the present inventive method on a real seismic event are shown. The left panel shows the velocities obtained via Snell's Law as a function of frequency. The right panel shows an inversion result using these velocities where the inversion is an iterative least-squares relaxation technique.

[0043] The foregoing patent application is directed to particular embodiments of the present invention for the purpose of illustrating it. It will be apparent, however, to one skilled in the art, that many modifications and variations to the embodiments described herein are possible. All such modifications and variations are intended to be within the scope of the present invention, as defined in the appended claims.

• [0044] Abt, .D. L., K. M. Fischer, S. W. French, H. A. Ford, H. Yuan, and B. Romanowicz, “North American lithospheric discontinuity structure imaged by Ps and Sp receiver functions,” J. Geophys. Res., in press (2010).
• [0045] Dziewonski A, and D. Anderson, “Preliminary Reference Earth Model,” Phys. Earth Planet. Int., 25(4) 297-356 (1981).
• [0046] Park, J, F. L. Vernon, and C. R. Lindberg, “Frequency dependent polarization analysis of high frequency seismograms,” J. geophys. Res. 92(B12), 12,664-12,674 (1987).
• [0047] Kennett, B. L. N., “The removal of free surface interactions from three-component seismograms,” Geophys. J. Int. 104, 153-163 (1991).
• [0048] Sheriff, R. E. and L. P. Geldart, Exploration Seismology, p 134-134 (1982).