Sunteți pe pagina 1din 9

1

Chapter 4
Seismic Signal and Noise

The goal of this chapter is to be able to distinguish signal from noise on the seismic
record.

In addition, you will be able to distinguish the characteristics of different types of


seismic noise, their causes, and methods to attenuate them.

Furthermore, you will learn about types of seismic wavelets and seismic resolution.

Seismic signal - Primary reflections

A primary reflection is the first reflection of the source wavelet from an interface.

A primary reflection is basically the source wavelet after multiplying it by the


reflection coefficient and displacing it by a time equal to the TWTT from the ground
surface to the interface.

Primary reflections carry useful information about the velocities and thicknesses of
subsurface layers.

Primary reflections are considered the desired signal in the seismic section, while
everything else is considered unwanted noise.

We have already studied primary reflections in detail in Chapter 2.

Seismic noise

Signal is any event on the seismic record from which we wish to obtain subsurface
information. Primary reflections are considered signal.

Noise is everything in the record that we did not want to record.

The signal-to-noise (S/N) ratio is the ratio of the signal energy in a specific portion of
the record to the noise energy in the same portion.

On poor records, S/N ratio is < 1.

There are two types of seismic noise:


Coherent noise which can be followed across at least a few traces. It includes
surface waves (ground roll), direct waves, refractions, diffractions, and multiples.
Incoherent noise which is random on all traces. It includes noise generated by
near-surface scatterers, wind, rain, humans, and machines.

Stacking usually suppresses most of the coherent and incoherent noise.

Stacking means summing all the traces in an NMO-corrected common midpoint


(CMP) gather to generate one trace.

If we stack M traces, the S/N ratio will be enhanced by about

M times. Therefore,

stacking 100 traces enhances the S/N by 10 times.

Direct wave

It is the P-wave that travels directly from the source to the receiver along the ground
surface.

The time-distance equation of the direct wave (for an inline source) is:
T = X/V1,

(1)

where T is the one-way traveltime from the source to the receiver at offset X and V1
is the velocity in the first (uppermost) layer.

Equation (1) represents a straight line with slope = 1/V1 and intercept = 0.

The direct wave always arrives before the reflected P-wave in the first layer. (Why?)

The direct wave is usually the first arrival at near offsets. The head wave usually
becomes the first arrival on far offsets (after the crossover distance).

It is usually attenuated by muting (after NMO correction) in the upper part of the
record and by stacking in the rest of the record because it has a linear moveout
(LMO) while a primary reflection has a hyperbolic moveout (NMO or DMO).

Figure: Direct wave on real seismic records.

Head wave (refraction)

Head waves are generated when the angle of incidence equals the critical angle (c).

The critical angle is given by: c = Sin-1(V1/V2), where V1 and V2 are the velocities in
the incident and refraction media, respectively.

A necessary condition for head wave generation is that the velocity in the refraction
medium (V2) be greater than that in the incident medium (V1).

The head wave travels in the refraction medium (with V2) along the interface
(Fermats principle) emitting seismic energy into the incident medium (Huygenss
principle) at an angle equal to the critical angle (Snells law).

The time-distance equation (for an inline source) for a single horizontal refractor is:
T = T0 cosc + X/V2,

(2)

which represents a straight line of slope = 1/V2 and intercept = T0 cosc; where
T0 = 2H/V1 is the TWTT at X = 0 and H is the thickness of first layer.

The head wave will not be observed at offsets less than the critical distance (Xc):
Xc = 2H tanc.

(3)

The head wave will become the first arrival (i.e., arrive before the direct wave) after
the crossover distance (Xo):
Xo = T0 cosc/(1/V1 1/V2).

(4)

Refractions are usually attenuated by muting (after NMO correction) in the upper part
of the record and by stacking in the rest of the record because they have linear
moveouts while primaries have hyperbolic moveouts.

Figure: Head wave on real seismic records.

Movie of spherical-wave refraction (Courtesy of J. Barker)

Ground roll

They are surface (Rayleigh) waves traveling along the ground surface.

They have generally low velocities (100-1000 m/s), low frequency (< 15 Hz), and
high amplitudes.

Their time-distance curves are straight lines with low velocities and zero intercepts.

There might be several modes of ground rolls in the record because of their dispersive
nature (i.e., different frequency components travel with different velocities).

They are attenuated using source and receiver arrays in the field and various
processing methods (e.g., frequency filtering, f-k filtering).

Figure: Ground roll on a real seismic record.

Diffraction

Diffraction occurs when the wavefront encounters the edge of a reflector or any
obstacle whose size is of the same size as the wavelength.

Snells law does not apply in the case of diffraction because the incident wave gets
spread out in all directions.

The T-X curve of a diffraction is a hyperbola whose apex is located at the projection
of the diffracting point on the seismic profile and has a TWTT corresponding to the
normal distance of the diffracting point from the seismic profile (Figure).

Diffraction moveout (hyperbolic curvature) will vary depending on the geometry of


the source and receiver (Model of a diffraction).

The amplitude of a diffraction is highest at its apex and decays rapidly as we get away
from the apex. This property is useful in recognizing diffractions from reflections
only when true amplitudes are preserved because reflections do not change
considerably.

Diffractions are attenuated using seismic migration during seismic data processing.

Figure: Real data examples.

Multiples

They are events that have undergone more than one reflection from the same
interface.

The T-X curve of the Nth multiple is a hyperbola given by:

X2
T ( X ) ( N 1) T 2 ,
V
2
N

2
0

where T0 and V are the zero-offset TWTT and velocity within the layer.

The amplitude of the Nth multiple is related to the reflection coefficients at the top
(RT) and bottom (RB) of the multiple-generating layer as follows:

(5a)

RN RTN RBN 1 .

(5b)

Note that N=0 in equations (5) indicates the primary.

There are two types of multiples (based on their paths): short-path (period) multiples
and long- path (period) multiples.

A short-path multiple is generated close to the primary so that it interferes with the
primary reflection from the same reflector. Its effect is to lower the vertical
resolution.

A long-path multiple is generated far from the primary so that it is totally separated
from the primary reflection from the same reflector. Therefore, it appears as a
separate event on the seismic record.

Short-path multiples
Peg-leg: It is a multiple that has been reflected successively from the top and base
of a thin layer on its way down to or up from a deeper reflector. They effectively
lower the signal frequency with time (Figure, courtesy slb.com).
Ghost: it is a multiple reflected from the base or surface of the weathering layer in
land surveys or water surface in marine surveys (Figure, courtesy slb.com).
It involves a 180 phase shift (i.e., polarity reversal).
Its effect on the waveshape depends largely on the source depth below the
ghost-generating surface.
In marine surveys, the optimum depth for sources and receivers is 10-15 m
because the ghost and primary will interfere constructively for frequencies of
20-40 Hz, which is the normal seismic range. (Analysis)

Water reverberation: it is produced by multiple reflections in a shallow-water


layer. It is a strong reflection that dominates the whole record and makes the
picking of primaries difficult (Figure, courtesy slb.com).

Long-path multiples
The strongest long-path multiples involve water reverberations in a deep-water
layer (Figure) and reflections at the base of a thick weathering layer (Figure).
They cause serious problems especially when the amplitude of the multiple is
strong and the amplitude of a primary arriving at the same time is weak
(Example).
Multiples usually exhibit more NMO (curvature) than primary reflections with the
same traveltime because of the general increase of velocity with depth.

Multiples are distinguished and removed using:


Their high NMO.
Their low velocities.
Predictive deconvolution during seismic data processing.

Summary: T-X curves of various seismic events.

Shape of seismic wavelet

The ideal wavelet is a zero-width spike located at the reflection coefficient time.
Practically, the best we can attain is a narrow wavelet with minimal sidelobes.

The Ricker wavelet, W(t), is a good approximation of a zero-phase seismic wavelet:


( )

where 0.1 < k < 1 and fd is the dominant (peak) frequency.

(6)

Most impulsive sources (e.g., dynamite) produce minimum-phase wavelets (i.e., ones
whose energy is concentrated at the start).

Ricker wavelet is commonly used for modeling if the actual seismic wavelet cannot
be extracted from seismic data that was generated by an impulsive source.

A Vibroseis produces ultimately a zero-phase Klauder wavelet (see Chapter 5).

Figure.

Seismic resolution

It is the minimum separation required to distinguish two closely-spaced objects.

The resolution problem occurs because seismic wavelets have finite lengths and
tend to interfere when they get close to each other.

In seismic studies we can define two types of resolutions: vertical and horizontal.

Vertical resolution

It refers to the minimum separation (in time or depth) between two interfaces
required to show two separate reflectors.
In general, we can distinguish two events vertically depending on: events
separation, wavelet dominant frequency, and reflection coefficients.

Rayleighs criterion states that two wavelets of wavelength (period T) can be


resolved only if the separation between them is greater than /4 (T/4) (Figure).

A thin bed is a bed whose thickness is less than /4 (T/4).

Tuning is the change in the peak amplitude of two closely spaced wavelets as the
separation between them increases (Figure).

Tuning is an important modeling tool in seismic interpretation studies.

Horizontal resolution

It refers to the minimum horizontal distance between two features on an interface


required to distinguish them as two different features on the seismic record.

The limiting horizontal resolution on unmigrated sections is the effective Fresnel


zone.

Fresnel zone:
Fresnel zone is defined as the area from which reflected energy arriving at a
detector has phases differing by no more than half wavelength (period)
(Figure).
Therefore, reflections come from an area rather than a point on the interface.
Most of the reflected energy comes from the first Fresnel zone circle.
For a spherical wave, the effective radius of the first Fresnel zone is:

R1 12 Z ,

(7)

where is the wavelength and Z is the depth to the reflector.


For a plane wave, the effective radius of the first Fresnel zone is:
R1 Z .

The smaller the first Fresnel zone the better is the horizontal resolution
because properties within the Fresnel zone are averaged into one reflection.

(8)

S-ar putea să vă placă și