Transcript
7/30/2019 0.7. Seismic Processing
1/18
SeismicProce
ssing
Synthetic Seismogram School
7
Seismic Processing
7/30/2019 0.7. Seismic Processing
2/18
SeismicProce
ssing
2
Introduction
If youre not achieving a good well tie it must be either your ...
Well data
This is something that you can check in OpenWorks.
Seismic Data
This takes a bit more of an investigation.
Many times it is the result of noise!
7/30/2019 0.7. Seismic Processing
3/18
SeismicProce
ssing
3
Propagation
The earth is much more complex then our simple RC model. As the diagram
illustrates, various earth effects are captured in recording. Processing acts to suppress
Most, but not all, of these effects. Those that remain can distort the final trace.
7/30/2019 0.7. Seismic Processing
4/18
SeismicProce
ssing
4
Convolution
The mathematical representation for Convolution is given by:
Seismict = n
s=0 Wavelets.RC t-s
With noise the equation becomes:
Seismict = n
s=0 RCs. Wavelet t-s + Noise
It is the noise term that can seriously degrade the matchbetween your synthetic seismogram (generally noise-
free) and the seismic signal.
7/30/2019 0.7. Seismic Processing
5/18
SeismicProce
ssing
5
Signal to Noise
Gathers with varying signal to noise. The gather on the left is pure signal. The signal-
to-noise is varied to the right from 3 to 0.2. Having gathers with good signal-to-noise iscritical to seismic studies. Gathers give valuable insight into the quality of amplitude
information in stacks and angle stacks.
Decreasing Signal to Noise
7/30/2019 0.7. Seismic Processing
6/18
SeismicProce
ssing
6
Coherent and Random
Coherent Noise includes:
Shot-generated noise such as ground roll
Propeller noise
Diffractions Cable Vibration
etc.
Random Noise includes:
Culture such as traffic
Weather such as rain or wind or lightning
Wildlife
etc.
What Are the Types of Noise?
7/30/2019 0.7. Seismic Processing
7/18
SeismicProce
ssing
7
Pre-Stack Noise Example
Processed Raw
RMS 31 743
Diffracted Multiple
Multiples
Random
The gather on the right has some serious noise
issues. The gather on the left is at the samelocation as the right, after processing for noise
suppression.
7/30/2019 0.7. Seismic Processing
8/18
SeismicProce
ssing
8
Seismic Processing:Pre-Stack Example
For a good seismic-to-synthetic ties, noise must be suppressed as much
as possible! The benefit of suppressing noise is obvious below!
Decreased noise with increased processing
CDP Gathers
7/30/2019 0.7. Seismic Processing
9/18
SeismicProce
ssing
9
Seismic ProcessingPost-Stack Example
A post-stack example of signal-to-noise enhancement. Obviously the section on
the right is superior to that on the left. The improved data will provide a superior
well tie.
Increased post-stack processing
7/30/2019 0.7. Seismic Processing
10/18
SeismicProce
ssing
10
Compare gathers near well with synthetics.(enhance S/N if necessary)
=
-
Final Contractor
Gather
After DSC P + PS
Synthetic
Excellent Tie
Comparison of gather and synthetic
Synthetic Corridor
Stack
Stack from Final Contractor CDP gathers
Stack after FK/Radon and DSC
Example from Norway
Note the improved synthetic-to-seismic
tie after suppression of noise.
7/30/2019 0.7. Seismic Processing
11/18
SeismicProce
ssing
11
The Effects of OverburdenAnisotropy and NMO on AVO
With greater offsets, the effects of anisotropy increase. In the example below, NMO was used to move out the cdps.
The anisotropic gather illustrates that it was insufficient and that higher-order moveout is required. If the amplitude is
measured as a function of constant time (typical among contractors), two very different profiles are obtained. In addition,
the phase change with offset is not identified on the anisotropic gather. Processed gathers should be reviewed for the
effects of anisotropy or inappropriate moveout. Stacking the gathers would give a very different result.
Soft
Hard
Soft
Impedance
- +
Isotropic Anisotropic
=0.3
=0.05
=0.3 =0.05
= 0
= 0
Input
Wavelet
Offset Wider Offset
Amplitude Profiles as a function of offset
0.05
0
7/30/2019 0.7. Seismic Processing
12/18
SeismicProce
ssing
12
Flat
Spot
Fourth order move out
FS2
Interval
Example from Angola, Block 14, at the FS-2 well location: Inline 4806, Xline 3100. Mutes created using
straight-ray approximation of 15 and 35 degree angle-of-incidence are displayed.
15 35
The Effects of Fourth-Order Moveouton the Stacked Section
7/30/2019 0.7. Seismic Processing
13/18
SeismicProce
ssing
13
Flat
Spot
Fourth order move out
FS2
Interval
Example from Angola, Block 14 at the FS-2 well location: Inline 4806, Xline 3100. Gathers muted to
include fourth-order moveout using a straight-ray approxim at ion for the angles. The fourth-order
moveout is the result of anisotropy.
Angle Muted Gathers
7/30/2019 0.7. Seismic Processing
14/18
SeismicProce
ssing
14
Flat
Spot
Fourth order move out
FS2
Interval
Example from Angola, Block 14 FS-2 at the FS-2 Well location: Inline 4806, Xline 3100. Gathers
muted to remove fourth-order moveout usin g a bent-ray estim ate for th e angles.
Angle Muted Gathers
7/30/2019 0.7. Seismic Processing
15/18
SeismicProce
ssing
15
Example from Angola, Block 14 FS-2. Note the area within the white box and compare to that on the
following slide. This product is superior for use in well ties.
Flat Spot
Frequency Spectrum of Stack
Stack of Gathers without Fourth-Order Moveout
7/30/2019 0.7. Seismic Processing
16/18
SeismicProce
ssing
16
Example from Angola, Block 14 FS-2. Although the events look higher in amplitude, they are also
lower in frequency. The effect of the fourth-order moveout is a smearing of the events. Although the
image may look stronger, it is an inferior product for seismic-to-well ties.
Flat Spot
Frequency Spectrum
Stack of Gathers with Fourth Order Moveout
7/30/2019 0.7. Seismic Processing
17/18
SeismicProce
ssing
17
Wavelet Extraction
Angle mutes generated with curved rays from November 2002 velocity model.
0-15 (Near) 25-35 (Far) 40-60 (VFO)
Comparison of the extracted wavelets and angle-of-incidence-limited gathers at
the FS-2 well location. The near and the far offset gathers, when stacked, willretain the embedded wavelet shape (zero phase). Because of residual moveout
on the very-far offsets, the embedded wavelet is distorted.
Zone
of
Extraction
Gathers from G. Delorme
Far
Near
Vfar
7/30/2019 0.7. Seismic Processing
18/18
SeismicProce
ssing
18
Summary
Seismic data need to be properly conditioned for a goodwavelet tie.
Coherent and random noise should be suppressed.
Stack and pre-stack data should be carefully reviewed.
Understand the l im i tat ions im posed by the seismic data!
top related