Overview ta3520

Introduction to seismics
Fourier Analysis
Basic principles of the Seismic Method
Interpretation of Raw Seismic Records
Seismic Instrumentation
Processing of Seismic Reflection Data
Vertical Seismic Profiles
Practical:
Processing practical (with MATLAB)

Signal and Noise

Signal: desired
Noise: not desired
So for reflection seismology:
- Primary reflections are signal
- Everything else is noise!

Signal and Noise (2)

Direct wave: noise

Reflection: (desired) signal

Refraction: noise

Signal and Noise (3)

Direct wave: noise

Multiply reflected : noise
Reflection: signal
Refraction: noise

Signal and Noise for P-wave survey

Desired signal:
primary reflected P-waves

Noise:
direct wave through first layer
direct air wave
direct surface wave
S-wave
Multiply reflected wave
Refraction / Head wave

Signal and Noise for P-wave survey

Signal
Noise

Primary P-wave Reflected Energy

=

All but Primary Reflection Energy

Goal of Processing:
Remove effects of All-but-Primary-Reflection Energy

Processing of Signal
(Primary-reflected energy)

Goal of processing:
Focus energy to where it comes from

Understanding signal and noise:

wave theory
Basic physics underlying signal is captured by wave equation
Ray theory: approximation of wave equation (high-frequency)
Resonances: modes expansion of wave equation
S-waves, P-waves: elastic form of wave equation

Seismic Processing
Basic Reflection and Transmission
Sorting of seismic data
Normal Move-Out and Velocity Analysis
Stacking
(Zero-offset) migration
Time-depth conversion

Basic Reflection and Transmission

(pdf-file with eqs)

shot data

Processing flow

CMP sorting
velocity model
NMO correction
stack
velocity model
(zero-offset) migration

image

Processing
Input: Multi-offset shot records
Results of processing:

1. Structural map of impedance contrasts

2. Velocity model

Sorting:
Common Shot gather
Seismic recording in the field:
Common Shot data
(Each shot is recorded sequentially)
Nomenclature:
- common-shot gather
- common-shot panel

Common Shot gather

Sorting:
Common Receiver gather
Gather all shots belonging to one receiver position in the field
Analysis/Processing: shot variations
(e.g., different charge depths)
(Also in common-shot gathers: receiver variations, e.g.,
geophones placed at different heights)

Common Receiver gather

Sorting

Sorting:
Common Mid-Point gather

Sorting:
Common Mid-Point gather
Mid-points defined as mid-points between source and receiver
in horizontal plane
Since reflections are quasi-hyperbolic:
Seismograms not so sensitive to laterally varying structures
Good for velocity analysis in depth
Stacking successful (noise suppression)
In practice, not really a point but an interval: BIN

CMP gather over structure

Sorting:
Common Offset gather
Purpose:
Very irregular structures (in which stacking does not work)
Application of Dip Move-Out (correction for dip of reflector)
Checking on migration: small and large offsets should give the
same picture : otherwise velocities are wrong
In practice, not really a point but an interval: BIN

Zero-offset gather over structure

Sorting
Common-Mid-Point (CMP) gathers: xs + xr = constant
Common-Offset gathers (COG): xs xr = constant
Multiplicity = Fold:

Nrec
2 xs / xr

Nrec = Number of receivers

xs = Spacing between subsequent shots
xr = Spacing between subsequent receivers

Sorting

shot data

Processing flow

CMP sorting
velocity model
NMO correction
stack
velocity model
(zero-offset) migration

image

Reflection 1 boundary

Normal Move-Out: 1 reflector

T =

R
c

(4d2 + x2)1/2
c
x = source-receiver distance
R = total distance travelled by ray
d = thickness of layer
c = wave speed

We do not know distance, but we know time:

x2
1/2
T = T0 ( 1 +
)
c2 T02
where T0 is zero-offset (x=0) traveltime: T0 = 2d/c

x2
1/2
T = T0 ( 1 +
)
c2 T02
Extra time shift compared to T0 called:

NMO- Normal Move-Out

TNMO

x2
= T T0 = T0 ( 1 + 2 2 )1/2 T0
c T0

Normal Move-Out (NMO)

Normal Move-Out (NMO)

 TNMO

x2
= T T0 = T0 ( 1 + 2 2 )1/2 T0
c T0

Larger TNMO for larger offset

Smaller TNMO for larger T0
(deeper layers have smaller move-out)
Smaller TNMO for larger wave speed c
(deeper layers usually larger velocities so smaller move-out)

Normal Move-Out (NMO)

Input CMP-gather

NMO-corrected CMP gather

(with right velocity)

NMO: effect velocity

NMO with right

velocity

NMO with too small

correction: too high
velocity

NMO with too large

correction: too small
velocity

NMO: 1-layer approximation

TNMO

x2
= T T0 = T0 ( 1 + 2 2 )1/2 T0
c T0

Use Taylor expansion of square root:

TNMO

x2
x2
) T0 =
= T T0 T0 ( 1 +
2
2
2 c T0
2 c2 T0

NMO: 1-layer approximation

NMO: 2-layer

Position depends on
velocities of layer 1 and 2
(via Snells Law)

NMO: 2-layer

(pdf-eqs)

NMO: multi-layer approximation

Velocity model: RMS model

velocity
Interval velocity
(velocity of layer)

time
Root-mean-square velocity
(weighted-average velocity of layers above)

Velocity Analysis
Approaches:
T2 x2 analysis
Alignment of reflectors: visually or mathematical
expression of coherence
With T2 X2 analysis we depend on picking travel-times,
and thus signal-to-noise ratio

Velocity Analysis:
Original CMP gather

Starting CMP gather

2
2
T -x

analysis

Velocity Analysis: aligning

reflectors

Starting CMP gather

Constant-velocity NMO

Velocity too low:

correction too much

Velocity too high:

correction too little

Velocity panels

CMP gather
with velocity
1300 m/s

CMP gather
with velocity
1700 m/s

CMP gather
with velocity
2100 m/s

Velocity as function of time

Velocity panels:real data example

Coherence measures for velocity analysis

Stacked amplitude (normalized or not)
Cross-correlation (normalized or not)
Semblance:
related to cross-correlation

( m A (xm , t , c) )2

m A2 (xm , t , c)

S (t , c) =

A = amplitude

Sum m over traces m in CMP

Semblance for CMP gather

Velocity panels:real data example

Velocity panels:real data example

Velocity panels:real data example

Factors affecting velocity estimation

(Yilmaz, 1988)
Spread length
Stacking fold / Signal-to-Noise ratio
(fold = multiplicity in CMP)
Choice of coherence measure
Departures from hyperbolic move-out
NMO stretch

Effect spread length on velocity estimation

Effect spread length on velocity estimation

Effect spread length on velocity estimation

Effect spread length on velocity estimation

Velocity model: RMS model

velocity
Interval velocity
(velocity of layer)

time
Root-mean-square velocity
(weighted-average velocity of layers above)

Velocity model: RMS model

Time (m-seconds)
CMP location

shot data

Processing flow

CMP sorting
velocity model
NMO correction
stack
velocity model
(zero-offset) migration

image

Applying NMO
Amount x2/(c2 T02) never exactly on a sample:
INTERPOLATION

NMO stretch
(via picture)
Since T0 is smaller:
shift is larger
T
T

Since T0 is larger:
shift is smaller
Hyperbolic shift

NMO stretch
(mathematically)
Due to differential working on T as function of T0:

T0

TNMO =

T0

x2
2 c2 T0

x2
=
2 c2 T02
This is called NMO-stretch

NMO stretch

NMO stretch

Amplitude spectra

Mute: too much NMO-stretch

We do not want too much distortion: setting it zero.
This called muting

NMOstretch on
field data

shot data

Processing flow

CMP sorting
velocity model
NMO correction
stack
velocity model
(zero-offset) migration

image

Stacking
Add traces from NMO-corrected, CMP gather into ONE trace
Number of traces = stack fold
Events that are not hyperbolic, do not add up nicely and
destructively interfere
Goal of stacking : to increase signal-to-noise ratio

Primaries and multiple

primary
multiple

primary