Depth Conversion of Time Interpretations Volume Models

Depth Conversion
Depth Conversion
of Time
Interpretations
~
Volume Models
Depth Conversion
Based on the different types of velocity models that can be
derived from well data produce a ranked list of approaches
to depth conversion with the simplest, least accurate at the
top and most accurate at the bottom.
12.1
Depth Conversion
Long Period Static Anomalies

Before we begin depth conversion
it is necessary to recognise, and
correct, any long wavelength static
anomalies in the time data.
These anomalies will probably be
seen in the times sections, time
map(s) and possibly the stacking
velocity sections or maps. The
problem and solution was first
discussed by Booker et al, 1976,
then by Pickard 1992, Musgrove
1994 and Armstrong et al 2001.
No LVL Static Correction
After Musgrove, 1994, Time Variant Statics

Corrections During Interpretation, Geophysics v. 59,
no. 3, p. 474.
Depth Conversion

The time delay due to near
surface anomalies is estimated
from the regional residual
separation of time delays on
shallow reflectors. For deep
anomalies it may be estimated
from well depths and velocities.
The width of the time distortion
at the target horizon is derived
from the width of the velocity
anomaly, its depth and the
target horizon depth. Fresnel
zone effects are often ignored.
From Armstrong et al, 2001, Removal of overburden

velocity anomaly effects, Geophysical Prospecting v.
49, no. 1, p. 79.
12.2
Depth Conversion

The technique then simulates
the CMP stack at the target
horizon by modelling the time
delay on each of the traces in
the CMP gather with respect to
distance along the seismic line.
This step requires a knowledge
of the mute pattern at the time
of stacking velocity analysis
(just as the bias correction did).

49, no. 1, p. 79.
Tx2 = To2 + x2 / VRMS2 - x4(VRM44 - VRMS4) / 4 To2 VRMS8

with
VRM44 = VIi4ti / to.
Depth Conversion

The time delay on the
stacked traces (the
required correction) is
then found from the
time axis intercept of a
least squares best fit
trend line of the time
delay on the different
traces in the CMP
gather plotted against
the offset squared.

49, no. 1, p. 79.
12.3
Depth Conversion
Ranked Methods for Functions

The least accurate methods are at the top and most accurate at the bottom.
1. Constant average velocity.

2. Mapped average velocity.
3. Average velocity function.
4. Instantaneous velocity function.
5. Instantaneous velocity function with mapped parameter.
6. Constant interval velocities.
7. Mapped interval velocities.
8. Interval or instantaneous velocity functions.
9. Interval or instantaneous velocity functions with one
parameter mapped.
10. Interval or instantaneous velocity functions with all
parameters mapped.
One Function
Depth Conversion
Depth conversion may use a

single velocity function from
the surface down to the layer
of interest.
Fast
Less accurate
0
1
T (sec)
An average velocity function

or an instantaneous velocity
function.
4
12.4
5000 Vi (ft/sec) 20,000
One Function
Depth Conversion
One Function
Depth
Map
Time
Map
Velocity
Function
Velocity
Map
Depth conversion process
One Function
Depth Conversion
Constant Average Velocity

Depth
Time
After Marsden, Layer cake depth conversion, Leading Edge, January 1989.
Apparent closed area depends of choice of contour

interval with respect to spill points in flat areas.
12.5
One Function
Depth Conversion
Constant Average Velocity with External Drift

The structure in depth
map based on well control
fails to represent the true
structural picture.
The well data is used to
provide a constant average
velocity based on a time
depth plot.
Example after Laurtent Moinard, Application of Kriging to the Mapping of a Reef from Wireline Logs and Seismic Data : a Case
History, in Geostatistical Case Studies, G. Matheron and M. Armstrong (editors) 1987, D. Reidel Publishing Co.
One Function
Depth Conversion

Structure in time map;
plenty of detail due to
abundant seismic control.
This map is used to derive
the semivariogram. A
plane least squares
surface was used as the
drift so that the
semivariogram is derived
from time residuals.
Example after Laurtent Moinard, Application of Kriging to the Mapping of a Reef from Wireline Logs and Seismic Data : a Case
History, in Geostatistical Case Studies, G. Matheron and M. Armstrong (editors) 1987, D. Reidel Publishing Co.
12.6
One Function
Depth Conversion

The depth map produced
from the time surface
using the constant
average velocity and the
semivariogram. This depth
map follows the shape of
the time map but departs
from it in the vicinity of the
well locations where it
matches the measured
depths.
Example after Laurtent Moinard, Application of Kriging to the Mapping of a Reef from Wireline Logs and Seismic Data : a
Case History, in Geostatistical Case Studies, G. Matheron and M. Armstrong (editors) 1987, D. Reidel Publishing Co.
One Function
Depth Conversion
Summary - Single Function

Depth conversion by
a single function is
well suited to areas
with dense well
control and simple
structure.
25 km
Leman field with production platforms
12.7
Depth Conversion
Single Function
In the marine environment we may be tempted to use a single layer
for depth conversion when the water layer appears to be relatively
uniform and the depth to the first interface appears to be relatively
deep. There is one anomalous well data point.
Top Sele
0.88
5950
One Way Time sec

0.9
0.92
0.94
TVDSS ft
6000
6050
6100
6150
6200
6250
6300
6350
Depth Conversion
Multiple Functions
In the marine environment if we separate out the water layer from
the underlying Tertiary we will obtain a much better function. The
figure shows the same formation as the previous slide with the
water layer removed. The scatter is reduced to give a better result.
isopach ft
Undifferentiated Tertiary
5900
5850
5800
5750
5700
5650
5600
5550
5500
5450
0.780 0.790 0.800 0.810 0.820 0.830 0.840 0.850
Isochron sec
12.8
Multiple Functions
Depth Conversion
Multiple Functions
A multi-layer approach
should be used in areas
where the overburden
displays lateral velocity
inhomgeneities, i.e. the
velocity structure is not
simple. Each of a number of
layers are then represented
by different functions.
5000 Vi (ft/sec) 20,000
T (sec)
Slower
Increased accuracy ?
Multiple Functions
Depth Conversion
Multiple Functions
Isochrons
Velocity
Functions
Isochores
Layers
Time
Maps
Depth conversion process

12.9
Depth
Maps
Average
Velocity
Maps
Multiple Functions
Depth Conversion
Multi-Layer Example
Time to upper surface
From a Sattlegger brochure
Multiple Functions
Depth Conversion
Multi-Layer Example
Depth to upper surface
12.10
Multiple Functions
Depth Conversion
Multi-Layer Example
Map of Vo coefficient
from Fausts equation
Vi = Voz1/n (n=3)
Multiple Functions
Depth Conversion
Multi-Layer Example
Map of Vo coefficient
after smoothing with a
16th order polynomial
12.11
Multiple Functions
Depth Conversion
Multi-Layer Example
Time to lower surface
Multiple Functions
Depth Conversion
Multi-Layer Example
Lower surface depth
converted using
Vi = Voz1/n (n=3)
12.12
Multiple Functions
Seismic
horizon
times
Checkshot
or VSP
Depth Conversion
Sonic log
Velocity log
Macrovelocity
model
Depth maps
Integration
Interval,
average,
instantaneous
Velocity
Maps
Analytic
functions
Depth
conversion
Z.O. or
image ray
modelling
Compare
Multiple Functions
Depth Conversion
Summary - Multiple Functions

Depth conversion by multiple
functions is well suited to
areas with moderate well
control and moderate
structural complexity.
The functions will account for
vertical gradients and rapidly
changing bed thicknesses.
1 km
When we have little well control then we have to make use of

seismic velocities to interpolate the well velocities.
12.13
Multiple Functions
Depth Conversion
Summary - Multiple Functions

Multi-layer depth conversions essentially predict variations in the
average velocity that cannot be handled by a single function
M ulti-layer Depth Conversion
10000
D epth (ft)
9000
Observed
8000
7000
Predicted multi-layer
6000
5000
0.4
0.5
0.6
0.7
0.8
0.9
Time (sec)
Depth Conversion
12.14
Exercise 12.1
Depth Conversion
You have to recommend a well on the structure seen at about 1.7 secs., on the
accompanying seismic section. This is a wildcat area with few wells having been
drilled. Make your depth prognosis using the function VA = 5000 + 2500t where t is the
one way time in seconds.
This function comes from good scout information which you trust. Your supervisor is
not so comfortable however and wants you to give an estimate of the error in your
depth conversion.
Make an initial guess at how accurate you think your depth prognosis is.
List the potential sources of error and assign estimates to the magnitude of each.
12.15
Depth Conversion
Exercise 1
12.16
Velocity Grids
Depth Conversion
C.I. = 100 m/s
~
Grid Models
0
8 miles
Average Stacking Velocity
Velocity Grids
Depth Conversion
How might we use seismically derived velocities for depth

conversion?
12.17
Data courtesy of Amoco (U.K.) Ex. Co.
Depth Conversion
of Time
Interpretations
Exercise 12.2
Depth Conversion
Now suppose that no well velocity information was available to you. The
only velocity data are stacking velocity functions every 2 km along the line
and they were derived without the benefit of DMO.
What is your depth prognosis now given the two nearest stacking velocity
functions?
SP 253
Time VS
msec m/s
0
152
384
601
859
1401
1756
2151
2621
1472
1472
1717
1865
2070
2317
2441
2616
3390
SP 155
VIS
m/s
Time VS
msec m/s
VIS
m/s
1472
1861
2102
2483
2662
2879
3283
5725
0
165
439
744
968
1438
1713
2045
2572
1478
1891
2081
2367
2525
3463
3661
5688
1478
1478
1747
1891
2011
2192
2441
2677
3511
Note:
Dips are relatively gentle so any dip correction will probably do more harm than
good.
The data are relatively old and were probably acquired with a cable short enough
that the bias correction would make no appreciable difference to the results.
How accurate do you suppose this depth conversion is?

12.18
Velocity Grids
Depth Conversion
Sparse Well Control

When there is only sparse
well control we usually
generate grid velocity
models from the seismic
data. By calibrating the
grids to the well velocities
we are making use of the
grids to interpolate the well
velocities.
In unexplored basins we
dont always have any wells
to interpolate or extrapolate
from.
Velocity Grids
Depth Conversion
Substitutes
5000 Velocity - ft/sec 10,000
Bias Corrected Models
0
Interval Velocity m /s
1000
2000
3000
4000
5000
6000
4
1000
6
Depth m
Depth - x1000 feet
2000
8
10
4000
12
5000
14
16
3000
6000
Average
Velocity
Stacking
RMS
Velocity Velocity
12.19
Original Model
Ray Trace + Semblance Analysis
Bias Corrected
Velocity Grids
Depth Conversion
Ranked approaches to depth conversion with seismic

velocities.
1. Kriging with seismic velocities and well velocities
* these approaches have constraints
2. Estimate average velocity from interval RMS velocities
* can be done without well control
3. Use seismic velocities to augment well data in deriving
functions
Velocity Grids
Depth Conversion
For Kriging:
Histogram of
velocities to be
kriged should show
a normal
distribution.
Enough data points

are needed to
produce a
reasonable
variogram (a
minimum of 8 or 10).
From a GX Technology brochure
12.20
Cokriging, kriging
with external drift
etc., require a good
linear correlation
between the different
parameters.
Velocity Grids
Depth Conversion
For Kriging
Stacking,
RMS or
Average
Velocity
Simple or
Common
Kriging
Smoothed
Velocity
Final
Velocity
Kriging
with
External Drift
Cross Plot
with Well
Velocities
Velocity Grids
Depth Conversion
Regional/Residual Calibration
A conventional
horizon oriented
stacking velocity
map. This map
can be smoothed
first by Kriging.
From Francis, Geostatistical Applications in Asset Valuation Uncertainty, PETEX 94.
12.21
Velocity Grids
Depth Conversion
The variogram from the stacking velocities. The noise
seen in the map produces the large nugget.
1000
Range
Variance (m2)
Sill
750
500
250
Nugget
0
0
5000
10000
15000
20000
25000
Sample Separation (m)

Velocity Grids
Depth Conversion
Well velocities cokriged with the drift
supplied by the
seismic velocities.
This is an average
velocity map to the
horizon of interest
that ties the well
control and
honours the trends
in the seismic
velocities.
12.22
Depth Conversion
Case History / Example
GOM: Mississippi Canyon
Calibrate Stacking Velocities with VSP Data
Create Depth Map for the 10-5 Sequence
Time Horizon
Quantitative
Geosciences, LLP
Depth Conversion
Data:
77,000 seismic stacking velocities
X = Y = 2000 ft CDP spacing
Z (time in ms) = variable (5 15 picks)
2 wells with VSP time-velocity-depths
10-5 Sequence travel times
23,837 travel time values
Grid mesh: 1000 x 1000 ft
Quantitative
Geosciences, LLP
12.23
Depth Conversion
Stacking Velocity and VSP Locations
Quantitative
Geosciences, LLP
Depth Conversion
10-5 Sequence Time Structure
Quantitative
Geosciences, LLP
12.24
Depth Conversion
Calibration
VSP Two-Way Time
VSP Two-Way Time
VSP Average velocities and Stacking velocity functions at

the well locations.
VSP Average Velocity
Stacking Velocity
Quantitative
Geosciences, LLP
Depth Conversion
Calibration
Calibrated and uncalibrated stacking velocity functions
Calib Stacking Velocity
Uncalib Stacking Velocity
Note the linear relationship between the uncalibrated velocities.
VSP Velocity
VSP Velocity
Quantitative
Geosciences, LLP
12.25
Depth Conversion
Variograms in three directions
Quantitative
Geosciences, LLP
Depth Conversion
Deterministic Velocity Cube from Kriging
Quantitative
Geosciences, LLP
12.26
Depth Conversion
10-5 Sequence Average Velocity
Quantitative
Geosciences, LLP
Depth Conversion
Calibrated Velocity Cube with Time Horizon
10-5 Time Horizon
Quantitative
Geosciences, LLP
12.27
Depth Conversion
10-5 Sequence Depth Map
Quantitative
Geosciences, LLP
Depth Conversion
Work Flow
Quantitative
Geosciences, LLP
12.28
Velocity Grids
Depth Conversion
Kriging 1 many wells

This approach is best if the seismic velocities are noisy.
Stacking,
RMS or
Average
Velocity
Variogram
Well
Average
Velocity
Variogram
Cokriging
Velocity Grids
Simple
Kriging
Smoothed
Seismic
Velocity
Cross Plot
Time
Map
Final
Velocity
Deterministic
Depth
Map
Depth Conversion
Kriging 1 few wells

This approach is best if the seismic velocities are noisy.
Stacking,
RMS or
Average
Velocity
Variogram
Well
Average
Velocity
Kriging
with
External Drift
12.29
Simple
Kriging
Smoothed
Seismic
Velocity
Cross Plot
Time
Map
Final
Velocity
Deterministic
Depth
Map
Velocity Grids
Depth Conversion
Kriging 2 many wells

Will produce unreliable results with noisy seismic velocities.
Stacking,
RMS or
Average
Velocity
Variogram
Well
Average
Velocity
Cross Plot
Variogram
Cokriging
Time
Map
Final
Velocity
Velocity Grids
Deterministic
Depth
Map
Depth Conversion
Kriging 2 few wells

Will produce unreliable results with noisy seismic velocities.
Stacking,
RMS or
Average
Velocity
Well
Average
Velocity
Variogram
Time
Map
Cross Plot
Kriging
with
External Drift
12.30
Final
Velocity
Deterministic
Depth
Map
Velocity Grids
Depth Conversion

velocities.
functions
Velocity Grids
Depth Conversion
For Average Velocity

Stacking,
RMS
Velocity
Interval RMS
Velocity
Robust Filter,
Smooth
Calibrate to
Well Interval
Velocities
Robust Filter,
Smooth
Robust Filter,
Smooth
Interval RMS
Velocity
Calibrate to
Well Interval
Velocities
Calibrate to
Average
Velocity
Calibrate to
Average
Velocity
Calibrate to
Well Interval
Velocities
Robust Filter,
Smooth
12.31
Final
Average
Velocity
Velocity Grids
Depth Conversion
Calibration
Before depth conversion it is necessary to calibrate the
seismically derived velocities to the velocities measured in
the wells.
Calibration is the process of using the abundant seismic
estimates, which probably reflect regional geological
variations, to interpolate and extrapolate the sparse well
control which aliases the geological variations.
The result of calibration is that we have velocities in our
model which honour the well measurements and display the
spatial sampling of the seismic data.
Velocity Grids
Depth Conversion
Calibration
The calibration is frequently performed in two steps.
The first is a regional calibration which takes care of any shift
between the trends of the two data sets.
The second is a residual calibration which accounts for the
local variations, the residuals, after the first calibration, and
ensures the wells are tied exactly.
12.32
Velocity Grids
Depth Conversion
Regional Calibration
We can correlate the VSP
or checkshot data to
stacking velocity (or VRMS
or VAS) functions. Linearity
is not required. Correlated
data points must relate to
the same points in the
subsurface, i.e. we need
the VRMS value at the same
travel time as the observed
VA.
Well v. Seismic Average Velocity

7000
6500
w ells
6000
5500
5000
4500
4500
seism ic
5000
5500
6000
6500
7000
7500
y = 3.7050934E-07x 3 - 6.7773959E-03x 2 + 4.1711981E+01x 8.0315444E+04
This approach correlates

the whole velocity volume.
Velocity Grids
Depth Conversion
Level 8 Interval Velocities
8000
7000
6000
5000
Well
To calibrate the stacking

velocities (or VRMS or VAS) to
a particular horizon a
percentage calibration
factor is required or used
(93.35% in the example).
4000
3000
2000
A percentage calibration
factor is equivalent to a
linear trendline being fitted
to the data, which goes
through the origin.
1000
0
Seism ic
2000
4000
IntVel = 0.9335Vdix
12.33
6000
8000
Velocity Grids
Depth Conversion
For interval velocity the
interval stacking (or RMS)
velocities are cross plotted
against the corresponding
well velocities.
Level 9 Interval Velocities

8000
7800
Well
7600
A good correlation is
usually observed but the
trend does not always go
through the origin.
7400
7200
7000
6800
7400
Seism ic
7600
7800
8000
8200
8400
8600
IntVel = 0.9662Vdix - 329.57
Velocity Grids
Depth Conversion
After calibration the trend
does go through the origin.
As with checkshot function

misties these misties are a
measure of the overall
accuracy of the method.
7000
6500
Seism ic
The remaining scatter in

the data means that none
of the wells will be tied
exactly.
Calibrated Average Velocities
6000
5500
5000
4500
4500
5000
5500
6000
Well
12.34
6500
7000
Velocity Grids
Depth Conversion
Misties
After regional calibration the misties at the wells are greatly reduced.
Data from a Paradigm Geophysical brochure
Velocity Grids
Depth Conversion

velocities.
functions a hybrid approach
12.35
Velocity Grids
Depth Conversion
We now have either a calibrated volume of seismically

derived velocities or a number of calibrated interval velocity
grids associated with different horizons which are used to
derive an average velocity to the horizon of interest.
Velocity functions from the calibrated volume may be treated
as additional checkshot values and used to augment sparse
checkshot data to derive analytical functions.
Velocity Grids
Depth Conversion
Determining K and V0 From Seismic

Given an estimate of the VRMS curve, obtained by the
correction of VS, estimates of analytical function parameters
can be obtained directly from the seismic data and mapped.
K
From Arnaud et al, K coefficient

determination of an interval velocity
law Vo + Kz from stacking velocity
analyses, EAEG-95 Workshop
Depth Conversion
V0
12.36
Velocity Grids
Depth Conversion

Vrms2 = [VI2.t] / [t] from our definitions
Vrms2 = [ Vi2.dt] / [ dt] and Vi = dz / dt
Vrms2 = [ (dz / dt)2.dt] / [ dt]

Vrms2 = [ (dz / dt).dz] / [ dt]
Substitute any expression for dz/dt and integrate.
Velocity Grids
Depth Conversion

For a single layer model:
For Vi = V0 + Kz,
VRMS2 = V02(e2Kt - 1)/2Kt
For Faust VRMS2 ={ nK/(n + 1)t}{(n - 1)Kt/n}[(n + 1)/(n - 1)]

For Evjen VRMS2 ={ V0K/(1 + n)t}{[1 + V0(1 - n)Kt] [(1 + n)/(1 - n)] - 1}
It is therefore possible to use the corrected VRMS data to map
directly the parameters of standard analytic functions.
12.37
Velocity Grids
Depth Conversion

It is also possible to use the
calibrated interval and
average velocity values as
checkshots thus permitting a
wider range of values to be
used in determining the
analytical function
parameters.
VELOCITY
DEPTH
After Marsden et al, Leading Edge, 1995
Velocity Grids
Depth Conversion
Summary - Grid Models

Grids of seismically derived RMS velocities can be used when
we have no well control or sparse well control.
We can use kriging to smooth the generally noisy grids and tie
them to the well control. This approach usually yields better
results than the more traditional ways of smoothing the noisy
grids.
The velocities have to be calibrated in some way to well
velocities.
The velocities may also be used to augment well checkshot
data.
12.38
Depth Conversion
Tying
the Well
Control
PGS Reservoir (U.S.) Inc. Doe Contract #DE-AC-22-94-PC 91008
Residual Calibration
Depth Conversion
Residual Calibration - Tying Well Control

We have our preliminary depth map that does not tie the
well control exactly. We have analysed our misties and
quantified the accuracy of our depth conversion.
How might we make the map tie the well control?
What are the disadvantages of the different methods?
12.39
Depth Conversion
Tying Well Control

How might we make the map tie the well control? What are the
disadvantages of the method?
For multi-layer depth conversion only tie the target horizons.
Intermediate errors tend to cancel out.
If the errors are random, distribute over an area whose radius
is half the average well spacing. Can produce bulls-eyes at the
wells.
Autocontour the errors. Produces unreasonable gradients and
error values outside the limits of well control.
Kriging. Will separate trend and random components of the
errors.
Depth Conversion
One-Step Calibration - Wells

A popular approach when using checkshot data from
multiple wells:
Fit a simple (linear?) function
Fix all but one of the parameters
Vary the one parameter to effect a tie to each data point
Map the variation of the parameter
Use the parameter grid in depth conversion
12.40
Depth Conversion
Interval Velocity (m/sec)

2000
Fit used
Slope (K) = 0.73
1900
1800
1700
1600
0
100
200
300
400
500
One Way Time (msec)
Keep the slope K constant and derive a Vo for each data

point..
Data courtesy of Amoco
Depth Conversion

Map of Vo from
previous plot
1900
1800
1700
1600
10 Miles
Data courtesy of Amoco
CI = 50 m/sec
What are the advantages and disadvantages of this

approach to macrovelocity model building?
12.41
Depth Conversion
One-Step Calibration - Seismic

The regionally
calibrated seismic
velocities display
variations of geological
significance even
though they do not tie
points of well control
with the desired
accuracy.
C.I. = 100 m/s
8 miles
Average Stacking Velocity
Depth Conversion

Velocity maps based
on well control will tie
the wells, more or less
exactly depending on
the contouring
algorithm, but alias the
geological trends.
C.I. = 100 m/s
8 miles
Average Velocity from Wells
12.42
Depth Conversion

The calibration factor or
residual is determined at
each well control point.
C.I. = 2 %
Note that steep gradients

can be introduced which
may not be geologically
reasonable.
These values are then

gridded to determine the
values to be applied to the
seismic grid.
8 miles
Calibration Factor
Depth Conversion

C.I. = 100 m/s
The calibrated interval

velocity map ready for
use in depth conversion.
Kriging can be used

instead of this traditional
approach.
8 miles
Calibrated Average Velocity
12.43
This map ties the well

control and honours the
trends seen in the
seismic data.
Depth Conversion
One-Step Calibration
Gridding the calibration
factor or residual from
each well control point can
produce undesirable
trends and steep gradients
when dissimilar values
occur in closely spaced
wells.
Data from a Paradigm Geophysical brochure
Depth Conversion
Random Errors
When the residuals are small and
random then the errors are
dispersed over an area with a
radius of up to one half the
average inter-well spacing. This
approach is only acceptable when
there is no spatial correlation
between the residuals.
12.44
Depth Conversion
Random Errors
Mistie grid.
Mistie values
when flexing
surface over too
small a radius
around wells.
Depth Conversion
Random Errors
Tied map.
When the radius
of flexing is too
small then the
circular nature of
the flexing will
show in the final
map.
12.45
Depth Conversion
Random Errors
Mistie grid.
Mistie values
when flexing
surface over a
distance of about
half the average
well spacing.
Depth Conversion
Random Errors
Tied map.
The circular
nature of the
mistie contour
values does not
show up.
12.46
Depth Conversion
Random Errors
Mistie grid.
Minimum
curvature gridding
with bicubic
interpolation
of mistie values.
The extrapolation
is geologically
unreasonable.
Depth Conversion
Random Errors
Mistie grid.
Inverse distance
weighted gridding
with bicubic
interpolation
of mistie values.
Produces almost
exactly the same
correction grid as
the previous
example.
12.47
Depth Conversion
Kriging
A semivariogram of the residuals
is a powerful tool that will find
any spatial correlation.
Kriging residuals results in a
calibration grid or error grid that
shows both any remaining possibly undetected - trend and
the true random residual error.
The effect is a series of bullseyes in a regional smooth trend.
Depth Conversion
Kriging
A semivariogram of the residuals
is a powerful tool that will find
any spatial correlation.
Kriging residuals results in a
calibration grid or error grid that
shows both any remaining possibly undetected - trend and
the true random residual error.
The effect is a series of bullseyes in a regional smooth trend.
12.48
Examples
Depth Conversion
Structure-in-time Map
Examples
Depth Conversion
Depth conversion using five layers and seismic interval

velocities, the one step calibration method was used.
12.49
Examples
Depth Conversion
Depth conversion by average velocity.
Examples
Depth Conversion
Depth conversion using six analytic functions based on wells.
12.50
Velocity Grids
Seismic
horizon
times
Seismic
Velocities
Depth Conversion
Well
Velocity
Dip correct,
Interpolate
Invert,
(Dix/Bias)
Macrovelocity
model
Depth maps
Velocity
Maps
Edit,
Smooth
Calibrate
Interval,
Average,
velocity
Analytic
functions
Depth
conversion
Z.O. or
image ray
modelling
Compare
Depth Conversion
Quality Control
Depth converting the same structure-in-time map by
different methods can result in different depth maps even
though all of the well control is honoured.
So how do we know which depth conversion is the most
accurate?
A final quality control step on our depth conversion, one
that is rarely applied, should be to model our seismic data
by ray tracing through our macrovelocity model.
12.51
Depth Conversion
Forward Modelling
Normal incidence ray trace
modelling on the velocity /
depth section generates
synthetic event with diffractions
to overlay on the stack section.
Data courtesy of Paradigm Geophysical (UK) Ltd.
Depth Conversion
Summary
When well control is adequate to define the velocity
distribution in the macrovelocity model analytical functions
are used.
When well control is inadequate then seismic velocities may
be used. The seismic velocities have to be calibrated to well
velocities.
The residual misties at the well locations are used to quantify
the accuracy of the depth conversion.
The residual error adjustment of the depth maps is made
when depth maps are required that tie the well control
exactly.
12.52
Depth Conversion
Summary
Accuracy of depth conversion: Rank Wildcat (50 km to well control) ~5%
Exploration well (10km to well control) ~2.5%
Appraisal well (2 or 3km to well control) ~1%
Development wells <0.5%
The final depth structures should be no more complex than
the time structures. If they are there needs to be a very good
explanation for the complex velocity model used.
Depth Conversion
Contractor
Depth Conversion
CGG
GeoQuest
GX
Paradigm
Sattlegger
Scott-Pickford
Landmark/Promax
Time Depth Mapper

InDepth
GX VM
Earth Model
Cubit/Velit
TDQ/Mimic/Raymap/SigmaView
12.53

Depth Conversion of Time Interpretations Volume Models

Uploaded by

Copyright:

Available Formats

Depth Conversion of Time Interpretations Volume Models

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Depth Conversion of Time Interpretations Volume Models

Uploaded by

Copyright:

Available Formats

Depth Conversion

Long Period Static Anomalies

No LVL Static Correction

After Musgrove, 1994, Time Variant Statics

Long Period Static Anomalies

From Armstrong et al, 2001, Removal of overburden

Long Period Static Anomalies

From Armstrong et al, 2001, Removal of overburden

Tx2 = To2 + x2 / VRMS2 - x4(VRM44 - VRMS4) / 4 To2 VRMS8

Long Period Static Anomalies

From Armstrong et al, 2001, Removal of overburden

Ranked Methods for Functions

1. Constant average velocity.

Depth conversion may use a

An average velocity function

5000 Vi (ft/sec) 20,000

Depth conversion process

Constant Average Velocity

Apparent closed area depends of choice of contour

Constant Average Velocity with External Drift

Constant Average Velocity with External Drift

Constant Average Velocity with External Drift

Summary - Single Function

One Way Time sec

5000 Vi (ft/sec) 20,000

Depth conversion process

From a Sattlegger brochure

From a Sattlegger brochure

From a Sattlegger brochure

From a Sattlegger brochure

From a Sattlegger brochure

From a Sattlegger brochure

Summary - Multiple Functions

When we have little well control then we have to make use of

Summary - Multiple Functions

C.I. = 100 m/s

Average Stacking Velocity

How might we use seismically derived velocities for depth

Data courtesy of Amoco (U.K.) Ex. Co.

How accurate do you suppose this depth conversion is?

Sparse Well Control

Bias Corrected Models

Depth - x1000 feet

Ranked approaches to depth conversion with seismic

Enough data points

From a GX Technology brochure

From Francis, Geostatistical Applications in Asset Valuation Uncertainty, PETEX 94.

Sample Separation (m)

VSP Two-Way Time

VSP Two-Way Time

VSP Average velocities and Stacking velocity functions at

VSP Average Velocity

Calib Stacking Velocity

Uncalib Stacking Velocity

Note the linear relationship between the uncalibrated velocities.

10-5 Time Horizon

Kriging 1 many wells

Kriging 1 few wells