RECORDER

TRUE DEPTH CONVERSION: MORE THAN A PRETTY PICTURE

Edward L. Etris, Nick J. Crabtree, Jan Dewar
Scott Pickford, A Core Laboratories Company

A good seismic image is not enough for an exploration or interpretation in the time domain is a riskier business.
field development interpretation. Good well ties and reliable Interpreting structure i n the time domain means accepting
depth conversion are also required. The authors have found the risk of assuming a constant velocity model, or that all
that geologists and geophysicists tend to approach the depth p o s s i b l e v e l o c i t y a b e r r a t i o n s c a n be c a u g h t b y the
conversion process quite differently. interpreter. Further, even simple geology can produce false
The geologist says, " I f I don't have "True Depth": highs (or can obscure true highs) - a 'velocity anomaly' is
wells, how can I do depth?" - often the actual not required in order to have a time structure. A thick zone
unaware that seismically-derived depth in the of high velocity material can masquerade in the time domain
velocities exist. T h e geophysicist subsurface. as an evenly deposited layer of rock overlying a structural
says, " I have all these velocities from high {Figure 1). Many good interpreters have fallen into this
my seismic," - and needs to be cautioned that these imaging classic pitfall! Similarly, structures can be concealed by the
velocities are not right for true depth conversion. o v e r b u r d e n , a n d a good d e p t h c o n v e r s i o n can s h o w
structures where none were thought to exist, revealing
We have also seen that there is sometimes confusion about potentially bypassed reserves.
w h a t the deliverables of a depth conversion project are.
These can be 1) seismic data v o l u m e ( S E G - Y ) i n depth D e p t h c o n v e r s i o n is a w a y to r e m o v e the s t r u c t u r a l
instead of time, 2) maps and/or computer grids of depth from ambiguity inherent i n time and verify structure.
the seismic a n d w e l l s , 3) a velocity model i n the f o r m of a Explorationists need to v e r i f y structures to c o n f i r m the
2 D profile or 3 D cube data v o l u m e ( S E G - Y ) , 4) another presence of a structural trap when planning an exploration
possible deliverable is an uncertainty analysis on the final well, or to determine the spill point and gross thickness of a
'best' result. prospect to establish volumetrics for economic calculations,
or to define unswept structural highs to drill with infill wells
Recently we have also seen confusion over the meanings of to tap attic oil.
"depth migration" and "depth conversion," which are two
different processes. M i g r a t i o n is an i m a g i n g issue; W h a t ' s more, there is an increasing use of seismically-
conversion is a calibration issue (although some blurring of derived rock property data in reservoir studies. Geological
the lines has arisen recently w i t h the advent of anisotropic, and engineering reservoir modeling studies are inherently in
pre-stack depth migration, or A P S D M ) . The differences are depth. By translating seismic interpretations from time to
discussed later i n this article. depth, we enable the integration of the seismic asset w i t h
geologic, petrophysical, and production data.
This article w i l l describe various methods to perform depth
conversion, including how m u c h sophistication is needed for
v a r i o u s objectives. We w i l l d i s c u s s a c c o u n t i n g for r e a l
geologic structure and stratigraphy, proper calibration of ^ ^ • a n d s & Shales
seismic velocities, proper honouring of w e l l data versus
s e i s m i c d a t a , a n d s u i t a b i l i t y to m e e t t i m e a n d c o s t
constraints.
Limestone
First things first: why depth?

One thing there is no confusion about is that subsurface Target Unit

Target Unit
rocks exist i n depth. Seismic reflection data portray this
subsurface i n recorded t w o - w a y time. Most seismic
interpretation is done in the time domain, w h i c h is quick and True Structure Apparent Structure
is acceptable for many situations. Stratigraphic interpretation
in the time domain is usually fine for seismic facies and Figure 1: The perils of interpreting time sections.
sequence stratigraphy analyses, because their interpretation
remains largely the same w i t h changing structure. Structural

Reprinted from the November 2001 issue of RECORDER

Copyright: Canadian Society of Exploration Geophysicists (CSEC), Calgary
There are many methods to convert seismic times to depths, "I did a depth migration, haven't I converted to depth?"
too many to cover i n one article. Depth conversion methods
can be separated into two broad categories: direct time-depth T h i s commonly asked question is a good one because it
conversion, a n d velocity modeling for depth conversion. forces us to examine our understanding of imaging. T h e
Whichever method is selected, a n accurate and reliable depth truth is that depth migrated data sets are not depth converted.
conversion is one that w i l l 1) tie the existing wells, and 2)
accurately predict depths at new well locations. Whenever the subsurface layers that we are trying to image
seismically are not flat-lying, the reflected image w e see on
A word to the geophysicist: imaging is not depthing an u n m i g r a t e d section w i l l not correspond to the real
position of the structure.
Recognizing that s i m p l e vertical 'stretching' of seismic
times to depth cannot correct for lateral position errors that O n the unmigrated seismic section, for example:
may be present in the seismic image, w e must ensure that a • T h e observed d i p of a sloping reflector w i l l be less than
suitable image be produced before w e attempt a depth the true d i p .
conversion. T h i s is largely an independent step from the
• Synclines w i l l appear narrower than they really are.
d e p t h c o n v e r s i o n , a n d is c a l l e d ' i m a g i n g ' . Imaging
• Severe synclines w i l l appear as 'bow-ties'.
addresses the proper focusing and lateral positioning of
reflectors. Depthing addresses the vertical positioning of • Anticlines w i l l appear w i d e r than they really are.
seismic times to true depth, using true vertical propagation • A n y abrupt structural edge acts as a point scatterer and
velocities. w i l l appear as hyperbolic diffraction.

Although both imaging and depthing require velocity, the

type of velocity used is different (Al-Chalabi, 1994; Schultz,
1999). Imaging uses velocities designed to flatten gathers
d u r i n g stacking, or derived from migration. (Here the
authors adopt the terminology of A l - C h a l a b i (1994), w h o
suggests the term "provelocities" to
refer to imaging velocities obtained
"Provelocities":
from seismic processing.) Depthing,
imaging
onthe other h a n d , requires true
velocities
vertical propagation velocities,
obtained from
s u c h as are obtained from w e l l
seismic
measurements (Table I ) . Figure 2. a) Syncline feature in depth, b) Resultant bozo tie
processing.
reflection event in time (before migration), c) Seismic section
A depth conversion consists of imaging first, to obtain the after migration, d) Before migration.
best image, then depthing, to tie wells and predict depths
away from the wells. Migration is a seismic processing step to reposition reflections
under their correct surface location (Figure 2).

Table 1

Imaging Depthing
j Purpose: Purpose :
• reposition reflectors • convert seismic times to actual
• collapse diffractions depths
• produce correct relative structural • produce predictions of depths av/ay
pictu re from the wells.
• absolute vertical calibration
! Imaging is done first. Depthing is done after Imaging.
Imaging is mainly a lateral correction. Depthing is strictly a vertical correction.
Imaging uses imaging velocities Depthing u s e s vertical propagation
(horizontal and vertical components). velocities (vertical component only).

2 C S E G Recorder Novcnihi-r, 2001

Migration collapses diffractions. Migration puts the reflected s i g n i f i c a n c e is that it is the p a r a m e t e r that
energy back where it came from. p r o d u c e s o p t i m u m a l i g n m e n t of the p r i m a r y
reflection on the traces of the C M P gather, purely
These simple geometric examples listed above show that the that. Similarly, 'velocities' obtained v i a pre-stack
need for migration arises when reflectors are dipping. The m i g r a t i o n v e l o c i t y a n a l y s i s t e c h n i q u e s are
need for migration also arises w h e n the subsurface velocities p r i m a r i l y parameters that produce o p t i m u m
v a r y laterally, as v a r i a t i o n s i n v e l o c i t y w i l l also cause imaging of migrated energy. I n general, they are
reflections to be recorded at surface positions different from quite unrepresentative of velocity in the ground."
the subsurface positions. (Al-Chalabi, 1994, p. 589)

T i m e migration is strictly v a l i d only for vertically v a r y i n g Transverse isotropy (seismic waves traveling horizontally
velocity; it does not account for ray bending at interfaces. through a geologic layer w i l l normally travel at a higher
Depth migration accounts for ray bending at interfaces but velocity than a similar w a v e traveling vertically) is often the
requires an accurate velocity model. Depth migration is typi- cause of the disparity between the best depth-imaging
cally called for w h e n there is significant lateral variation of velocities and the best depth-conversion velocities (Schultz,
velocities. 1999). Provelocities are generally very different from the
true vertical velocity field. For this reason, pre-stack depth
Imaging addresses the proper lateral positioning of reflectors, migration ( P S D M ) does not provide the correct depth of
but does not result in a true depth data set, even if depth migra- events and should just be used for lateral positioning, not for
tion is used (Al-Chalabi, 1994; Schultz, 1999). Depth migration depthing (Al-Chalabi, 1994).
'depths' often mistie known well depths; errors of over 100
metres are still common after depth migration (Haskey et al., Depth migration output is in the depth domain, but does not
1998). The "depth" in "depth migration" is not true depth. Why? result in accurate depths of reflectors because the velocities
Because provelocities, those that do the best job of N M O and used are provelocities. T h i s is w h y depth migration results
migration, are not the same as true verhcal propagation veloci- do not tie wells accurately. Schultz (1999, pp. 2-7 and 2-8)
ties. Seismic energy, after all, does not travel vertically. There is says:
a strong horizontal element to the travel path of energy that w e
record in any seismic surface data (Reilly, 1993; Schultz, 1999). " E v e n though velocity model-building and depth
Even if you do a zero-offset survey, and you send the source sig- imaging create a seismic depth volume, their main
nal down vertically, the raypaths refract in accordance w i t h contribution is an improved image. The depth
Snell's law whenever velocity variations are encountered. rendering, v i a the [imaging] velocity model used
Because of Snell's law and ray-bending, the signal that departed for depth migration, is not sufficiently accurate to
vertically will be unlikely to travel verhcally. It is compelled to tie the wells. A major reason for the misties is
travel along at directions that are bent away from vertical. v e l o c i t y anisotropy. The migration velocity
analysis measures the horizontal component of
Nonetheless, provelocities are the right values to use for imag- velocity, and the depth conversion requires the
ing. What makes them so fit for their purpose, though, makes vertical component of velocity. The horizontal
them unfit for the purpose of true depth conversion, because component is often faster, and commonly makes
they are designed to correct a different problem. You neither the well depth markers come in at a shallower depth
want to use vertical propagation velocities to do depth migra- than the corresponding seismic reflection event."
tion, nor use provelocities to do depth conversion {Table 1).
Another reason for the mistie is nonuniqueness: there are
Is this unsettling? Intuitively, geophysicists feel that there many velocity models that w i l l produce an equivalent image
must be an actual velocity at w h i c h the seismic wavefront (Tieman, 1994; Ross, 1994). E v e n in totally isotropic media,
travels through the ground. Over the years, though, velocity therefore, unless well data are incorporated into the velocity
terminology has suffered casual use a n d often misuse. m o d e l ( A l k h a l i f a h & T s v a n k i n , 1995), there w i l l probably
Unfortunately, what is commonly called 'velocity' obtained be misties - especially due to the tendency to pick on the
from seismic processing; fast s i d e w h e n p i c k i n g p r o c e s s i n g v e l o c i t i e s so as to
discriminate against multiples.
"has the dimensions of velocity but is generally or
only remotely or vaguely related to the actual A good a p p r o a c h to depth c o n v e r s i o n , e s p e c i a l l y i n a
velocity i n the groimd. The most common type of complex geological environment, is first to perform a depth
s u c h ' v e l o c i t y ' is w h a t i n the i n d u s t r y is
commonly k n o w n as stacking velocity. ... Its real

November. 2001 C S E G Recorder 3

migration w i t h a velocity model optimized for structural answer ( I know the depth at the well); let me come up w i t h a
imaging, second to render the resulting laterally positioned translation function to predict that answer." This can consist
depth image to time using the provelocities, a n d finally to of a p p l y i n g a fixed translation equation, as regression
convert the depth-migrated seismic data - now i n the time models do, or a spatially-oriented function, as geostatistical
domain - to true depth using a true vertical velocity model procedures do.
(Schultz, 1999; Crabtree et al., 2001). (Figure 3).
The depths calculated v i a the direct time-depth conversion
Perhaps 'Depth Migration' should more accurately be called method can only be assessed by calculating the prediction
'Lateral Imaging Migration' - food for thought. error at k n o w n well locations ( I D ) , but this is a potentially
flawed Q C method because the depths being predicted are
To s u m m a r i z e : i m a g i n g f i r s t , to a c c o m p l i s h l a t e r a l the ones used to develop the prediction equation in the first
positioning; then true depth c o n v e r s i o n using vertical place. E v e n the c r o s s - v a l i d a t i o n method ( c o m m o n i n
propagation velocities. geostatistics), w h i c h w i t h o l d s a g i v e n w e l l f r o m the
i n t e r p o l a t i o n a n d then compares the p r e d i c t i o n at the
location to the real data, is not fully independent in that a l l
Process Result the wells were used to develop the interpolation parameters.
Depth migration T h e direct time-depth conversion method leaves us w i t h
little idea of the v a l i d i t y of the t i m e - d e p t h t r a n s l a t i o n
using optimal velocities for relationship between the wells. We can therefore have only
structural imaging as much confidence in these depths as our w e l l control
Image correctly a l l o w s , w h i c h is u s u a l l y w e a k . Moreover, this method
positioned laterally
prevents the incorporation of any velocity data from seismic,
Convert to time which may provide valuable additional information between
using optimal velocities for well control.
structural imaging Image correctly .
positioned laterally, T h e authors describe this approach as "direct time-depth
~ A now in time domain conversion" because the velocity modeling step is essentially
True Depth Conversion implicit - that is, velocity is not truly modeled, but rather
using vertical propagation reduced to a translation function. This translation function is
Actual depths: fit so as to result i n a predicted depth that either minimizes
velocities
laterally positioned the error or is back-calculated to tie the well depth exactly.
and vertically correct W h y is this not truly velocity modeling? Because it is hiding
many error factors w i t h i n this translation function. One of
Figure 3: A good approach to lieplh conversion. This luill
the major causes of error i n depth predictions is misties in
accomplish lateral positioning and vertical positioning to
achieve actual depths. time between s e i s m i c h o r i z o n s a n d the c o r r e s p o n d i n g
geologic w e l l pick i n time. Direct methods hide these errors
by forcing the wells to tie, thus altering the velocity provided
independently by the w e l l a n d creating (fudging) a n e w
Depth Conversion Methods back-calculated velocity to ensure a correct tie. T h i s means
that the translation function is now no longer simply a model
Once a suitable seismic image is obtained, the depthing step of the t r u e v e l o c i t y i n the g r o u n d - it i s a c o m p o s i t e
can commence. There is no single method by w h i c h this is correction factor (Figure 4).
done, but rather many methods exist a n d are i n use today,
not all of which are published in the literature. Each method Although direct conversion can be done so as to guarantee
has its o w n advantages and disadvantages, and the choice of exact well ties, which are desirable, the loss of independence
method is often a subjective one, or dictated by time a n d means that the reliability of predicted depths where no wells
cost constraints. T h i s is because no single method can be have yet been drilled is compromised.
shown to be superior i n all cases.
Velocity modeling for depth conversion
Direct time-depth conversion
Different kinds of velocity models are required for different
The simplest approach is to convert a time horizon to depth purposes (e.g., stacking, migration, depth conversion). When
directly - that is, without regard to the structure of velocity velocity modeling is done as an explicit intermediate step in
variations. Direct time-depth essentially says, " I know the time-depth conversion, the goal is to derive a robust model

4 C S E G Recorder NovcmlKr. 2001

that accurately predicts true vertical velocity at and between I n other w o r d s , choice of a conversion method depends
w e l l s by l e v e r a g i n g k n o w l e d g e about v e l o c i t y as a n partly on the data available and partly on the objectives of
additional tool. I n velocity modeling, the description of the study.
velocities is the goal, and then the depth conversion falls into
place. I n direct conversion methods, depths that tie at the What makes a velocity model robust?
w e l l s are the goal, a n d the velocity is a by-product. I n
velocity modeling, unlike direct conversion, the ability to T h e most reliable velocity model possible is one that is 1)
predict depth to a minimal or zero error is something that is geologically consistent, 2) uses appropriately detailed
c h e c k e d after the m o d e l i n g , as a test, rather than as a v e l o c i t i e s , a n d 3) incorporates a l l a v a i l a b l e v e l o c i t y
constraint on the procedure itself {Figure 4). information, weighting different types {seismic and wells)
properly.

Geologically consistent means building a velocity model that

Model T-D Model T-D follows the appropriate l a y e r i n g scheme: i n h a r d rock
e n v i r o n m e n t s this u s u a l l y m e a n s f o l l o w i n g the true
g e o l o g i c a l s t r u c t u r e , t a k i n g i n t o account l i t h o l o g i c a l
contrasts (e.g., bedding), folding, and faulting; in soft rock
environments the layering may simply parallel the structure
of the topography or bathymetry, because velocity may be
mainly a function of depth of burial (Schultz, 1999).

Depth (a) Depth (b)

I n a multi-layer depth conversion, the section is divided into
Zero error in depth at base, Better velocity model: separate g e o l o g i c a l l a y e r s , each of w h i c h l i k e l y has a
but poor velocity model closer fit to actual T-D different, but i n t e r n a l l y consistent, i n t e r v a l v e l o c i t y or
curve overall v e l o c i t y v e r s u s d e p t h f u n c t i o n (Figure 5). A separate
Figure 4: The paradox of velocity modeling: (a) when one velocity model is built for each layer, and results in a depth
minimizes the error at the tie point (direct conversion) the prediction of the base of the layer, given the top of the layer
model can become unrepresentative of the true function, but from the previous calculation. For example, the top of the
when one honors the true function the error at the tie point first layer is usually the seismic datum, then the base of that
increases (b).
layer becomes the top of the next layer and the conversion is
repeated, layer by layer, d o w n to the last horizon of interest.
These layers may not be of exploration interest on their own,
V e l o c i t y m o d e l i n g is a s t e p f o r w a r d b e y o n d d i r e c t but are important because they form the overburden above
conversion because velocity information adds two features the zones of interest and m a y contain significant velocity
to the conversion to depth. First, the velocity model can be variation.
e v a l u a t e d n u m e r i c a l l y , v i s u a l l y , a n d i n t u i t i v e l y for
reasonableness (i.e., tested independently of its ability to There are three levels of detail in modeling velocity,
predict depth, thus increasing its reliability), something that depending on how the velocity behaves w i t h depth. T h e
cannot be done w i t h a global time-depth correlation. Second, s i m p l e s t l e v e l is average velocity, w h e r e w e ignore the
velocity modeling enables the use of velocity information from layering and just go straight to the target horizon (Figure 5a).
both seismic and wells, p r o v i d i n g a m u c h broader data set T h i s s i n g l e - l a y e r a p p r o a c h has the advantage of being
for critical review and quality control. simple and quick to implement. The obvious disadvantage is
that such a model does not describe the subsurface in detail,
Nonetheless, a direct time-depth c o n v e r s i o n is often the so our confidence in the depths predicted may be reduced.
preferred approach in certain circumstances. I t u s u a l l y There may be good reasons to ignore the detail, though, such
offers the quickest solution, a n d it m a y be the o n l y one as a lack of consistency in the pattern of velocity behavior
acceptable within the project's budget or time constraints. w i t h depth, or a lack of easily definable horizons.
T h e conversion m a y only be an intermediate step, intended
to be repeated again soon w h e n more data are available. Or, A d d i n g more d e t a i l , w e can m o v e on to u s i n g i n t e r v a l
guaranteeing w e l l ties in the immediate vicinity of the wells velocities. Here, w e assign a constant velocity to each layer
may be the primary goal of the conversion, regardless of the within a given w e l l (Figure 5b). Using average or interval
accuracy a w a y from the wells. F r o m a technical perspective, velocities allows spatial variation of velocity between well
p r o v e l o c i t i e s m a y be u n a v a i l a b l e , or too n o i s y or locations. We can accomplish this by cross-plotting interval
untrustworthy to be of use, and time-depth curves from wells
may not be available.

November, 2001 C S E G Recorder 5

 velocity best fits the actual V ( z ) curve over the entire depth range for

s
the given layer, not just the one w i t h the best tie at the well
(i.e., the base of the geological layer) (Figure 4b). But h o w

-a-
1 •
can we evaluate goodness of fit? There is a unique quantitative
method for determining the accuracy of the fit of the models.
V -V The authors call this approach "discrepancy a n a l y s i s . " It
* 1 ' i 'intoW
was derived and patented by Al-Chalabi (1997a), and has
been used extensively for several years. What follows is a
+<F -V )e discussion of Ai-Chalabi's approach.
max Y mm max/^

a) Average veL b) Interval vei c) Instantaneous veL This approach makes use of the fact that most analytic expres-
sions of velocity variaHon with depth, whether a linear or
Figure 5
curvilinear expression, have two parameters. (The ideas
presented here are extendible to functions w i t h more than
two parameters. For simplicity, the two parameter case is dis-
velocity v e r s u s m i d p o i n t depth, for example, or w e can cussed.) For example, i n the commonly used linear equa-
contour our w e l l average or interval velocities - perhaps tion of the form V ( z ) ^ VQ + k Z , the two free parameters are
contour them geostatistically u s i n g seismic processing Vg and k. Within a given rock layer, the variation of velocity with
velocities at distances far from the wells. depth can be described equally well by a range of V O a n d k
parameter values. These analytic functions describe a smooth
A d d i n g still more detail, w e w o u l d like our model layer variation of velocity with depth, much smoother than the high
velocities to include variation with depth in some cases, frequency fluctuations observed on sonic logs (Figure 6).
because velocities often increase w i t h greater degrees of
compaction caused by thicker overburden (Figure 5c). For velocity
these situations we w i s h to have an instantaneous velocity
data set to model, such as a time-depth curve from a vertical Actual V(z)
seismic profile, or check shot survey, or an integrated sonic black curve
log. T h i s type of curve provides velocity variation over very
small depth increments, hence "instantaneous" velocity.

The simplest w a y to describe such variation is to model

instantaneous velocity as a linear function of depth: V ( z ) =
V|, + k Z , where V ( z ) is the instantaneous velocity at depth Z , Figure 6
and VQ and k are the intercept and slope of the line.
N u m e r o u s other functions have also been proposed
(Kaufman, 1953; Al-Chalabi, 1997b), some linear and some In practice no analytic function could represent the actual
c u r v i l i n e a r These functions are fit separately for each layer high frequency flutter of instantaneous velocity w i t h depth
to ensure geological consistency. The authors advocate using precisely - nor should it, because its purpose is not to describe
the simplest model that fits the data acceptably w e l l . So, the geology in that specific well, but rather the typical veloc-
i n t e r v a l v e l o c i t y is u s e d w h e r e a p p r o p r i a t e ( i . e . , no ity w i t h i n the geological unit overall. The goal is not to tind a
consistent increase i n velocity w i t h depth), then preferably a function that is an exact fit to the velocity vs. depth data for that
l i n e a r m o d e l , a n d f i n a l l y a c u r v i l i n e a r m o d e l o n l y if layer for any one specific well; the goal is to find a specific
necessary. parameter combination that produces a closer fit than any
other combination for all wells, and that tits the real func-
Instantaneous velocity modeling tions adequately. H o w do we assess which parameter pair is
the best to use, among the range of possible parameter pairs?
For those cases best suited to a velocity versus depth function, The goodness-of-fit between the well velocity data and the
the i s s u e a r i s e s of h o w to c h o o s e the b e s t f u n c t i o n . calculated function curve can be calculated. Both parameters
A simple w a y to check the correctness of a V ( z ) function is are varied and the goodness-of-fit calculated for each pairing,
to calculate the depth it predicts for a given geologic top at a w h i c h is termed "discrepancy." The value of the discrepancy
w e l l location, where the top depth is k n o w n . H o w e v e r , it at each parameter pairing is given by
can q u i c k l y be seen that m a n y different V ( z ) models w i l l
calculate the correct depth of a g i v e n geologic marker. (V,-C/
F(V„. k) -
W h i c h is the best V ( z ) f r o m a m o n g the possible candi-
dates? The best one is the one that w i l l effectively predict (4) from Al-Chalabi (1997a)
depths at locations a w a y from the wells, which is the one that

6 C S E G Recorder Noiwiibcr. 2001

where V, and denote the i actual (observed) and function
velocity values respectively, in is the number of sampled
depth points, and q is the norm (q^2 in this case).

I n the crossplot space of the two free parameters (V(|, k ) , the

discrepancy values for each pairing are contoured. Each iso-
discrepancy contour delimits a region in the parameter space
inside w h i c h any (V,,, k) combination produces a function
that fits the w e l l velocity data more closely than the value of
that de-limihng contoun The area w i t h i n an iso-discrepancy
contour is an area of equally good parameter pairs. T h e
Region of equally good parameter
discrepancy contour corresponds to a margin of tolerance. choice for BOTH wdb
(Figure 7)
Figure 8
Single well
Contours of iso-tUscrepancy
different fault blocks, or different facies associations. These sit-
uations can be handled by holding one parameter constant, such
as k, and then solving for the other, allowing it to vary. O n c e c a l -
c u l a t e d for a l l w e l l s , it c a n t h e n be m a p p e d , providing a
map of anticipated differences in uplift or facies.

Parameterl (eg.
Yellow area: equally good
parameter choice

Figure 7

WeU2

There is no single parameter pairing that can be considered

the ' e x a c t ' s o l u t i o n , e s p e c i a l l y w h e r e a s i n g l e w e l l is Differences may indicate different fault
concerned. A g i v e n parameter combination may, however, blocks or changing facies

satisfy the data from more than one w e l l . B y m a k i n g a

Figure 9
composite discrepancy overlap plot of the discrepancy
contours for the same layer i n two wells (or three wells, or
m a n y w e l l s ) , the region of overlap between the contours The discussion thus far has focussed on calculating velocity mod-
represents the (V,),k) p a i r s that w o u l d p r o d u c e a V ( z ) els from wells. The patient reader has been waiting for the dis-
function that w o u l d fit both wells to w i t h i n the appointed cussion to open up to the possibilities that seismic data offer to
m a r g i n of t o l e r a n c e . T h a t i s , a n y s u c h p a r a m e t e r velocity modeling. The despairing reader may even have seis-
combination w o u l d provide a single function that applies to mic but no w e l l data. A r e the benefits of v e l o c i t y m o d e l i n g
the whole area adequately and correctly. Through the use of s t i l l a v a i l a b l e i n this situation? T h e following sections show
discrepancy overlap plots the range of acceptable parameter that velocity modeling, including instantaneous velocity model-
pairings can be reduced, thus increasing the confidence in the ing, is still available even w h e n only provelocities are available.
applicability of the parameters over a large area. (Figure 8)
Use all available velocity data to build a robust velocity
model for depth conversion
If a single region of overlap can be found, then the reliability
of the model is high since it applies to all wells used in the W h i l e no-one w o u l d disagree w i t h the advice, " U s e a l l avail-
analysis. T h u s , predicted values between the wells should be able data", we must bear in m i n d that different types of data have
reliable. If the wells don't all overlap, but instead break into different degrees of certainty, particularly w e l l data versus seis-
clusters (Figure 9), it m a y indicate that there are several mic data. Well data can consist of vertical seismic profiles ( V S P ) ,
different sub-areas w i t h i n the o v e r a l l area. These are often

November, 2UUI C S E G Recorder 7

check shot s u r v e y s , sonic logs, or some combination of are w h a t w e require for depthing. We can correct them
these in several wells. V S P and check shots may be used substantially, however. The provelocities require corrections
directly, but sonic logs require corrections for " d r i f t " to be for anisotropy (depthing demands the vertical velocity, and
comparable to a V S P or check shot survey i n the s a m e w e l l provelocities contain a horizontal velocity element), for
( R e i l l y , 1993). G e n e r a l l y s p e a k i n g , V S P are preferred heterogeneity (due to lateral facies changes and such), and for
most, then check shots, then integrated sonic logs, but the ray bending (Al-Chalabi, 1994). These corrections produce
more w e l l s available the better, e v e n if it means mixing dif- closer estimates of the true vertical velocities for accurate
ferent types of time-depth curves. depthing.

Well data are h a r d measures of depth - not completely Although provelocities w i l l always have more error and
w i t h o u t error, but the w e l l d e p t h m e a s u r e m e n t s c a r r y uncertainty than well velocities, w e can at least calibrate them
relatively low uncertainty. However, wells present us w i t h to the w e l l s a n d then benefit f r o m their a d d e d s p a t i a l
velocity information that is spatially sparse, often clustered, coverage. Calibrated provelocities can be converted either to
and l i m i t e d by w e l l total depth. Further, w e l l data over- average or to interval velocities, and then combined with well
represent anomalous locations, such as structural highs. average or interval velocities, preferably using an appropriate
Seismic data offer a spatially dense, regular, and objective geostatistical approach (e.g., k r i g i n g w i t h external drift,
sampling, and cover the entire depth range evenly throughout collocated cokriging, etc.). This approach is often very useful
the survey area. These traits offer the opportunity to overcome in the first layer of a multi-layer depth conversion, where
many of the limitations of using w e l l data alone. However, instantaneous well velocities are often limited or absent (due
seismic data are a measure of time rather than depth or to no logging behind surface casing), and where average
velocity directly, and the provelocities derived from seismic provelocities handle the overburden and provide extensive,
are imaging velocities, not vertical propagation velocities such unbiased areal coverage.
as in wells.
Perhaps more usefully, though, provelocities can also be used
A n y effort that undertakes to combine hard (well) data (high for instantaneous velocity modeling, using several different
certainty and low sampling density) and soft (seismic) data approaches. I n one such approach, well data are used to
(low certainty and high sampling density) must honour the derive the gradient parameter (k) in the Vo,k function, and
h i g h e r c e r t a i n t y of the w e l l d a t a . Geostatistics ( s p a t i a l interval velocities extracted from seismic and converted to V,,
statistics) is the proper w a y to combine these two diverse are used in combination w i t h well VQ values in a V,) map.
types of data and retain proper weighting of well control, as Another approach is to derive the VQ from the provelocities
w e l l as to capture and maintain spatial trends (Chambers et after they have been converted into "pseudo w e l l s , " described
al., 2000). For instance, kriging (which here is taken to include next.
the various versions of kriging and cokriging) is a method of
interpolation that uses specially-weighted combinations of The advantage of these approaches is that the wells provide
data observed at k n o w n locations (such as wells) to predict the necessary detail in the vertical direction (k gradient), the
u n k n o w n v a l u e s at other locations. K r i g i n g also p r o v i d e s seismic provides the necessary detail in the lateral directions
estimates of the accuracy of the predicted values. (VQ map), and geostatistics ties them together w i t h proper spa-
tial weighting.
The field of geostatistics presents many interesting techniques
for integrating and mapping velocity, and analyzing spatial Pseudo-wells from seismic
structures of velocity. We must restrict the scope of this short
article to the topic at hand - depth conversion - and simply One of the unique things that can be done w i t h provelocities
note that geostatistical analysis offers us tools to combine all is to compute "pseudo-wells."
available velocity data.
From provelocities that have been calibrated to true vertical
Extending the velocity model to make use of velocities, time-depth curves ( T - D curves) are c o m p u t e d at
velocities from seismic each s t a c k i n g location. ( T - D curves are just another w a y of
representing velocity-depth functions.) T h e n T - D curves can
H o w can w e extract good quality v e r t i c a l propagation be amalgamated (averaged) into pseudo-wells to be used in
velocity information from seismic data? Recall our earlier instantaneous velocity function modeling just as the T - D curves
discussion that the provelocities used i n processing the from wells are used for instantaneous velocity modeling,
seismic data to a stacked, migrated, laterally focused picture albeit at a coarser time sampling {Figure 10). T h e averaging
of the subsurface reflectors are not the same entity as true is used to smooth the error inherent in stacking velocity
vertical propagation velocities i n the same ground, w h i c h analysis.

8 C S E G Recorder November. 2001

 propagation velocities. T h i s is not an error - i m a g i n g is a
separate issue from true depth calibration.

A l l imaging processes use a category of velocity that is more

properly called "provelocity," although "imaging velocity" or
"seismic velocity" suffice as well. Provelocity is appropriate
for imaging because seismic acquisition and processing
involve both vertical and horizontal velocity to varying
extents, but it is inappropriate for depth conversion - or
"depthing" - because depthing requires strictly actual vertical
propagation velocity ("true" velocity). True velocity is best
obtained from vertical seismic profiles, check shot surveys, or
calibrated sonic logs.
Figure W
Depthing can be done v i a a w i d e range of existing methods,
too many to cover in any article, but which can be separated
It can quickly be seen that even if we have seismic only and no into two broad categories: 1) direct time-depth conversion,
well data or sparse wells, we can derive pseudo-wells and do instan- and 2) velocity modeling for time-depth conversion.
taneous velocity modeling for our depth conversion. I n this w a y w e
can use V { z ) gradient functions to model velocity even if w e Direct time-depth conversion ignores the structure (spa-
are using seismically-derived velocities. I n order to do this, tial patterns) of velocity, and operates at k n o w n depth
though, seismic velocities need to be sufficiently detailed points only (i.e., at w e l l s ) by forcing a n exact or m i n i m a l
vertically to allow a robust V ( z ) c u r v e to be d e r i v e d . A n error m a t c h b e t w e e n a c t u a l a n d p r e d i c t e d depths.
approach to pre-stack velocity analysis has been developed to Moreover, direct conversion only i n v o l v e s seismic times at
produce "geologically consistent velocities" in seismic pro- w e l l points - velocity information from seismic, a n d all
cessing (Crabtree, et al., 2000). I n addition to providing a finer the spatial benefits that go w i t h it, cannot be u s e d .
spatial sampling along the time axis, this technique forces a
closer fit to w e l l velocities and generally reduces the arti- Velocity m o d e l i n g for time-depth conversion i n v o l v e s
facts that are typically present i n provelocities. building a true velocity model using all available velocity
data. T h i s modeling may include various types of well
Because of the greater areal coverage of seismic, one of the velocities only, or calibrated provelocities only, or both.
significant uses of pseudo-wells is to create many wells spread Modeling may use simple average velocity (single layer), or
out across a study area and perform discrepancy contouring interval velocity (multi-layer), or instantaneous velocity (vari-
and overlap plots to look for clustering of pseudo-wells into ation of velocity w i t h depth). The goal is to determine a
areas of different velocity behavior. T h i s points out areas of model that has some likelihood of w o r k i n g adequately
major facies changes or differences in uplift caused by between the k n o w n depth points, in addition to matching the
faulting. The pseudo-well technique is a geological tool as k n o w n points. Some techniques can be used involving con-
w e l l as a velocity modeling tool for time-depth-conversion. It ditions other than final depth prediction accuracy, and can
can be used for the detection and evaluation of anomalously- then be tested against the known points to determine their
pressured geological units, such as geopressured units effectiveness. T h i s is an independent way to predict depth
(Gordon et al., 2000) that must be dealt w i t h during drilling because it uses velocity functions as the input rather than
(or avoided), and basin-centered gas accumulations (Surdam, horizon depth and time at wells, and because it can involve
1997). provelocities in addition to or even instead of w e l l velocities.

Summary The choice of a depthing method depends on data availability

and quality, depthing objecrives, and time and cost constraints
In this article w e have touched on a number of issues w i t h on the depthing process. Direct methods are fast and accurate at
regards to translating seismic from time to true depth. Seismic the wells, which may be all that is required. Some forms of
imaging is a separate step and must be addressed before velocity modeling can also be fast and exact, whereas other
depth conversion. N o depth conversion can correct for forms require significant data resources, modeling expertise,
improper lateral positioning of events, because depth conver- and time to administer, yet offer greater confidence in the
sion is a vertical process only. Depth migration is currently the results, particularly between well control, where it really counts.
ultimate tool for lateral imaging, but it does not calibrate the
seismic to true depth, because it does not use true vertical

November, 2001 C S E G Recorder 9

References & Further Reading
_ E D W A R D L. ETRIS
Al-Chalabi, M., 1997a, Parameter nonuniqueness in velocity versus depth
functions, Geophysics, 62, no. 3, 970-979. ^^^^^k Edward (Ned) Etris received M.Sc.
• ^ 1 (1987) and Ph.D. (1991) degrees in geology
Al-Chalabi, M . , 1997b, Time-depth relationships for multilayer
•^r from the University of South Carolina,
depth conversion. Geophysical Prospecting, 45, 715-720.
specializing in quantitative sedimentol-
Al-Chalabi, M., 1994, Seismic velocities - a critique, First Break, 12, no. ogy. Ned is currently Manager of Geology
12, 589-596. at Scott Pickford Canada. He is an experi-
enced reservoir characterization geologist
Alkalifah, T., Tsvankin, I., 1995, Velocity analysis for transversely isotropic
and numerical reservoir model developer,
media. Geophysics, 60, no. 5,1550-1566.
with particular strengths in the evaluation
Chambers, R. L., Yarus, J. M, Hird, K.B, 2000, Petroleum geostatistics for and integration of core, log, and seismic
non-geostatisticians. The Leading Edge, May 2000 {Part 1), June data, including statistical and geostatisti-
2000 (Part 2). cal analysis. He came to Canada in 1991 to work for Canadian
Hunter Exploration Ltd., then later Ranchmen's Resources Ltd. and
Crabtree, N . , Hill, D., Veltmeijer, H . , 2001, Depth prediction from a
Crestar Energy Inc., gaining a wide range of prcxluction company
prestack depth image: a Dutch North Sea case study, S E G Aimual
experience in Western Canada. His work has ranged from research
Meeting, San Antonio, September 2001.
to development drilling and pool exploitation, and from regional
Crabtree, N.J., Etris, E . L , Eng, J., Brewer, G . , and Dewar, J., 2000, prospecting to detailed reservoir characterization. Since joining Scott
Geologically consistent seismic processing velocities improve time to Pickford in 1999 he has directed work on time-depth conversion, 3D
depth conversion, poster session presentation at GeoCanada2000 geological modeling, petrophysical evaluations, and tight gas sand
combined Conference of the C S F G . C S E G . C G U . C W L S . M A C . evaluations, and played a key role in integrated geophysical-geolog-
G A C . Calgary, Alberta, May 29-June 2, 2000 ical-engineering studies within Scott Pickford's Reservoir Modeling
Group. Ned has numerous papers and professional presentations to
Gordon, G . , et al.. Integrated science to predict overpressure in a new
his name, and has taught industry courses on geostatistics.
deep-water frontier area - N W Europe, Overpressure 2000
Workshop. London, U K , April 2000.

Haskey, P., Faragher, J. R., Raymond!, M. J., Dangerfield, J. A . and

Fjeld, O., 1998, Embla: An interpretive case history: Depth imaging M I C K J. C R A B T R E E
with well controlled signal estimation, 68th Ann. Internal. Mtg: Soc. of
Nick Crabtree received an M . A .
Expl. Geophys., 1174-1177
(Hons) Natural Sciences (Geological
Kaufman, H . , 1953, Velocity functions in seismic prospecting. Geophysics, Sciences) degree in 1992 from Jesus
18, 289-297. College, Cambridge, U K . (The college,
part of the University of Cambridge, was
Reilly, M., 1993, Integration of well and seismic data for 3D velocity model
founded in 1496!) Nick is currently
building. First Break, 11, no. 6, 247-260.
Technical Manager of Depthing Research
Ross, W. S., 1994, The velocity-depth ambiguity in seismic traveltime data. and Services, at Scott Pickford's Croydon
Geophysics, 50, no. 5, 830-843. U K office. H e has worked on the
"Velocity Modeling Using Statistical
Schultz, P., 1999, The Seismic Velocity Model as an Interpretation Asset,
Analysis of Seismically Derived
1999 Distinguished Instructor Short Course, Distinguished
Velocities" project; served as technical lead on the development of
Instructor Series, No. 2, S E G .
V E L I T , Scott Pickford's velocity modeling software; and presented
Surdam, R.C., 1997, A New Paradigm for Gas Exploration in Anomalously numerous papers at the S E G , C S E G , and E A G E . In 1999, after a
Pressured "Tight Gas Sands" in the Rocky Mountain Laramide Basins. year in snowy Calgary, Nick returned to the Croydon office to
in R . C Surdam, ed.. Seals, traps and the petroleum system: A A P G head up the RISKIT research project studying uncertainty and sen-
Memoir 67, 283-298. sitivity analysis of velocity modeling and depth conversion. He
provides technical support and mentoring in the areas of depth
Heman, H.J., 1994, Investigating the velocity-depth ambiguity of reflection
conversion, velocity modeling and reservoir characterization.
traveltimes. Geophysics, 59, no. 11, 1763-1773.
Nick's professional interests also include seismic attribute analy-
sis, reservoir characterization, integration, and mapping.
For Further Reading
Yarus, J.M. and Chambers, R.L. (editors), 1994, Stochastic Modeling and
Geostatistics, A A P G Computer Applications in Geology No. 3,
1994. R. JAN DEWAR

Jan Dewar graduated from the

University of Alberta in 1981 with a B.Sc.
in Physics. Jan is currently working with
Scott Pickford in Calgary, with a special
enthusiasm for communicating technical
concepts including A V O , Inversion,
Modeling, VSP and Transfer Filter pro-
cessing. Rock Physics, and just about
anything else that can be puzzling to the
average tiear. jdewar@scopica.com

10 C S E G Recorder November. 2001