The role of AVO in prospect risk assessment

Rocky Roden1, Mike Forrest1, Roger Holeywell2, Matthew Carr3, and P. A. Alexander4

Abstract

Essentially all companies exploring for oil and gas should perform a risk analysis to understand the uncer-tainties in their interpretations and to properly value order prospects in a company’s drilling portfolio. For con-ventional exploration in clastic environments, primarily sands encased in shales, a key component of the riskanalysis process is evaluating direct hydrocarbon indicators, which can have a significant impact on the finalrisk value. We investigate the role AVO plays in the risk assessment process as a portion of a comprehensive andsystematic DHI evaluation. Documentation of the geologic context and quantification of data quality andDHI characteristics, including AVO characteristics, is necessary to properly assess a prospect’s risk. A DHIconsortium database of over 230 drilled prospects provides statistics to determine the importance of data qual-ity elements, primarily in class 2 and 3 geologic settings. The most important AVO interpretation characteristicsare also identified based on statistical results and correlated with well success rates. A significant conclusion isthe relevance of AVO in risk analysis when it is the dominant component in the DHI portion of the risk. Critical inthe risk assessment process is understanding the role AVO and DHI analysis play when prospects approachclass 1 geologic settings. The impact that hydrocarbons have on the seismic response is significantly diminishedin this setting versus the other AVO classes. All of these observations confirm the necessity of properly evalu-ating a prospect’s geologic setting and implementing a consistent and systematic risk analysis process includingappropriate DHI and AVO components.

IntroductionThe evaluation of amplitude variation with offset

(AVO) data, or better stated an amplitude variation withangle, has provided interpreters a powerful technicaltool in prospecting for oil and gas. The hydrocarbon ef-fect on the AVO response was first recognized in the1970s and put into practice in the 1980s (Ostrander,1984). However, it was not until Rutherford and Wil-liams (1989) showed the industry that AVO responsesfrom the top of gas sands differ by geologic setting(classes 1-3) that AVO for interpreters became moreunderstandable and rock physics modeling more rou-tine. With the addition of a class 4 AVO setting byCastagna et al. (1998), essentially all geologic settingswere accounted for in clastic environments. The under-standing and interpretation of the geologic setting as itrelates to the AVO response is critical in the risk assess-ment process.

The background theory for AVO is accredited toKnott (1899) and Zoeppritz (1919) who developed equa-tions describing elastic waves as a function of reflectionangle at an interface. Through the years, there have

been several approaches developed to simplify theseequations (Bortfeld, 1961; Richards and Frasier, 1976;Aki and Richards, 1980; Shuey, 1985; Smith and Gidlow,1987; Fatti et al., 1994; Verm and Hilterman, 1995; Grayet al., 1999) each with a different emphasis.

The most commonly used linear approximation usedin the industry is from Shuey, 1985, who took the com-plicated Zoeppritz equations and produced approxima-tions that could be measured and calculated fromconventional prestack data. There are two forms ofthe Shuey approximation typically employed in explo-ration AVO analysis.

Shuey two-term∶ RðθÞ ¼ Aþ B sin2 θ. (1)

Shuey three-term∶

RðθÞ ¼ Aþ B sin2 θ þ Cðtan2 θ − sin2 θÞ.(2)

In these equations, θ is the angle of incidence, A isthe intercept and represents the reflection coefficientat normal incidence, B is the gradient and denotesthe slope of the reflection coefficients with offset or

1Consultant Rose & Associates. E-mail: rodenr@wt.net; forrestm33@sbcglobal.net.2Marathon Oil Company. E-mail: rcholeywell@gmail.com.3QI Petrophysics. E-mail: mcarr@qipetrophysics.com.4Consultant Michael C. Forrest. E-mail: alexander.pa@att.netManuscript received by the Editor 31 July 2013; revised manuscript received 20 September 2013; published online 20 March 2014. This paper

appears in Interpretation, Vol. 2, No. 2 (May 2014); p. SC61–SC76, 20 FIGS., 3 TABLES.http://dx.doi.org/10.1190/INT-2013-0114.1. © 2014 Society of Exploration Geophysicists and American Association of Petroleum Geologists. All rights reserved.

t

Special section: Interpreting AVO

SC61SC61

Dow

nloa

ded

09/2

8/15

to 1

24.1

95.4

.82.

Red

istr

ibut

ion

subj

ect t

o SE

G li

cens

e or

cop

yrig

ht; s

ee T

erm

s of

Use

at h

ttp://

libra

ry.s

eg.o

rg/

angle (actually sin2 θ), and C is the curvature term anddescribes the behavior at large angles/offsets that areclose to the critical angle. The Shuey two-term approxi-mation is typically good for angles less than 30°, wherethe third term is dropped.

These approximations began the evolution of AVOinterpretation techniques which are important in pros-pect risk assessment. Table 1 lists a few of the key au-thors and associated AVO technology advances that areespecially relevant to AVO interpretation. From the late1990s on, prestack data became more routinely em-ployed in various inversion approaches including simul-taneous inversion (Ma, 2002; Hampson et al., 2005),elastic impedance (Connolly, 1999), and extended elas-tic impedance (Whitcombe et al., 2002). However,

the accuracy of these prestack inversion techniquesrequires relatively local and quality well control for cal-ibration that represents similar stratigraphy and depo-sitional environment as the prospect. But how do theseAVO interpretation approaches ultimately relate to howcompanies risk prospects?

This paper primarily addresses the role of AVO inprospect risk assessment for exploration prospectswith sand reservoirs encased in shales. It is importantto determine the benefits, limitations, and strengths ofAVO analysis in exploration because this provides thefundamental guidelines and foundation of AVO technol-ogy when used for exploitation and development.

Geologic setting for AVO analysisThe geologic setting interpretation is very important

in understanding how an AVO analysis contributes tothe risking process. Figure 1 displays the typical AVOresponses and designated AVO classes as they relateto the top of gas sands (Rutherford and Williams,1989; Ross and Kinman, 1995; Castagna et al., 1998).These reflection coefficient responses with offset areassociated with specific geologic settings and provideguidelines in the AVO interpretation process.

Class 3 represents low-impedance unconsolidatedsands and is usually associated with the “bright spot”environment. Class 4 is also associated with unconsoli-dated sands; however, they are overlain by hard silts,shales, or carbonates that have a higher shear wavevelocity than the gas sand which produces a decreasein negative amplitude with offset. Class 2 and 2P (latterhas a phase change with offset) relate to a more con-solidated sediment setting than class 3 with interceptshaving very small to no amplitudes but often produce

Table 1. Evolution of AVO interpretation techniques— Condensed list of key contributors.

Author Description

Ostrander (1984) Described AVO response in lowimpedance sands

Castagna et al. (1985) P and S wave relationships andmudrock line

Smith and Gidlow (1987) Geostack scheme and fluid factor

Rutherford and Williams(1989)

AVO classes 1, 2, and 3

Greenburg and Castagna(1992)

P and S wave relationships formixed lithologies

Allen and Peddy (1993) AVO case studies

Castagna and Backuseditors (1993)

Book on theory and practice ofAVO analysis

Castagna and Smith(1994)

Comparison of AVO indicators

Fatti et al. (1994) P wave reflection coefficients interms of RP and RS

Ross and Kinman (1995) AVO class 2P

Verm and Hilterman(1995)

Linear approximation usingnormal incidence reflectivity andPoisson reflectivity

Foster et al. (1997) AVO crossplotting andbackground trends

Goodway (1997) Computed lambda-mu-rho fromAVO

Castagna et al. (1998) Relationship of fluid line slopeand VP/VS ratio, AVO class 4

Hendrickson (1999) Effectiveness of interpretingintercept and gradient for fluids

Ross (2000) AVO crossplot modeling tutorial

Simm et al. (2000) Anatomy of AVO crossplots

Hilterman (2001) Comprehensive seismicamplitude course manual

Avseth et al. (2005) Application of rock physics toAVO interpretation

Hampson et al. (2005) Prestack simultaneous inversion

Li et al. (2007) Practical aspects of AVOmodeling

Foster et al. (2010) Interpretation of AVO anomalies

Figure 1. The AVO classes related to specific geologic set-tings. Curves represent the AVO responses from the top ofgas sands. Shaded regions represent approximate locationsof the specific classes. The area within the dashed curvesidentifies the approximate class 4 location. (classes 1–3 fromRutherford and Williams, 1989; class 2P from Ross and Kin-man (1995); and class 4 from Castagna et al., 1998)

SC62 Interpretation / May 2014

Dow

nloa

ded

09/2

8/15

to 1

24.1

95.4

.82.

Red

istr

ibut

ion

subj

ect t

o SE

G li

cens

e or

cop

yrig

ht; s

ee T

erm

s of

Use

at h

ttp://

libra

ry.s

eg.o

rg/

very strong gradients (usually greater than class 3).Class 1 rocks are very hard in relation to their encasingshales, and it is very difficult to identify the hydro-carbon effect on the AVO response for this setting inexploration. Roden et al. (2005b) and Forrest et al.(2010) summarize AVO classes 1-4 and their associatedgeologic setting characteristics in more detail. TheseAVO classes are general guidelines and the boundariesbetween classes are gradational and subjective at times,but the accuracy of the geologic setting interpretationhas a direct bearing on the AVO assessment and asso-ciated risk.

Crossplotting of AVO attributes helps determine thepotential hydrocarbon AVO responses and how they re-late to the appropriate geologic setting. The interceptversus gradient crossplot of Figure 2 is a common toolto interpret prospective AVO anomalies and how theyrelate to background trends (Castagna and Swan,1997; Castagna et al., 1998; Ross, 2000). Foster et al.(2010) have produced an excellent intercept versus gra-dient crossplot displaying how the crossplot pointsmove as reservoir properties such as porosity, fluidcompressibility, and increases in clay content change.The background trend in Figure 2 has additional signifi-cance in that the angle of this trend relates to the VP/VS

ratio (given certain density assumptions) and yieldsinsight into the properties of the encasing shales (Cas-tagna et al., 1998).

Figure 3 displays another useful crossplot employedin exploration where the near-offset stack (or near-angle stack) and far-offset stack (or far-angle stack)are plotted. This crossplot is especially useful in explo-ration because there are no calculations involved, onlythe crossplotting of common offsets/angles which isusually near versus far. The dashed lines through thebackground trends in Figures 2 and 3 theoretically

denote the location of shales; however, in practice,the cloud of points in the background trend ellipse usu-ally represents shales and wet sands. On these cross-plots, key interpretation elements relative to prospectrisking include the deviation from background trendof the anomalous points, their relationship to theinterpreted geologic setting, and the location of thesepoints on structure maps and vertical seismic displays.In other words, these anomalous points in crossplotspace must make geologic sense.

Prospect riskingMost oil companies risk prospects in their explora-

tion portfolio by determining a risk value associatedwith a series of geologic chance factors (Rose, 2001).The reasoning behind this approach is that each ofthe geologic chance factors are independent variables,and therefore when the risk for each of the chance fac-tors are multiplied, they produce the geologic chance ofsuccess (Pg) for a prospect. The probability of geologicsuccess represents the chance of attaining flowable hy-drocarbons that do not deplete upon test. In otherwords, it signifies the chance that a petroleum systemis working. This Pg does not pertain to potential vol-umes of hydrocarbons, which is addressed in the eco-nomic and commercial portion of the overall riskanalysis process (not addressed in this paper). Table 2shows a list of typical geologic chance factors and theirassociated elements or subfactors. Some companiesmodify and recategorize these geologic chance factorsinto as few as three or as many as eight categories, butin every case the same information is evaluated. What isimportant in evaluating the primary geologic chancefactors is that each one is equally important and ifone fails there will be no trapped hydrocarbons.

Figure 2. Intercept versus gradient crossplot displaying loca-tion of AVO classes. Shaded regions display approximate lo-cations of classes from the top and base of the reservoir. Thedashed red line and ellipse denote general location of thebackground trend typically representing wet sands andshales.

Figure 3. Near- versus far-offset crossplot displaying loca-tion of AVO classes. Location of the top and base of a reservoirare designated with their associated AVO class. The dashedred line and ellipse denote general location of the backgroundtrend typically representing wet sands and shales.

Interpretation / May 2014 SC63

Dow

nloa

ded

09/2

8/15

to 1

24.1

95.4

.82.

Red

istr

ibut

ion

subj

ect t

o SE

G li

cens

e or

cop

yrig

ht; s

ee T

erm

s of

Use

at h

ttp://

libra

ry.s

eg.o

rg/

An important question for seismic interpreters is:How does the presence of a positive AVO anomaly af-fect the Pg, and does this anomaly relate to any one ortwo geologic chance factors more than the others?Therefore, it is necessary to understand how AVO fitsin a comprehensive and systematic seismic direct hy-drocarbon indicator (DHI) evaluation.

DHI analysisA DHI is a seismic amplitude anomaly caused by the

presence of hydrocarbons. The DHI interpretation proc-ess involves identifying any number of DHI character-istics and determining their relationship to a potentialoil and gas accumulation. A DHI evaluation is part ofa risk analysis work process to assess the probabilityof success that can typically range from 5% to 95%.

One approach in an exploration risk analysis processfor DHI prospects is to determine an initial Pg based onthe geologic chance factors independent of the ampli-tude anomaly as a fluid indicator or DHI. This estab-lishes an initial geologically based Pg and allows theDHI component of risk to be assessed separately.For example, in a pure stratigraphic play where thereis no structural closure or well control, the initial Pg willusually be very low because most of the risk assessmentwill be in analyzing the amplitude defined prospect. Atthe other extreme, if there have been several wellsdrilled near a prospect establishing the reservoir pres-ence and properties, the structure is well defined, andhydrocarbons are present, then the initial Pg will be rel-atively large before the DHI component of risk is deter-mined. This methodology allows for an overall DHIevaluation and does not bias the assessment of any ofthe geologic chance factors. To have a true DHI presenton seismic data, all of the geologic chance factors mustbe working (100% chance) because a petroleum systemexists. The exception to this is the presence of low-saturation gas which can produce seismic DHI charac-teristics that are difficult to distinguish from commer-cial quantities of hydrocarbons. The presence of low-saturation gas sands typically indicates there has beena trapping configuration present at some time in thepast, but the seal or containment of the trap has beenbreached or broken and hydrocarbons have leaked out,

leaving behind residual gas trapped bycapillary pressures (Holtz, 2002).

It is important to understand thatAVO is one component in the toolkitof a comprehensive seismic DHI evalu-ation. A good DHI prospect has multiplepositive DHI characteristics. Therefore,a systematic and comprehensive DHIevaluation, including AVO assessment,is necessary to determine the inter-preter’s confidence or risk that the pro-spective anomaly being evaluated trulyis a DHI. Not all seismic amplitudeanomalies are DHIs, and not all DHIshave the same characteristics. An indus-try-wide DHI consortium has developeda methodology to evaluate DHI pros-pects and determine their associatedrisks (Figure 4). A description of thisconsortium’s background and method-ology can be found in Forrest et al.(2010) and Roden et al. (2012). Overthe last 13 years, this consortium has

Table 2. Geologic chance factors.

Source Rock

Kitchen area and thickness

Richness

Thermal maturity

Hydrocarbon type

Migration and Timing

Closure forms before/during migration

Migration distance and pathways

Reservoir Rock

Facies and extent

Minimal thickness

Reservoir quality

Trap

Confidence of depth/shape of trap

Structural and stratigraphic traps

Confidence in mapping

Containment

Sealing capacity/top and bottom

Preservation

Figure 4. DHI prospect risking methodology workflow chart based on an indus-try-wide DHI consortium over the past 13 years.

SC64 Interpretation / May 2014

Dow

nloa

ded

09/2

8/15

to 1

24.1

95.4

.82.

Red

istr

ibut

ion

subj

ect t

o SE

G li

cens

e or

cop

yrig

ht; s

ee T

erm

s of

Use

at h

ttp://

libra

ry.s

eg.o

rg/

accumulated a database of over 230 DHI drilled pros-pects (97% are class 2, 2P, and 3) from around the worldcontributed by over 40 oil companies. About half of theprospects are from the offshore Gulf of Mexico, with50% from deepwater and 50% from the continental shelf.The database contains 175 class 3 prospects (53% suc-cess rate), 53 class 2/2P prospects (60% success rate),five class 1 prospects (20% success rate), and two class4 prospects (0% success rate). The prospects in this da-tabase are predominantly exploratory (86%) and rangein age from Triassic to Pleistocene (90% Tertiary). Pros-pect target depths range from 3000 to 20,000 feet. Muchof the DHI evaluation methodologies and AVO statisti-cal information in this paper come from the DHI con-sortium and consortium database.

The initial risk assessment process requires putting aprospect in its proper geologic context where it is nec-essary to document the prospect’s location, well class(e.g., development, wildcat, etc.), type of trap, terrain(onshore/offshore), depth-to-target, anomaly size andthickness, etc. The expected reservoir geologic charac-teristics such as age, lithology, lateral geometry, andvertical boundaries are also input along with the pres-ence and location of analog fields, discoveries, anddry holes.

In a comprehensive DHI evaluation, it is necessary tocalculate the data quality, which includes seismic androck/fluid property data. Table 3 lists the general com-

ponents of data quality that must be evaluated andquantified to determine its impact on prospect risking.From an AVO perspective, there are numerous techni-cal approaches and products to evaluate prestack dataincluding lambda-mu-rho, fluid factor, prestack inver-sion techniques, etc., but all of these approaches essen-tially are derived from the fundamental prestack datalisted in Table 3 and in most cases are derived from in-tercept and gradient values or equivalents (Ross, 2010).

The DHI consortium has compiled statistics on all ofthe information in Table 3 and correlated this informa-tion with success rates. Figures 5, 6, and 7 display thestatistics on three of the most important rock and fluidproperty data quality characteristics (especially rel-evant to AVO). The success rates for class 2 and 3are displayed for a prospect’s proximity to well control,a prospect’s similarity of depth, stratigraphy and depo-sitional environment to well control, and whether anyAVO modeling was performed. Other well log charac-teristics such as the source and quality of density,P-wave, and S-wave information are important in quan-tifying data quality and their statistics show similartrends as the graphs in Figures 5, 6, and 7.

Figure 5 illustrates proximity to well control is cor-related with higher success rates. For prospects withwell control less than a mile from the prospect, the suc-cess rates for class 2 and 3 wells are well above theaverage total success rate for each class. For well

Table 3. Quantified data quality elements for DHI analysis.

SeismicRock and fluid property data(VP, VS, density)

Type of data (2D, 3D, single component, multicomponent) General setting

Processing/migration (pre- or poststack migrations, time or depth, etc.) Proximity to prospect

Overall imaging quality (focusing, defocusing, ray path geometries, etc.) Similar depth, stratigraphy anddepositional environment

Acquisition vintage Source/quality of rock density andvelocity data

Processing vintage Pressure environment of prospect

Amplitude preservation Modeling (synthetics, offset andmulticomponent modeling)

Processing type (speculative or specifically for prospect)

Data coverage (line spacing, bin size, offsets, etc.)

Phase estimation and accuracy

Vertical resolution (tuning thickness)

Horizontal resolution (relationship to amplitude size)

Prestack data (offset and angle gathers, offset and angle volumes;

AVO, angle [θ] or sin2 θ graphs, intercpet versus gradient crossplots, near versus far offsetcrossplots, etc.)Amplitude preservation

Sufficient offsets/angles for target depth

Accuracy of offset angle calculations

Reflectors distinguished from coherent and random noise

Proper NMO corrections

Anomaly and background trends in crossplotting

Interpretation / May 2014 SC65

Dow

nloa

ded

09/2

8/15

to 1

24.1

95.4

.82.

Red

istr

ibut

ion

subj

ect t

o SE

G li

cens

e or

cop

yrig

ht; s

ee T

erm

s of

Use

at h

ttp://

libra

ry.s

eg.o

rg/

control within 1–5 miles, the success rates are close tothe overall success rates for their associated classes.For well control greater than five miles, the success rateshows a significant decrease.

Figure 6 displays success rates related to how similarthe well control is to the prospect in depth, stratigraphy,and depositional environment. When these propertiesare interpreted to be nearly identical from several wellties and seismic information to the prospect, the suc-cess rates are significantly above average for the overallsuccess rates for each AVO class. For the situationswhere there is only one or very few wells and the seis-mic correlations are less, the results are almost exactlycorrelated with the overall success rates and may re-present the most common situation of the prospectsin the database. When correlation to depth, stratigra-phy, and depositional environment are unknown or evi-dence suggests they may be different, the success ratesdrop below average success rate.

Good well control not only helps in better definingthe geologic setting but provides for more accurate in-put into AVO modeling. Also shown are the statisticson success rates when AVO modeling was performed(Figure 7). When a well-defined model with reliable in-puts closely matches the seismic, the success rates forclass 3 prospects is approximately 30% above average,and for class 2 prospects 40% above the average suc-cess. In fact, for class 2 wells there was a 100% successrate when the model closely matched the real datawhich was based on a limited sample of four successfulwells. With reasonable inputs for modeling and a some-what reliable match with the data, the success ratesclosely matched the overall success rates of class 2and 3. When no modeling was performed or inputs be-come less reliable, the success rates drop below theclass range averages. If the model does not fit regard-less of inputs, class 2 and 3 success rates fall almost 40%below their associated averages. This demonstratesthat AVO modeling is positively correlated with highersuccess rates, especially for class 2 rocks where theAVO effect can be dramatic.

The interpretation and quantification of risk for DHIcharacteristics requires making an assessment of spe-cific DHI characteristics based on observations of thepertinent data and modeling comparisons. As previ-ously stated, the DHI characteristics assessment mustbe related to their interpreted geologic setting. TheDHI consortium has identified 37 DHI characteristicsfor class 4 anomalies, 38 characteristics for class 3,32 characteristics for class 2, and as few as 26 charac-teristics for class 1 anomalies. These DHI characteris-tics can be divided into different categories with theamount in each category and content of each character-istic varying depending on the geologic setting. The DHIconsortium has divided these DHI characteristics intonine categories including potential pitfalls as delineatedon the flowchart of Figure 4. Quantification of AVOcharacteristics is an important and critical componentin the overall DHI risk assessment. All of the DHI

Figure 6. Class 2 and 3 well success rates based on how sim-ilar the depth, stratigraphy, and depositional environment areto the well control. Overall success rate averages for each en-tire class are designated by dashed lines.

Figure 7. Success rates for class 2 and 3 wells based on per-formance of AVO modeling. Overall success rate averages foreach entire class are designated by dashed lines.

Figure 5. Success rates for class 2 and 3 wells based on theirproximity to well control. Overall success rate averages foreach entire class are designated by dashed lines.

SC66 Interpretation / May 2014

Dow

nloa

ded

09/2

8/15

to 1

24.1

95.4

.82.

Red

istr

ibut

ion

subj

ect t

o SE

G li

cens

e or

cop

yrig

ht; s

ee T

erm

s of

Use

at h

ttp://

libra

ry.s

eg.o

rg/

characteristics must be quantified to determine the im-pact of direct hydrocarbon indicators have on the over-all risk of the prospect.

AVO characteristics for riskingThe DHI consortium has compiled a list of AVO char-

acteristics that have been determined to be necessary inrisking prospects (Roden et al., 2012). These character-istics have been statistically evaluated and quantified tounderstand their impact on prospect risk assessment.

AVO observations using gathers, offset/angle stacks,

or windowed (gated) derived amplitudes. This charac-teristic relates to an interpreter’s confidence that theAVO response is proper for the associated AVO class(see Figure 8). In other words, for the AVO class inthe interpreted geologic setting, the normal incidentand amplitude change with offset (gradient) are trend-ing appropriately. Noisy gathers, incorrect NMO correc-tions, multiples, insufficient offset during acquisition,and processing artifacts often complicate evaluationof this characteristic.

Consistency of AVO in mapped target area (typi-

cally on gathers, far offset stacks, or windowed AVO

attributes). The internal consistency of the AVO re-sponse in a prospective reservoir was found to be asignificant DHI characteristic. This relates to the uni-formity of the AVO response within the mapped targetarea. Figure 9 displays guidelines for the consistency ofAVO responses within a defined DHI anomaly (Rodenet al., 2012). It should be noted that internal consistencyfor class 3 wells is determined from the stackeddata. When evaluating this characteristic, considerationshould be given to possible faulting and stratigraphicchanges that may modify internal consistency.

Change in AVO compared to model (wet versus

hydrocarbons). Modeling of the AVO response, usuallyapplying Gassmann’s equation, typically involves sub-stitution modeling of gas, oil, and water responses. Acomparison of the in situ and modeled responses to ac-tual gathers provides confidence that hydrocarbons arepresent or not. Distinguishing the AVO response of wetsands from hydrocarbons is one of the most criticalcomponents of AVO interpretation, and modeling is cru-cial in understanding these differences.

Excluding possible stacked pays, the AVO effect

response is anomalous compared to events above

Figure 8. Modeled AVO curves and associ-ated wiggle trace gather responses fromAVO classes 1, 2, 2P, 3, and 4. The gather mod-els represent a 50-foot sand with a 25-HzRicker wavelet applied.

Figure 9. Examples of AVO consistencywithin a defined anomaly from Grade 1(worst) to Grade 5 (best). The AVO responseis typically mapped from gathers, far offset/angle stacks, and/or any windowed AVOattributes.

Interpretation / May 2014 SC67

Dow

nloa

ded

09/2

8/15

to 1

24.1

95.4

.82.

Red

istr

ibut

ion

subj

ect t

o SE

G li

cens

e or

cop

yrig

ht; s

ee T

erm

s of

Use

at h

ttp://

libra

ry.s

eg.o

rg/

and below. On prestack data, usually gathers, this char-acteristic refers to whether the events above and belowthe targeted anomaly look similar (Figure 10). If theanomaly is relatively unique and the events aboveand below don’t display similar AVO trends, there isa better chance of a hydrocarbon bearing reservoir.

The AVO event is anomalous compared to the same

event outside the closure. This characteristic describeswhether the appropriate AVO class response is uniquecompared to the correlative event outside the closure.This characteristic is often interpreted from gathers,far offset/angle stacks, as well as intercept, gradient, in-tercept times gradient, far–near, and (far–near) timesfar volumes. In Figure 11, even though the seismic gath-ers are not flat, the differences in AVO responses arequite clear as one moves from off structure in thewet leg of the reservoir to the gathers under closurecontaining hydrocarbons.

How well-defined is the background trend (cross-

plots)? When crossplotting intercept versus gradientor far versus near offsets/angles, the interpretationand accuracy of defining the background trend is keyto understanding if the targeted event is anomalous(Figure 12). Interpreting the background trend usuallyencompasses wet sands and/or shales, depending onthe data window selected. Optimally, defining a back-

ground trend from a known wet sand response closeto the anomaly level can impact the risk in distinguish-ing the hydrocarbon AVO signature.

How distinct is the anomaly from the background

trend? Once the background trend has been estab-lished, interpreting how well the appropriate pointsfrom the prospect in crossplot space deviate fromthe background determine the risk for this characteris-tic (Figure 12). Identifying where the anomalous pointsfall on associated vertical seismic displays and struc-ture maps provide insight into the risking of this char-acteristic.

Can the differences from the background be ex-

plained by hydrocarbon substitution modeling? Afterthe background trend and the prospective anomalyhave been interpreted in crossplot space, comparingto modeled responses from the appropriate AVO class(geologic setting) helps confirm the reliability of theAVO interpretation. Of course, in a wildcat setting withlittle or no well control, the interpretation of this char-acteristic can be difficult (higher risk).

From the DHI consortium database, the statistics foreach of these AVO characteristics was compiled andwell success rates were compared to the interpretationgrade of the characteristics. In this convention, the AVOcharacteristics were objectively graded on a scale of

Figure 10. Gather showing the AVO response at the 1.5 second gas pay is anomalous compared to events above and below. Thisexcludes possible stacked pays. The uniqueness of the AVO response was found to be an important DHI characteristic. This gatheris displayed in different formats with the left in wiggle-trace variable area, the middle in color raster format, and the right in colorraster format with a wiggle-trace overlay. The angled colored lines on the right gather represent angle guides from 10° to 50° in 10°increments.

SC68 Interpretation / May 2014

Dow

nloa

ded

09/2

8/15

to 1

24.1

95.4

.82.

Red

istr

ibut

ion

subj

ect t

o SE

G li

cens

e or

cop

yrig

ht; s

ee T

erm

s of

Use

at h

ttp://

libra

ry.s

eg.o

rg/

1–5, with 1 being the worst and 5 being the best. Themost important class 2 and class 3 AVO characteristicswere determined based on valid statistics and successrate trends. There are 175 class 3 wells evaluated withthe overall success rate 53%. In Figure 13, the four mostimportant AVO characteristics for class 3 wells are dis-played by grades and associated success rates. Thegrade 4 values for successful prospects range from52%–69% and the grade 5 values 55%–86%. These highvalues are significant considering the overall successrate for the class 3 prospects is 53%.

For class 2, the database contains 53 prospects witha success rate of 60%. Figure 14 displays the five mostimportant AVO characteristics for this class. Except forthe characteristic, change in AVO compared to model,all of the AVO characteristics for grades 4 and 5 re-vealed success rates well above the overall averagefor class 2 wells of 60%. Grade 4 values for successfulprospects range from 76%–84%, and grade 5 values from50%-100%. The change in AVO compared to model did

not show as striking results as the other AVO character-istics. This may be due to the fact that the database iscompiled from predominantly exploration wells (86%)and well control for modeling may not have been veryclose or indicative of the specific prospect beforedrilling.

When all the AVO characteristics for a prospect hadgrades of 4 or 5, there were only six dry holes for class 3and five dry holes for class 2. For class 3 dry holes, threewells had low saturation gas, one had a thick wet sandwith possible low saturation gas, one had a wet sand,and one a hard streak above a wet sand that accentu-ated the amplitude response. The dry holes for the class2 wells included two associated with low-saturationgas, two were clean high-porosity wet sands, and onewas a shaley sand with possible low saturation gas.Therefore, when all the AVO characteristics were ratedvery high for a prospect, wet sands (often thick wetsands) and low-saturation gas were the reasons forfailure.

Figure 11. The time-structure map is based on the picks from the stacked data at the level designated by the arrows on thegathers. The six gathers on top are in wiggle-trace format, and the same gathers on the bottom are in color raster format. Thesegathers are designated by letters A–F located on the map. Moving from west to east, gathers A, B, and C display no anomalous AVOresponses at the picked horizon with gather D on the edge of the structure and gathers E and F displaying anomalous signatures.The anomalous AVO responses in gathers D, E, and F within the structure are associated with trapped hydrocarbons. Furtherprocessing is required to flatten these gathers, but the anomalous trends are evident.

Interpretation / May 2014 SC69

Dow

nloa

ded

09/2

8/15

to 1

24.1

95.4

.82.

Red

istr

ibut

ion

subj

ect t

o SE

G li

cens

e or

cop

yrig

ht; s

ee T

erm

s of

Use

at h

ttp://

libra

ry.s

eg.o

rg/

AVO relevanceAs Bill Fahmy indicated in his 2006 SEG/EAGE

Distinguished Lecture presentation on DHI/AVO best

practices methodology and applications: a historical

perspective, “standalone AVO does not equal hydrocar-bons.” (Fahmy, 2006). The DHI consortium prospect da-tabase was analyzed for success and failure of the wellsas it relates to the AVO contribution to the risk andmore specifically to the DHI component of the risk(Figure 15).

For the class 3 prospects where prestack data wasevaluated (60% of all class 3 wells or 105 prospects),

approximately half the wells were successful. For theclass 3 successful wells, the average AVO componentof the DHI risk portion was 35%. For unsuccessful wells,the average AVO contribution of the DHI risk portionwas 64%. These statistics suggest there is a greaterchance of failure if AVO is the dominant factor in theDHI analysis. This is probably because, for class 3prospects, many of the DHI characteristics manifestthemselves on stacked data better than AVO data, suchas amplitude conformance to structure, flat spots(hydrocarbon contacts), velocity pull down effects,shadow zones, etc. Another component of class 3

Figure 12. The intercept versus gradient crossplot is for data in a half-second window encompassing the high-amplitude eventdisplayed on the stacked seismic line. The background trend on the crossplot is well-defined, and the points within the red and blueellipses are from the top and bottom of a gas sand, respectively. These points within the ellipses are projected on the seismic line(Roden et al., 2005a).

Figure 13. Success rates for the four most im-portant AVO characteristics for class 3 wellsbased on grades 1 (worst) to 5 (best). Thedashed line represents the average successrate for all class 3 wells in the DHI consortiumdatabase.

SC70 Interpretation / May 2014

Dow

nloa

ded

09/2

8/15

to 1

24.1

95.4

.82.

Red

istr

ibut

ion

subj

ect t

o SE

G li

cens

e or

cop

yrig

ht; s

ee T

erm

s of

Use

at h

ttp://

libra

ry.s

eg.o

rg/

AVO responses is that the gradient is usually not verylarge so the stacking process tends to bring out many ofthe non-AVO DHI characteristics.

For class 2 prospects, the database contains fewerwells to derive good statistics. In the DHI consortiumdatabase, 92% of the class 2 wells had prestack data,which contains 49 prospects with a 65% success rate.The successful wells had an average AVO componentof 60% in the DHI analysis, almost double the AVO com-ponent for class 3 successful wells. This is reasonable

because AVO analysis plays a larger role in risking class2 prospects than class 3 prospects. For unsuccessfulwells, the statistics were poor but the AVO componentof the DHI analysis averaged about 80%, accounting forstatistical aberrations.

These statistics do not mean that AVO characteris-tics are not important, only that when the AVO charac-teristics are the dominant component in the DHI riskanalysis with an absence of other DHI characteristics,there is a statistically higher chance of failure. This

Figure 15. Bar graphs showing the averageAVO contribution to the DHI portion of Pg,based on failures and successes. For class3, the average is based on 51 successes and54 failures. For class 2, the average is basedon 32 successes and 17 failures.

Figure 14. Bar graphs showing success ratesfor the five most important AVO characteris-tics for class 2 wells based on Grades 1 (worst)to 5 (best). The dashed line represents theaverage success rate for all class 2 wells inthe DHI consortium database.

Figure 16. Class 3 AVO hydrocarbon mod-eled responses based on in situ wet sand onlogs. All curves are based on the top of thesand. Note the distance on the graphs be-tween the wet sand response and the gasand oil curves.

Interpretation / May 2014 SC71

Dow

nloa

ded

09/2

8/15

to 1

24.1

95.4

.82.

Red

istr

ibut

ion

subj

ect t

o SE

G li

cens

e or

cop

yrig

ht; s

ee T

erm

s of

Use

at h

ttp://

libra

ry.s

eg.o

rg/

reinforces the notion that a comprehensive DHI evalu-ation is necessary for proper risk assessment, includingAVO characteristics.

Limitation of AVOWhen exploring for oil and gas, a critical issue is

whether the prospect geologic setting is conducive toproduce AVO and/or DHI anomalies. In other words,as the geologic setting transcends from a class 2P toa class 1 environment, the impact of hydrocarbonson the seismic response decreases. Avseth et al.

(2005) refer this as being outside the “AVO window”

where depth and data quality play key factors. This isespecially important in risk analysis in determininghow much emphasis and weight to give to any inter-preted AVO anomaly. Figures 16–19 show a series ofhalf-space AVO models with in situ brine sands whereGassmann’s fluid substitution has been applied to deter-mine the hydrocarbon responses in each case. What isevident from these figures is that the separation be-tween the brine sand and the hydrocarbon (oil andgas) curves decreases moving from the class 3 to class

Figure 19. Class 1 AVO hydrocarbon mod-eled responses based on in situ wet sand onlogs. All curves are based on the top of thesand. Note how the wet sand and hydrocarboncurves are very close together.

Figure 18. Class 2P AVO hydrocarbon mod-eled responses based on in situ wet sand onlogs. All curves are based on the top of thesand. Compare distance between wet curvesand hydrocarbons to the wet curves and hy-drocarbons in Figures 16 and 17.

Figure 17. Class 2/3 AVO hydrocarbon mod-eled responses based on in situ wet sand onlogs. All curves are based on the top of thesand. The hydrocarbon and wet sand curvesare closer together than on Figure 16.

SC72 Interpretation / May 2014

Dow

nloa

ded

09/2

8/15

to 1

24.1

95.4

.82.

Red

istr

ibut

ion

subj

ect t

o SE

G li

cens

e or

cop

yrig

ht; s

ee T

erm

s of

Use

at h

ttp://

libra

ry.s

eg.o

rg/

1 case. This makes interpreting the class 1 AVO re-sponse as a DHI anomaly for risking purposes quitedifficult, especially in an exploration setting.

The elastic properties between shales and sandstransitioning to a class 1 setting can be quite compli-cated. By definition, class 1 reservoirs have highervelocities than their encasing shales. Figure 20 cross-plots density and P-wave velocity for sands and shalesfrom an array of wells. Seven sand cases (A-G) withfluid substitution display the sensitivity of gas, oil,and brine for classes 1, 2, and 3. Note how the brineand hydrocarbon points converge moving from SandE (class 2P/1) to Sand G (extreme class 1 case). It iscommon for class 1 reservoirs to increase in clay con-tent, decrease in porosity, and due to diagenetic proc-esses routinely increase in cementation. There areseveral other contributing factors that impact the seis-mic response including fluid properties (brine salinity,gas gravity, GOR, oil density), pressure, and tempera-ture. Any factors that increase the sand dry rock Pois-son’s ratio away from the quartz dry rock Poisson’sratio will impact the AVO response. Calcite cementationcan be particularly destructive because it not only has arelatively high Poisson’s ratio, but it occludes porosity,reduces permeability, increases velocity, and reducesthe sensitivity to fluids. In other words, it decreasesthe interpreter’s ability to distinguish a hydrocarbonAVO response from a brine response.

It is the interpreter’s responsibility to determinewhether a prospect is in a class 2P to class 1 setting.Knowledge of the local geology is obviously important.Substitution modeling accounting for mineralogy, claycontent, porosity, cementation, fluid properties, pres-sure, depth, and compaction must all be considered.Avseth et al. (2005) suggest a deterministic approachwhich includes comparing models, producing modelgathers with the appropriate parameters, and compar-ing against the real gathers. Avseth et al. (2003) employa depth-dependent probabilistic AVO technique thatenables the prediction of the most likely lithology

and pore fluid from seismic data even in areas of sparsewell control.

There are only five class 1 prospects in the DHIconsortium database. Only one of the five was success-ful, and each had very low DHI components of risk(<6%). This reinforces the notion that the risks onthe prospects in this setting are predominantly deter-mined by the geologic chance factors and the DHIand AVO impact are small to negligible.

ConclusionsThe interpretation of DHIs is a critical component in

the risk analysis process when exploring for oil and gas.A systematic and comprehensive methodology is re-quired to interpret DHIs and to understand the roleAVO interpretation plays in that process. Initially, aquantitative assessment of the seismic and rock physicsdata quality is necessary, followed by the interpretationof the prospect’s geologic setting. The geologic settingdictates which AVO characteristics are important andappropriate. The results from a DHI consortium com-pare various data quality elements and specific AVOcharacteristics with well outcomes. The grading ofthe most significant AVO characteristics for class 2and 3 wells demonstrates their importance as it relatesto the overall DHI database well success rates.

A significant conclusion from the DHI consortiumdatabase relates to the lower success rates of wellswhen AVO is the dominant component in the DHI riskanalysis process. This is especially pertinent for class 3prospects, where the AVO contribution of the DHI riskfor the dry holes is 64% versus 35% for the discoveries.This suggests that for class 3 prospects, non-AVO char-acteristics such as amplitude conformance to structureand flat spots play a bigger role in the DHI risk analysisevaluation. These results emphasize the necessity fora comprehensive and systematic DHI risk analysisprocess.

It is important to understand when evaluating pros-pects from a class 2P setting to a class 1, the DHI and

Figure 20. Crossplot displays P-wave veloc-ity versus density. The gray points are fromsands and shales from an array of wells fromvarious geologic settings. For sand casesA–G, the fluid substitution for gas is in redand for oil (35° API) in green. Blue pointsare 100% brine. Note all three fluids convergenear the extreme class 1 case (Sand G). Themodified Raymer Line denotes the limit ofshales with a small amount of quartz.

Interpretation / May 2014 SC73

Dow

nloa

ded

09/2

8/15

to 1

24.1

95.4

.82.

Red

istr

ibut

ion

subj

ect t

o SE

G li

cens

e or

cop

yrig

ht; s

ee T

erm

s of

Use

at h

ttp://

libra

ry.s

eg.o

rg/

AVO portion of the ultimate risk evaluation decreases.This is because the hydrocarbon contribution to theseismic response in these settings is small comparedto the rock matrix and its associated properties. Thisfurther emphasizes the importance of the geologic set-ting interpretation to perform an accurate risk analysisassessment. AVO class 1 sand prospects should berisked predominantly on a geologic basis, and thereare many oil and gas discoveries worldwide in thiscategory.

AcknowledgmentsThe authors thank the member companies of the

Rose & Associates DHI Risk Analysis Consortiumfor providing invaluable information necessary to de-velop the resulting interpretation process and prospectdatabase.

ReferencesAki, K., and P. G. Richards, 1980, Quantitative seismology:

Theory and methods: W. H. Freeman and Co.Allen, J. L., and C. P. Peddy, 1993, Amplitude variation with

offset Gulf Coast case studies: SEG.Avseth, P., T. Mukerji, and G. Mavko, 2005, Quantitative

seismic interpretation: Cambridge University Press.Bortfeld, R., 1961, Approximation to the reflection and

transmission coefficients of plane longitudinal andtransverse waves: Geophysical Prospecting, 9, 485–502.

Castagna, J. P., and M. M. Backus, editors, 1993, Offset-dependent reflectivity — Theory and practice ofAVO analysis: SEG, vol. 8.

Castagna, J. P., M. L. Batzle, and R. L. Eastwood, 1985,Relationship between compressional-wave and shear-wave velocities in clastic silicate rocks: Geophysics,50, 571–581, doi: 10.1190/1.1441933.

Castagna, J. P., and S. W. Smith, 1994, Comparison of AVOindicators: A modeling study: Geophysics, 59, 1849–1855, doi: 10.1190/1.1443572.

Castagna, J. P., and H. W. Swan, 1997, Principles of AVOcrossplotting: The Leading Edge, 16, 337–344, doi: 10.1190/1.1437626.

Castagna, J. P., H. W. Swan, and D. J. Foster, 1998, Frame-work for AVO gradient and intercept interpretation:Geophysics, 63, 948–956, doi: 10.1190/1.1444406.

Connolly, P., 1999, Elastic impedance: The Leading Edge,18, 438–452, doi: 10.1190/1.1438307.

Fahmy, W. A., 2006, DHI/AVO best practices methodologyand applications: A historical perspective: SEG/EAGEDistinguished Lecture presentation.

Fatti, J. L., G. C. Smith, P. J. Vail, P. J. Strauss, and P. R.Levitt, 1994, Detection of gas in sandstone reservoirsusing AVO analysis: A 3D seismic history using theGeostack technique: Geophysics, 59, 1362–1376, doi:10.1190/1.1443695.

Forrest, M., R. Roden, and R. Holeywell, 2010, Risking seis-mic amplitude anomaly prospects based on database

trends: The Leading Edge, 29, 570–574, doi: 10.1190/1.3422455.

Foster, D. J., R. G. Keys, and F. D. Lane, 2010, Interpreta-tion of AVO anomalies: Geophysics, 75, no. 5, 75A3–75A13, doi: 10.1190/1.3467825.

Foster, D. J., R. G. Keys, and J. M. Reilly, 1997, Anotherperspective on AVO crossplotting: The Leading Edge,16, 1233–1239, doi: 10.1190/1.1437768.

Goodway, W. N., T. Chen, and J. Downton, 1997, ImprovedAVO fluid detection and lithology discrimination usingLame petrophysical parameters: “λρ”, “μρ”, and “λ∕μ”fluid stack from P and S inversions: 67th AnnualInternational Meeting, SEG, Expanded Abstracts,183–186.

Gray, D., W. Goodway, and T. Chen, 1999, Bridging the gap:Using AVO to detect changes in fundamental elasticconstants: 69th Annual International Meeting, SEG, Ex-panded Abstracts, 852–855.

Greenberg, M. L., and J. P. Castagna, 1992, Shear-wavevelocity estimation in porous rocks: theoretical formu-lation, preliminary verification and applications: Geo-physical Prospecting, 40, 195–209.

Hampson, D., B. Russell, and B. Bankhead, 2005, Simulta-neous inversion of pre-stack seismic data: 75th AnnualInternational Meeting, SEG, Expanded Abstracts, 1633–1637.

Hendrickson, J. S., 1999, Stacked: Geophysical Pro-specting, 47, 663–705, doi: 10.1046/j.1365-2478.1999.00150.x.

Hilterman, F. J., 2001, Seismic amplitude interpretation:Distinguished instructor short course: SEG/EAGE.

Holtz, M. H., 2002, Residual gas saturation to aquifer influx:A calculation method for 3-D computer reservoirmodel construction: SPE 75502, SPE Gas TechnologySymposium.

Knott, C. G., 1899, Reflexion and refraction of elasticwaves with seismological applications: PhilosophicalMagazine, 48, 64–97.

Li, Y., J. Downton, and X. Yong, 2007, Practical aspects ofAVO modeling: The Leading Edge, 26, 295–311, doi: 10.1190/1.2715053.

Ma, X., 2002, Simultaneous inversion of prestack seismicdata for rock properties using simulated annealing:Geophysics, 67, 1877–1885, doi: 10.1190/1.1527087.

Ostrander, W. J., 1984, Plane-wave coefficients for gassands at non-normal angles of incidence: Geophysics,49, 1637–1648, doi: 10.1190/1.1441571.

Richards, P. G., and C. W. Frasier, 1976, Scattering of elas-tic wave from depth-dependent inhomogeneities: Geo-physics, 41, 441–458.

Roden, R., J. Castagna, and G. Jones, 2005a, The impact ofprestack data phase on the AVO interpretation work-flow — A case study: The Leading Edge, 24, 890–895,doi: 10.1190/1.2056369.

Roden, R., M. Forrest, and R. Holeywell, 2005b, The impactof seismic amplitudes on prospect risk analysis: TheLeading Edge, 24, 706–711, doi: 10.1190/1.1993262.

SC74 Interpretation / May 2014

Dow

nloa

ded

09/2

8/15

to 1

24.1

95.4

.82.

Red

istr

ibut

ion

subj

ect t

o SE

G li

cens

e or

cop

yrig

ht; s

ee T

erm

s of

Use

at h

ttp://

libra

ry.s

eg.o

rg/

Roden, R., M. Forrest, and R. Holeywell, 2012, Relatingseismic interpretation to reserve/resource calculations:Insights from a DHI consortium: The Leading Edge, 31,1066–1074, doi: 10.1190/tle31091066.1.

Rose, P. R., 2001, Risk analysis and management of petro-leum exploration ventures: AAPG methods in explora-tion No. 12.

Ross, C. P., 2000, Effective AVO crossplot modeling:A tutorial: Geophysics, 65, 700–711, doi: 10.1190/1.1444769.

Ross, C. P., 2010, AVO ritualization and functionalism (thenand now): The Leading Edge, 29, 532–538, doi: 10.1190/1.3422450.

Ross, C. P., and D. L. Kinman, 1995, Nonbright-spot AVO:Two examples: Geophysics, 60, 1398–1408.

Rutherford, S. R., and R. H. Williams, 1989, Amplitude-versus-offset variations in gas sands: Geophysics, 54,680–688, doi: 10.1190/1.1442696.

Shuey, R. T., 1985, A simplification of the Zoeppritzequations: Geophysics, 50, 609–614, doi: 10.1190/1.1441936.

Simm, R., R. White, and R. Uden, 2000, The anatomy ofAVO crossplots: The Leading Edge, 19, 150–155, doi:10.1190/1.1438557.

Smith, G. C., and P. M. Gidlow, 1987, Weighted stacking forrock property estimation and detection of gas: Geophy-sicsical Prospecting, 35, 993–1014.

Verm, R., and F. Hilterman, 1995, Lithology color-coded seismic sections: The calibration of AVO cross-plotting to rock properties: The Leading Edge, 14,847–853.

Whitcombe, D. N., P. A. Connolly, R. L. Reagan, andT. C. Redshaw, 2002, Extended elastic impedance forfluid and lithology prediction: Geophysics, 67, 63–67.

Zoeppritz, K., 1919, Erdbebenwellen VIIIB, On the reflec-tion and propagation of seismic waves: Göttinger Nach-richten, I, 66–84.

Rocky Roden received a B.S. inoceanographic technology-geologyfrom Lamar University and an M.S.in geological and geophysical ocean-ography from Texas A&M University.He runs his own consulting company,Rocky Ridge Resources, Inc., andworks with numerous oil companiesaround the world on interpretation

technical issues, prospect generation, risk analysis evalu-ations, and reserve/resource calculations. He is a principalin the Rose and Associates DHI Risk Analysis Consortium,developing a seismic amplitude risk analysis program andworldwide prospect database. He has also worked withSeismic Microtechnology, Geophysical Insights, and RockSolid Images on the integration of advanced geophysicaltechnology in software applications. He has been involvedin numerous oil and gas discoveries around the world and

has extensive knowledge of modern geoscience technicalapproaches (past chairman, The Leading Edge EditorialBoard). As Chief Geophysicist and Director of AppliedTechnology for Repsol-YPF (retired 2001), his role com-prised advising corporate officers, geoscientists, and man-agers on interpretation, strategy, and technical analysis forexploration and development in offices in the UnitedStates, Argentina, Spain, Egypt, Bolivia, Ecuador, Peru,Brazil, Venezuela, Malaysia, and Indonesia. He has beeninvolved in the technical and economic evaluation ofGulf of Mexico lease sales, farmouts worldwide, and bidrounds in South America, Europe, and the Far East. Pre-vious work experience includes exploration and develop-ment at Maxus Energy, Pogo Producing, Decca Survey,and Texaco.

Mike Forrest received a B.S. (1995)in geophysical engineering from St.Louis University. He had a 37-yearcareer with Shell Oil as a geophysi-cist and executive, including GeneralManager Exploration for Gulf ofMexico during the mid 1980s whenShell expanded exploration into deepwater, and he worked with Maxus

Energy for five years as an executive. Since 2001, hehas been Chairman of the Rose & Associates DHI (DirectHydrocarbon Indicator) Interpretation and Risk Ana-lysis Consortium that currently has 25 oil company mem-bers. He has over 40 years of experience interpreting andrisking seismic amplitude anomalies (DHIs). He is amember of SEG (60 years) and AAPG, a Director ofthe SEG Foundation, and Director of Global GeophysicalServices.

Roger Holeywell received a B.S. ingeology from Texas A&M University,an M.S. in geology from Brown Uni-versity, and a master’s in finance fromthe University of Texas at Dallas. Hehas been working in the field of riskanalysis since the early 1990s and re-cently celebrated his 15th year as afreelance software developer for Rose

& Assoc. His software projects include MMRA, PortfolioEssentials, Risk Essentials, Multizone Master, PlayRA,and Portfolio Rollup. He has also been involved inR&A’s DHI Consortium since its inception 12 years agoand is the author of the Consortium’s primary softwaretool, SAAM. He also holds a full-time position as AdvancedSenior Consultant for Marathon Oil Co. in Houston wherehe is currently the manager of geoscience data services.He has previously held positions as Senior ExplorationGeologist, Exploration Manager and Exploration Co-ordinator, among others. His work with Rose & Assoc. iswith the approval and cooperation of Marathon, who is along time R&A client as well as a previous member of theDHI Consortium.

Interpretation / May 2014 SC75

Dow

nloa

ded

09/2

8/15

to 1

24.1

95.4

.82.

Red

istr

ibut

ion

subj

ect t

o SE

G li

cens

e or

cop

yrig

ht; s

ee T

erm

s of

Use

at h

ttp://

libra

ry.s

eg.o

rg/

Matthew Carr received a B.S. fromOld Dominion University and M.S.and Ph.D. degrees from the Universityof South Carolina. His expertise in-cludes the integration of petrophysicsand seismic rock properties, rockphysics, sandstone and carbonatepetrology and its relation to petrophy-sics, petrographic image analysis, in-

terpretation of nuclear magnetic resonance data innatural porous media, as well as multivariate statisticalanalysis of geologic systems. He has pioneered severaltechniques and applications for the integration of well dataand seismic data. In addition, he possesses certification asa Professional Geophysicist from the State of Texas, he isan AAPG certified Professional Petroleum Geophysicist. Inhis more than 23 years in the industry, he has contributedto multiple large discoveries (both domestically and inter-nationally). In addition to major discoveries, he has helpedmany companies to quantify risky exploration targets, re-sulting in the redirection of exploration efforts that would

offer a better success rate. He is currently the managingdirector and lead scientist at QI Petrophysics, focusedon quantification of rock properties to mitigate risk.

Patsy Alexander received under-graduate degrees in geoscience (cumlaude), geography, and computer pro-gramming from the University ofTexas at Dallas, the University ofNorth Texas, and Mountain View Col-lege (Dallas), respectively. With morethan 20 years in the oil industry, shehas worked as an exploration geolo-

gist, E&P database administrator, computer programmer,and data analyst. Since 2007, she has performed databasemanagement and data mining and analysis for the DHIConsortium team to identify statistically significant trendsin the group’s DHI Prospect database, as a supplement tothe statistics generated by the SAAM software.

SC76 Interpretation / May 2014

Dow

nloa

ded

09/2

8/15

to 1

24.1

95.4

.82.

Red

istr

ibut

ion

subj

ect t

o SE

G li

cens

e or

cop

yrig

ht; s

ee T

erm

s of

Use

at h

ttp://

libra

ry.s

eg.o

rg/