Using Seismic Attributes

Presentation on theme: "Using Seismic Attributes"— Presentation transcript:

Using Seismic Attributes
Lecture 13 Using Seismic Attributes Horizon A Horizon B Slide 1 Introduction Images appear within lecture Good Seal Good Reservoir W1 W2 W3 W5 W4 W6 W7 W8 W9 Courtesy of ExxonMobil L Seismic Attributes

Outline Review causes of seismic response
Modeling the seismic response What are seismic attributes? Overview of seismic attribute applications Qualitative analyses Exercise: Mapping depositional environments Quantitative analyses Exercise: Predicting average porosity Slide 2 Introduction This is the outline for this unit We will: Review causes of seismic response Talk about modeling the seismic response Define what seismic attributes are Give an overview of seismic attribute applications Qualitative analyses Quantitative analyses Courtesy of ExxonMobil L Seismic Attributes

Thud! Ring! Seismic Response What causes a seismic response?
1. Changes in bulk-rock velocity or density Lithology (e.g., sandstone, shale, limestone, salt) Slide 3 What causes a seismic response? It is changes in the velocity and/or density of the rock with fluid in the pore space Several factors effect the velocity and density of bulk-rock The most common factor is lithology e.g., sandstones generally have different velocity and density values than shales or limestone. Consider the last time you struck a hammer or pick against a rock If you struck a dense limestone, the hammer would bounce back rapidly and the sound would have a high pitched ring However, if your hammer struck a less dense shale, it would only yield a low pitched ‘thud’ The frequency of the sound you hear when your hammer strikes a rock is proportional to the density of that rock Limestone Ring! Thud! Shale Courtesy of ExxonMobil L Seismic Attributes

Seismic Response What causes a seismic response?
1. Changes in bulk-rock velocity or density Lithology (e.g., sandstone, shale, limestone, salt) Porosity (e.g., intrinsic, compaction, diagenesis) Slide 4 Another factor is porosity Since the velocity of sound through water is much slower than the velocity of sound through rock, rocks that have a lot of pore-space are ‘slower’ than those that have lower porosity. Fast Slow Courtesy of ExxonMobil L Seismic Attributes

What causes a seismic response?
1. Changes in bulk-rock velocity or density Lithology (e.g., sandstone, shale, limestone, salt) Porosity (e.g., intrinsic, compaction, diagenesis) Mineralogy (e.g., calcite vs. dolomite, carbonaceous shales) Slide 5 The mineral composition of the rocks also affects the velocity and density For example, carbonaceous material (such plant and animal matter) has a lighter density than most minerals Therefore, organic-rich source rocks commonly have a lower velocity and density than surrounding rocks Unfortunately, this difference is often too subtle to detect with typical seismic data Courtesy of ExxonMobil L Seismic Attributes

Seismic Response What causes a seismic response?
1. Changes in bulk-rock velocity or density Lithology (e.g., sandstone, shale, limestone, salt) Porosity (e.g., intrinsic, compaction, diagenesis) Mineralogy (e.g., calcite vs. dolomite, carbonaceous shales) Slide 6 Velocity and density are also affected by the type and density of the pore fluid For example, assume we have a cubic meter of rock with a porosity of 30% Let’s also assume that the density of the solid matrix is equal 2.65 g/m3 If the pore spaces of the rock were filled with salt water, the bulk density would be g/m3 However, if the pore spaces were filled with methane, the bulk density would only be g/m3 Fluid type and saturation (water, oil, gas) Sandstone with 30% Porosity: Pore Fluid Density Salt Water Fresh Water Oil Gas Courtesy of ExxonMobil L Seismic Attributes

Modeling the seismic response:
Seismic Modeling Modeling the seismic response: Determine bulk-rock velocity and density Calculate impedance (Recall: I = ρ x v) Represent impedance changes as reflection coefficient Convolve seismic wavelet to reflection coefficients Slide 7 We can model the seismic response at the boundary between two rock units if we know the velocity and density of each unit We calculate the reflection coefficient using the formula shown We then use a mathematical process called convolution to model the seismic response I2 - I1 I2+ I1 RC= Courtesy of ExxonMobil L Seismic Attributes

The Convolution Method
Reflection Coefficients Lithology Velocity Density Impedance Wavelet Model = x * Shale Sand Slide 8 This slide shows the basic steps in modeling the seismic response We multiply the velocity and the density to get the impedance Next we calculate the reflection coefficient at each significant rock boundary an increase in impedance across a boundary gives a positive RC a decrease in impedance across a boundary gives a negative RC We extract or assume the pulse shape or wavelet Each RC is convolved with the wavelet The responses for each individual RC are summed to get a modeled trace Courtesy of ExxonMobil L Seismic Attributes

Seismic Modeling W E Wedge Modeling
A wedge model is used to display the interactions of reflection coefficients as the thickness changes Note how the ‘middle peak’ changes amplitude, shape, and duration as the sand thins to the east Slide 9 If RCs are close together (thin rock units), then the seismic wavelets will interfere with one another We can build simple wedge models to display the interactions of reflected waves as the thickness changes Note on this slide how the ‘middle peak’ changes amplitude, shape, and duration as the sand thins to the east The ‘middle peak’ disappears about mid way across the model Courtesy of ExxonMobil L Seismic Attributes

What are seismic attributes?
Definition What are seismic attributes? Seismic attributes are mathematical descriptions of the shape or other characteristic of a seismic trace over specific time intervals. Slide 10 What are seismic attributes? Seismic attributes are mathematical descriptions of the shape or other characteristic of a seismic trace over specific time intervals Courtesy of ExxonMobil L Seismic Attributes

Why are seismic attributes important?
Importance / Benefits Why are seismic attributes important? Our increasing reliance on seismic data requires that we extract the most information available from the seismic response Seismic attributes enable interpreters to extract more information from the seismic data Applications include hydrocarbon play evaluation, prospect identification and risking, reservoir characterization, and well planning and field development Slide 11 Why are seismic attributes important? Our increasing reliance on seismic data requires that we extract the most information available from the seismic response Seismic attributes enable interpreters to extract more information from the seismic data Applications include: hydrocarbon play evaluation, prospect identification and risking, reservoir characterization, and well planning and field development Courtesy of ExxonMobil L Seismic Attributes

Classes of seismic attributes?
Single-Trace Types Classes of seismic attributes? Horizon (loop) Horizon A Peak amplitude Duration Symmetry Slide 12 There are several classes of seismic attributes The more common attributes are calculated trace-by-trace; they are single-trace attributes Single-trace attributes can be measurements for a certain loop ( loop = a peak or a trough) a specific time interval (e.g., from Horizon A to Horizon B) a series of successive data samples – the instantaneous attributes Interval Average amplitude Maximum (Minimum) Duration Isochron Sample (volume, instantaneous) Amplitude Time Frequency Horizon B Courtesy of ExxonMobil L Seismic Attributes

Classes of seismic attributes?
Multi-Trace Types Classes of seismic attributes? Trace A Trace B Multi-Trace Dip / azimuth Coherency Slide 13 Another class of seismic attribute involves the simultaneous analysis more than one trace For example, an interpreter will map (correlate) a peak or trough through an area Software can then calculate two attributes the dip of the interpreted horizon (in msec/trace) the azimuth – the compass direction of the maximum dip direction (e.g., 15° East of North) Coherency is a popular multi-trace attribute in industry it is a measure of similarity between a window of one trace relative to its neighboring trace or traces we discussed coherency in the structural lecture (Unit 10) it is a volume-based attribute – not needing an interpreted horizon Correlation Window R2 = 0.92 Amplitude A Amplitude B Courtesy of ExxonMobil L Seismic Attributes

Multi-Trace Types Dip map Faults Stratigrahic features
Slide 14 This is a display of the DIP attribute calculated along an interpreted horizon Low dips are light; high dips are dark Two types of geologic features are revealed: linear segments of high dip associated with fault offsets of the horizon curvilinear segments of high dip associated with channels and other stratigraphic features How do we separate structural and stratigraphic features? structural features are consistent over many time slices stratigraphic features change rapidly from one time slice to the next (shallower or deeper) Stratigrahic features Courtesy of ExxonMobil L Seismic Attributes

Seismic attribute applications:
Qualitative Data quality; seismic artifact identification Seismic facies; depositional environment Slide 15 Seismic attributes serve two basic purposes: Qualitative evaluations checking seismic data quality – identifying artifacts (non-geologic features) in the data performing seismic facies mapping to predict depositional environments Quantitative analyses using well data to provide calibration, we find an equation that uses one or more attributes to predict a rock or fluid property, such as: Reservior thickness Lithology Porosity Type of fluid fill (water, oil, gas) The next few slides will show some examples Quantitative Equations relating rock property changes to changes in seismic attributes Reservoir thickness Lithology Porosity Type of fluid fill Courtesy of ExxonMobil L Seismic Attributes

Data Quality Analysis (Artifact detection):
Qualitative Analyses Data Quality Analysis (Artifact detection): Identify zones where seismic data quality is adversely affected by acquisition or processing methods or by geologic interference. Acquisition gaps, Inline-parallel striping Multiples, migration errors, incorrect velocities Improper amplitude and phase balancing Frequency attenuation Overlying geology (e.g., shallow gas, channel) Slide 16 This slide outlines some of the things we can use qualitative attribute analysis for to terms of data quality We can identify such things as: Acquisition gaps, Inline-parallel striping Multiples, migration errors, incorrect velocities Improper amplitude and phase balancing Frequency attenuation Artifacts due to the overlying geology (e.g., shallow gas, channel) Courtesy of ExxonMobil L Seismic Attributes

Data Quality Analysis (Artifact detection):
Inline-parallel acquisition striping at water bottom (~ 40 ms) Slide 17 Here is an amplitude-based attribute map at a level about 40 ms (10 data samples) below a shallow water bottom Note the east-west trending “stripes” This is an acquisition artifact associated with a multi-streamer boat collecting the data by sailing east-west and west-east These ”stripes” tend to “heal” with depth; they may not impact our analysis of a deep reservoir interval Inline Direction Courtesy of ExxonMobil L Seismic Attributes

Data Quality Analysis (Artifact detection):
Inline-parallel acquisition striping at 1000ms Slide 18 This is the same attribute that was displayed on the previous slide, but at a depth of 1000 msec (1 sec) below the seafloor The east-west “stripes” are still present, but at a much reduced level This demonstrates how the acquisition ”stripes” tend to “heal” with depth Inline Direction Courtesy of ExxonMobil L Seismic Attributes

Seismic facies mapping:
Qualitative Analyses Seismic facies mapping: Facies are packages of rocks that exhibit similar characteristics (e.g., lithofacies, petrophysical facies, depositional facies) Seismic facies are packages of seismically-defined bodies that exhibit similar seismic characteristics (e.g., reflection geometry, amplitude, continuity, frequency). Environment of Deposition (EoD) can be interpreted from patterns of seismic facies (i.e., similar seismic attributes) Slide 19 Another qualitative applications is seismic facies mapping Facies are packages of rocks that … Seismic facies are … Environment of Deposition (EoD) can be … Courtesy of ExxonMobil L Seismic Attributes

Seismic Facies Mapping Exercise
Qualitative Analyses Seismic Facies Mapping Exercise Slide 20 You will be doing a simple seismic facies mapping exercise This seismic line has been flattened (datumed) on the green horizon This removes post-depositional tilting Our interest is at the orange horizon We want to determine where there is good reservoir rocks below the orange horizon with good seal rock just above it Orange Datum Courtesy of ExxonMobil L Seismic Attributes

Seismic Facies Mapping Exercise
Conceptual Depositional Model: Stacked, prograding fluvial to nearshore to offshore siliciclastic parasequences Magenta Slide 21 This is the conceptual model for deposition during the period we are interested in This area was characterized by nearshore to offshore deposition Note at the orange horizon Below the horizon – there are fluvial and nearshore deposits Above the horizon – there are offshore shale deposits Orange Fluvial shales - sands Nearshore sands Offshore shales Courtesy of ExxonMobil L Seismic Attributes

Seismic Facies Mapping Exercise
Conceptual Depositional Model: Prograding sands increase in porosity upwards before being capped by variable quality marine shale. Marine Shale (seal) Magenta Slide 22 At any location: We could have poor to good reservoir quality (sand vs. silt and shale, reworking by waves, etc) We could have fair to excellent seal quality (marine shales diluted with some to no silt) Porous Sand (reservoir) Marine Shale (seal) Orange Fluvial (reservoir) Porous Sand (reservoir) Marine Shale (seal) Courtesy of ExxonMobil L Seismic Attributes

Seismic Facies Mapping Exercise
Modern Analog: Fluvial to nearshore progression resulting in wave dominated, barrier island complex (Texas Gulf Coast) Slide 23 These units were deposited in an environment like this – a barrier island The nearshore (beach) sands would be great reservoir rocks (medium to course sand with good sorting) Lagoonal muds and offshore shales would be good sealing lithologies Courtesy of ExxonMobil L Seismic Attributes

Seismic Facies Mapping Exercise
Modeled Seismic Response Seismic modeling indicates the following response to changes in reservoir and seal quality: Good Seal Good Reservoir Poor Reservoir Poor Seal Strong Peak Strong Trough Moderate Trough Moderate Peak Slide 24 Some seismic models were generated for this area The pulse is called quadrature phase – similar to a minimum phase response The orange horizon occurs at a zero-crossing Attribute of the peak above the orange horizon indicate the properties of the seal Attribute of the trough below the orange horizon indicate the properties of the reservoir a strong (high positive amplitude) peak over a strong (high negative amplitude) trough is what we want to find if the peak (trough) is moderate or low amplitude, it indicates less favorable seal (reservoir) Courtesy of ExxonMobil L Seismic Attributes

Seismic Facies Mapping Exercise
Objective: Identify areas where good-quality seal rocks overlay good-quality reservoir rocks Available data / tools: Slide 25 So the objective of the exercise is to identify areas where good-quality seal rocks overlay good-quality reservoir rocks You are given: Two seismic attribute maps Orange time structure map Depositional model and seismic response Tracing paper and pencils Seismic attribute maps Orange time structure map Depositional model and seismic response Tracing paper and pencils Courtesy of ExxonMobil L Seismic Attributes

Seismic Facies Mapping Exercise
What We Want Good Seal Good Reservoir Strong Peak Strong Trough Poor Seal Good Reservoir Good Seal Poor Reservoir Poor Seal Poor Reservoir Slide 26 Here again is the key – where is a strong peak followed by a strong trough centered on the orange horizon? Moderate Peak Strong Trough Strong Peak Moderate Trough Moderate Peak Moderate Trough Have Some Fun Mapping! Courtesy of ExxonMobil L Seismic Attributes

Seismic attribute applications:
Qualitative Data quality; seismic artifact identification Seismic facies; depositional environment Slide 27 We have seen a few qualitative applications of seismic attributes Now we will consider some more quantitative applications Quantitative Equations relating rock property changes to changes in seismic attributes. Reservoir thickness Lithology Porosity Courtesy of ExxonMobil L Seismic Attributes

Quantitative Analyses
Quantitative Seismic Attribute Analysis Requirements: Controlled Amplitude, Controlled Phase processing Data quality reconnaissance Good well-seismic ties Sufficient well control (additional seismic modeling is usually necessary) Slide 28 There are several requirements to perform a quantitative attribute analysis Some are related to the seismic data; others relate to calibration data from wells In terms of the seismic data: The data processing should be Controlled Amplitude and Controlled Phase Data quality should be high and checked by doing some reconnaissance All wells should be tied to the seismic data (synthetics) and the results should be high-quality ties In terms of calibration: There should be sufficient well control so that variations in rock or fluid properties can be related to variations in one or more seismic attributes We can use 1-D models to give more calibration points (e.g., if the well has a 50 m gas zone, we could model the seismic response and attribute characteristics of a similar gas sand that is 100 m or 25 m thick) Courtesy of ExxonMobil L Seismic Attributes

Quantitative Analyses
Quantitative Attribute Exercise Goal: Build a correlation between seismic attributes and sand thickness to predict areas of high reservoir producibility. Tools: Seismic - well log (i.e., rock property) models Slide 29 We will go through a quantitative exercise together Our goal is to build a correlation between seismic attributes and sand thickness If we can do this, it will allow us to predict areas of high reservoir producibility. We will use modeled seismic and well data Courtesy of ExxonMobil L Seismic Attributes

Geologic Description Backstepping, unconfined sheet-sands comprising two multicycle reservoirs separated by a marine shale Slide 30 Our modeled reservoir is shown on this slide There are two high sand-percent intervals (high Net:Gross) separated by a marine shale Courtesy of ExxonMobil L Seismic Attributes

Attribute Response Which seismic attributes differentiate average sand thickness? Sand Shale Sand Shale Sand Shale Slide 31 We can extract from the reservoir model the sand % and how the sands are stacked at various locations This slide shows 3 of the 9 locations we will use to calibrate seismic attributes to average sand thickness Well 2 Well 6 Well 9 Courtesy of ExxonMobil L Seismic Attributes

Calibration Duration Amplitude Maximum Average Minimum
50 100 150 200 250 300 30 40 60 70 80 90 Maximum Loop Duration (ms) Measured Average Sand Thickness (ft) 35 45 55 65 75 Average Loop Duration (ms) 10 20 Minimum Loop Duration (ms) 50 100 150 200 250 300 40 60 80 120 140 160 Average Positive Amplitude Measured Average Sand Thickness (ft) Amplitude 90 110 130 170 180 Maximum Amplitude 5 10 15 20 25 30 35 Average Amplitude Maximum Slide 32 Here we have 6 cross-plots – each has sand thickness on the vertical axis and a seismic attribute on the horizontal axis The attributes for the top row: maximum time duration and maximum amplitude The attributes for the middle row: average time duration and average amplitude The attributes for the bottom row: minimum time duration and minimum amplitude QUESTION: Of these 6 attributes, does any show a simple trend (linear is nice) so that from the attribute value we can predict sand thickness with a fair degree of confidence? ANSWER: There is one – average amplitude – has a simple, near-linear trend Average Minimum Courtesy of ExxonMobil L Seismic Attributes

Seismic Attribute Calibration
300 Thickness = APA R2 = 0.869 250 200 Slide 33 This is the “well behaved” attribute plotted against sand thickness From the last slide, the “well behaved” attribute is average amplitude A linear regression was obtained (an equation) that fits the data fairly well (R2 value of 0.87) As an example of how we can use this, if at a given location the attribute value is 100, then we would predict the sand thickness to be 150 ft (from 100 on the X axis, go up to the yellow line and then look across to the Y value at that level) Measured Average Sand Thickness (ft) 150 100 50 40 60 80 100 120 140 160 Average Positive Amplitude Courtesy of ExxonMobil L Seismic Attributes

Input Seismic Attribute
W1 W2 W3 W5 W4 W6 W7 W8 W9 Slide 34 Here is the input seismic attribute – average amplitude It varies from 55 to 145 amplitude units We can put the amplitude values into our equation and predict the sand thickness Average Amplitude Low High 55 70 95 110 125 140 Courtesy of ExxonMobil L Seismic Attributes

RESULT W1 W2 W3 W5 W4 W6 W7 W8 W9 Thin Thick Average Sand Thickness 60
Slide 35 Here is the result of using this equation Since the equation had only one attribute and a linear relationship, the colors are the same HOWEVER, the values are now predicted FEET of sand Thin Thick Average Sand Thickness 60 80 100 120 140 160 feet Courtesy of ExxonMobil L Seismic Attributes

Quantitative Analysis: A Brief Example
An Oil Field, Onshore Alabama Porosity in the Upper Smackover No Porosity in the Upper Smackover Impedance Impedance Slide 36 Next we will review an example of a quantitative attribute analysis using real data This study is for an oil field from onshore Alabama The reservoir is a carbonate interval called the Smackover formation In some locations, the upper Smackover is porous and contains oil; elsewhere it is tight (no porosity) We want to use seismic data to predict where the Smackover is porous Haynesville Haynesville Porous Zone Tight Smackover Smackover Norphlet Norphlet Courtesy of ExxonMobil L Seismic Attributes

Change Porosity -> Change Seismic Response
in the Smackover No Porosity in the Smackover Representative In-Line 2.82 2.82 2.84 2.84 Slide 37 Here is a seismic line that connects two well locations The well on the left has porosity (and oil); the well on the right has no porosity NOTE the black reflection cycle (a trough) associated with the upper Smackover on the left, it is not as black (not as negative) as it is on the right also, on the left there is a thin interval of white that dies out to the right These are subtle differences that may be related to the presence/absence of porosity It would be hard to visually use these observations; but seismic attributes make it relatively easy 2.92 2.92 The trough is lower in amplitude and loop duration is longer The trough is higher in amplitude and loop duration is shorter Mapped Horizon (white) Courtesy of ExxonMobil L Seismic Attributes

1-D Seismic Modeling Changing the porosity in the Upper Smackover in 1-D models confirms there is a seismic signature related to porosity 16 ft Porous Zone 10 ft Porous Zone 3 ft Porous Zone Slide 38 We can check to see if some attributes are sensitive to the amount of porosity Here we have a “real” well and two “pseudo-wells” The real well had a 10 foot thick porous zone We edited the well data to give us two “pseudo-wells” (modifications of a real well) On the left is a pseudo well with 16 feet of porosity On the right is a pseudo well with 3 feet of porosity We also show a modeled trace for all three NOTE how the trough varies in amplitude and shape as the thickness of the porous zone changes Using seismic attributes, we can capitalize on these subtle variations Haynesville Smackover Norphlet Courtesy of ExxonMobil L Seismic Attributes

Attribute Calibration & Evaluation
Porosity for the Smackover Predicted based on 4 attributes Calibration based on 8 wells Actual Average Smackover Porosity 5 3 2 1 8 9 7 6 4 Predicted Average Smackover Porosity Best Fit 95% C.I. Slide 39 Here is a cross-plot in which: The horizontal axis is porosity thickness (data points are only from actual wells) The vertical axis is our predicted porosity thickness from an equation that uses 4 seismic attributes as terms (each attribute has a different weight) A best linear fit was derived (mathematically) and 95% confidence intervals are shown We could use pseudo-well to add more calibrations points, but that was deemed unnecessary for this case since we have 8 wells with a good spread of values Courtesy of ExxonMobil L Seismic Attributes

A Predicted Porosity Map
Applying the derived attribute “equation” to the 3D seismic survey resulted in a Smackover porosity map Possible New Well Location Slide 40 We use the equation (based on 4 attributes) to derive a map of predicted feet of porosity throughout the area Hot colors are where we predict the thickest porosities exist There are some “edge effects” near the boundaries of the survey that have to be ignored The zoomed in area gives a high level of detail for a known producing zone We can use this detail to position additional production wells This map was also used to identify other drill locations for near-field wildcats 18% porosity Courtesy of ExxonMobil L Seismic Attributes

Potential Pitfalls / Solutions
Inadequate well control: Wells don’t represent all variability within reservoir Use seismic modeling to infill gaps Redundant attributes Different attributes highly correlated to one another Remove redundant attributes; keep one that correlates best with rock property Linear correlation Nonlinear correlation may be better representation Test other nonlinear correlation schemes but be aware of extrapolation problems Slide 41 This slide lists some attribute pitfalls and some of the possible solutions/remedies Courtesy of ExxonMobil L Seismic Attributes

Summary Seismic attributes describe shape or other characteristics of a seismic trace over specific intervals or at specific times Seismic attributes are important because they enable interpreters to extract more information from seismic data Seismic attributes can be derived from a single-trace or by comparison of multiple traces Three common types of single-trace attributes are horizon-, interval-, and sample-based Seismic attributes are used for qualitative analysis (e.g., data quality, seismic facies mapping) and quantitative analysis (e.g., net sand, porosity prediction) Slide 42 In summary …. Courtesy of ExxonMobil L Seismic Attributes

Similar presentations