Watch videos of Presentations from SEG 2017
Using Attributes to Interpret the Environment of Deposition - A Video Course. Taught by Kurt Marfurt, Rocky Roden, and ChingWen Chen
Dr. Kurt Marfurt and Dr. Tom Smith featured in the July edition of AOGR on Machine Learning and Multi-Attribute Analysis
Rocky Roden and Ching Wen Chen in May edition of First Break - Interpretation of DHI Characteristics using Machine Learning
Seismic interpretation and machine learning by Rocky Roden and Deborah Sacrey, GeoExPro, December 2016

LOGIN / MY PROFILE | LOG OUT

Reveal anomolies via SOM and PCA in interpretation

SEG 2015 | Patricia Santogrossi

An excerpt:

Extracted platform and basinal carbonate play areas. Optimized potential TEST location. Distinguished good from poor productivity areas (note limited productivity data). Helped assess efficacy of perfs for different play types. Identified under perfed zones. Enabled easy high-grading of leasehold for three plays

...read more >

Full Article Text...

Finding sub-seismic features using SOM and PCA

Slide 2

What Paradise Can Do That Other Software Cannot

  • Combines multiple attributes on 2D as well as 3D data
    • Quickly -a 2D line in a twinkling; 10 sq. miles w/ 40 attributes in minutes, 50 squares with 60-70 attributes in under an hour.
  • Software automatically documents analysis details, results.
  • Normalized results eliminates tuning, velocity concerns
  • Characterizations delimited by horizons; guide by key line (s)
  • Identifiesand quantifiesthe key attributes in the suite
  • Classifiesdata in clusters; shows likely anomalies
  • Utilizes transparency to focus

Slide 3

Results Part 1 - Viola

  • Characterization resultsof the Viola Chocolate Brown platform carbonates
    • Suggest TEST location could be moved
    • Identified the low productivity area tested by Key well #1
    • Characterized high productivity Key well #2 in different faciestract; indicated additional potential perftargets
  • Characterization results in Huntontrends
    • Explained moderate cumulative production in Key well #3
    • Identified 2 large play areas of Huntonpotential
  • Demonstrated ability to reveal possible karst features
  • Help to locate wells & perfzones; retain critical leases!

Slide 4 - Platform on Viola Base and Inline 121

Can identify 8 fault-bounded SW-NE segments, one NW-SE frontal segment!

Slide 5 - Will the Real Chocolate Brown Please Stand Up?

Slide 6

Client Provided Data

  • 10.6 mi2Enhanced Migrated Stack seismic data
    • Time volume 0-5s at 2ms sample interval
  • 3 interpolated horizons
    • Top Hunton, Top and Base Viola
  • Approximately 141 wells
    • 6 with log curves; 1 TEST well location
  • Base Map
  • 3 Well Tie Lines
  • Perforation data for 156 wells
  • Cumulative Production Data for 4 wells
    • Key well #0 (Not used in study), Key Well #1, Key Well #2, Key well #3

Slide 7

GI Produced Products

  • Mapped three additional events
    • Unconformity, Top Sylvan, and Arbuckle
    • Mapping combined and fitted for use in Paradise
  • 3 NW-SE diagonal arb lines resolved fault pattern
    • Mapped three faults
  • Ran Rock Solid Attributes (22) and Spectral Decomposition Volumes
    • Generated 15 volumes 4-74 Hz
    • Dominant Frequency is 24 Hz
  • Loaded all well, horizon, and segydata into Paradise

Slide 8 - Unconformity Structure Map with 58 Wells

Slide 9 - SW-NE Arb Line Through Test

Slide 10 - Northern Arbitrary Line NW-SE

Slide 11

Paradise - A Unique Integration of Processes

  • Principal Component Analysis (PCA)
    • Identifiesthe most significant attributes in the data set
    • Quantifiesthe relative contribution of each attribute
  • Self Organizing Maps (SOM)
    • Applies unsupervised neural networks to simultaneously classifymultiple attributes
    • Reveals natural clusters, shown by 2D colormaps, in seismic data
  • Results exposed critical aspects “hidden” in seismic data re
    • Structure
    • Stratigraphy
    • Rock properties such as porosity and facies
    • Possible diageneticor fluid changes

Slide 12

Workflow

  • Perform Principal Component Analysis
    • Developed SOMs from PCAs from first 4 Eigen Values, all attributes
    • Developed SOMs with a recipeof top 11 attributes from first 4 Eigen Values
    • Looked for auto-highlighted anomalous areas
    • Noted apparent facieschanges in color patterns
  • Tested varying Parameter sets
  • Exported Classification & Probability volumes to interpretation software

Slide 13 - Two Ways to Make a SOM

Slide 14

SOM3 to Evaluate Production from Key Well #2

"Recipe"

EV1: MM Trace Envelope
MM Sweetness
EV2: MM Imaginary Part
MM RAI
MM Instant. Phase
EV3: Instant. Frequency
Thin Bed
Smoothed Frequency
EV4: MM Normalized Ampl.
MM Enh. Migr. Stack
Phase Breaks/
MM Paraphase

Slide 15 - Parameter Sets and Granularity

Slide 16

Results Part 1B

  • Characterization results in the Viola Chocolate Brown platform carbonates
    • Suggest the TEST location should be moved
    • Identified the low productivity area tested by Key Well #1
    • Characterized high productivity in Key Well #2 in different faciestract; indicated additional potential perftargets
  • Suggest better Huntontrends
    • Explained moderate Key Well #3 cumulative production
    • Identified 2 large play areas for Huntonpotential
  • Demonstrated ability to reveal possible karst features
  • Help to locate wells & perfzones; retain critical lease!

Slide 17 - Inline 55 for Key Well #2 Well-Peres

Slide 18 - Inline 55 Through Key Well #2

Slide 19 - Inline 55 Through Key Well #2

Slide 20 - Inline 55 Through Key Well #2 - Basinal Facies?

Slide 21

Results Part 2 - Hunton

  • Characterized the Viola Chocolate Brown platform carbonates
    • Suggests TEST location should be moved
    • Identifies the low productivity area tested by Key Well #1
    • Characterizes high productivity Key Well #2 in different faciestract; indicates additional potential perftargets
  • Suggest better Huntontrends
    • Explains moderate Key Well #3 cumulative production
    • Identifies 2 large play areas of for Huntonpotential
  • Demonstrates ability to reveal possible karst features
  • Helps locate wells & perfzones; retain critical leases!

Slide 22 - Hunton "Type" Well

Slide 23 - Hunton "Type" Well

Slide 24 - Key Well #3 - Hunton "Type" Well

Slide 25 - Hunton Truncation Trap Anomaly

Slide 26 - Hunton/Sylvan 2nd Objectives at Key Well #2

Slide 27 - Uncalibrated hunton Anomalies

Slide 28 - Geometric SOM of Top 5 Curvature Attributes

Slide 29 - Hunton Anomalies on SOM 3 - Workflow

Slide 30

Value Adds Due to Paradise for This Project

In Summary

  • Extracted platform and basinalcarbonate play areas
  • Optimized potential TEST location
  • Distinguished good from poor productivity areas (note limited productivity data)
  • Helped assess efficacy of perfsfor different play types
  • Identified underperfedzones
  • Enabled easy high-grading of leasehold for three plays
  • Results exposed critical aspects “hidden” in Client’s seismic data re
    • Structure: Revealed unknown karst features
    • Stratigraphy: Enhanced key geometries
    • Rock properties: Flat spots in Hunton, Viola and Simpson indicated likely faciescontrasts, possible diageneticchanges, and or fluid effects