Pattern recognition methods

Pattern recognition methods - Patternrecognitionmethodsin

Info iconThis preview shows pages 1–8. Sign up to view the full content.

View Full Document Right Arrow Icon
    Pattern recognition methods in  seismic data interpretation
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
    Seismic data Seismic data: Pre-inversion: amplitude data methods:  geostatistical inversion    calibration of amplitude data into a  facies probability cube Post-inversion: elastic/acoustic impedance methods:   use as locally varying mean     co-simulation (using co-kriging)     block co-simulation    
Background image of page 2
    Using seismic data Seismic informs lateral continuity, various approaches   Locally varying mean  2D:  Vertical average  of petro-physical property       (Xu, 1992, Behrens, 1996, Yao, 1998)  3D:  Co-kriging models  (Journel, 1998) MM I  : screening of further away secondary by co-located primary MM II  : screening of further away secondary by co-located secondary
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
    Facies: amplitude calibration, example Amplitude Window with amplitude feature A Cluster label Cluster 15 B Cluster 15/16 C Channel sand probability D Filtered probability E Mudchannel fault
Background image of page 4
    Seismic data as local  probability “Seismic data s” porosity From well-logs 0 35 5000 1000 1500 3000 3500 4000 Porosity when s=1500 Porosity when s=3500 Porosity when s=3500 Porosity when s=4000 k k y( ; z ) Pr(Z(u ) z | S(u )=s) [0,1] α α α = u z k
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
    Need for improvement Common to most traditional approaches: Assumption of smoothness of seismic =>  Screening  of further away seismic by co-located events   Pr { Facies | Co-located soft datum } => Largely  neglecting the local spatial pattern  of seismic Possible solution : use co-located window       Pr { Facies | Co-located  WINDOW  of soft data }
Background image of page 6
    Obvious example Channels East North 0.0 100.000 0.0 130.000 Seismic Impedance East 0.0 100.000 0.0 130.000 7000.00 8000.00 9000.00 10000.00 11000.00 12000.00 Frequency t 5000. 7000. 9000. 11000. 13000. 0.000 0.040 0.080 0.120 0.160 Clustered Data Number of Data 10857 mean 9177.33 std. dev. 1065.61 coef. of var 0.12 maximum 12994.30 upper quartile 9936.33 median 9368.14 lower quartile 8606.53 minimum 5425.39 t 5000. 7000. 9000. 11000. 13000.
Background image of page 7

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Image of page 8
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 01/24/2011 for the course ERE 284 taught by Professor . during the Spring '10 term at Stanford.

Page1 / 43

Pattern recognition methods - Patternrecognitionmethodsin

This preview shows document pages 1 - 8. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online