information entropy monte carlo simulation · 2009-04-09 · monte carlo simulation ayato kato...
TRANSCRIPT
![Page 1: Information Entropy Monte Carlo Simulation · 2009-04-09 · Monte Carlo Simulation Ayato Kato March, 27th, 2009. Rock Physics 2 Outline Shannon’s Information Entropy Monte Carlo](https://reader034.vdocuments.us/reader034/viewer/2022042808/5f86bf4814f71d0ec8660d2a/html5/thumbnails/1.jpg)
Rock Physics
Information EntropyInformation EntropyMonte Monte Carlo Carlo SimulationSimulation
Ayato KatoAyato KatoMarch, 27th, 2009
![Page 2: Information Entropy Monte Carlo Simulation · 2009-04-09 · Monte Carlo Simulation Ayato Kato March, 27th, 2009. Rock Physics 2 Outline Shannon’s Information Entropy Monte Carlo](https://reader034.vdocuments.us/reader034/viewer/2022042808/5f86bf4814f71d0ec8660d2a/html5/thumbnails/2.jpg)
22Rock Physics
Outline
Shannon’s Information Entropy
Monte Carlo Simulation
![Page 3: Information Entropy Monte Carlo Simulation · 2009-04-09 · Monte Carlo Simulation Ayato Kato March, 27th, 2009. Rock Physics 2 Outline Shannon’s Information Entropy Monte Carlo](https://reader034.vdocuments.us/reader034/viewer/2022042808/5f86bf4814f71d0ec8660d2a/html5/thumbnails/3.jpg)
33Rock Physics
Which attribute(s) should We Use ?
Type SeismicAttribute
Major Geological significance
Amplitude Lithological contrast, beddingcontinuity
Interval velocity Lithology, Porosity, Fluid Content
Conventionalproperty
Acousticimpedance
Lithology, Porosity, Fluid Content
Reflectiongeometry
Reservoir Architecture,Sedimentary Structure
Trace continuity Fault geometry, Fault distribution,Stratigraphic continuity
Volume-relatedattribute(Multi-trace attribute)
Time curvature,Dip, Azimuth
Detailed Reservoir Architecture,Fault geometry, Fault distribution,Fracture density
AVOImpedance(elastic/S-wave)Poisson's ratioλ, μ (Lame constant)
Fluid Content, Lithology, PorosityPre-stack attribute
AVOZ Fracture Orientation, Fracture Density,Fluid Content
Instantaneousphase
Bedding continuityInstantaneousattribute
InstantaneousFrequency
Bed thickness, lithologic contrast, fluidcontent
FrequencyAttenuation
Fluid contentMiscellaneousattribute
Anythingcomputed fromseismic traces
???
Innumerable Seismic AttributesFor the purposes of predicting porosity, which attributes should we use?
- AVO P0 & G- AI & EI- λρ & μρ- Vp/Vs etc
Shannon’s Information Entropy can give us the solution quantitatively.
![Page 4: Information Entropy Monte Carlo Simulation · 2009-04-09 · Monte Carlo Simulation Ayato Kato March, 27th, 2009. Rock Physics 2 Outline Shannon’s Information Entropy Monte Carlo](https://reader034.vdocuments.us/reader034/viewer/2022042808/5f86bf4814f71d0ec8660d2a/html5/thumbnails/4.jpg)
44Rock Physics
Shannon’s Information Theory
Dr. Claude Shannon
Shannon and Weaver (1949)“The Mathematical Theory of Communication”
Defined Quantity of Information
Information Content
Information Entropy
Mutual Information
![Page 5: Information Entropy Monte Carlo Simulation · 2009-04-09 · Monte Carlo Simulation Ayato Kato March, 27th, 2009. Rock Physics 2 Outline Shannon’s Information Entropy Monte Carlo](https://reader034.vdocuments.us/reader034/viewer/2022042808/5f86bf4814f71d0ec8660d2a/html5/thumbnails/5.jpg)
55Rock Physics
It will snow !!
Very surprised !!
Not surprised
(valuable information)
(Not-valuable information)
Information Content
PI log−=Information Content (Quantity of information)
Isnow = 6.64
* Base = 2
Isnow = 0.51
How surprised one would be if the event happened.
HoustonVery rare (1% )
CalgaryVery often (70%)
Low Probability
High Probability
![Page 6: Information Entropy Monte Carlo Simulation · 2009-04-09 · Monte Carlo Simulation Ayato Kato March, 27th, 2009. Rock Physics 2 Outline Shannon’s Information Entropy Monte Carlo](https://reader034.vdocuments.us/reader034/viewer/2022042808/5f86bf4814f71d0ec8660d2a/html5/thumbnails/6.jpg)
66Rock Physics
Information Entropy
Expected Value of Information Content
( ) ∑ ⋅−=n
iii PPXH log
Houston
Calgary
1Snow
33Rain
33Cloudy
33Sunny
Pi (%)Xi
70Snow
10Rain
10Cloudy
10Sunny
Pi (%)Xi NY
25Snow
25Rain
25Cloudy
25Sunny
Pi (%)Xi
H=1.65
H=1.36 H=2.00
• Expected Surprise• Quantity of uncertainty associated with P
Forecast is easyUncertainty is small
Forecast is difficultUncertainty is large
Information Entropy { }NxxX ,,1 ⋅⋅⋅⋅⋅⋅=
![Page 7: Information Entropy Monte Carlo Simulation · 2009-04-09 · Monte Carlo Simulation Ayato Kato March, 27th, 2009. Rock Physics 2 Outline Shannon’s Information Entropy Monte Carlo](https://reader034.vdocuments.us/reader034/viewer/2022042808/5f86bf4814f71d0ec8660d2a/html5/thumbnails/7.jpg)
77Rock Physics
Conditional Entropy
( )AXH |Conditional Information Entropy
Conditional Information Entropy at a given additional information (A)
NY
25Snow
25Rain
25Cloudy
25Sunny
Pi (%)Weather
H(X)=2.00Uncertainty is large
5Snow
10Rain
25Cloudy
50Sunny
Pi (%)Weather
H(X|A)=1.54Uncertainties reduce
At given additional information A
( ) ( ) ( )AXHXHAXI || −= Quantity of uncertainty reduced by additional information A
Mutual Information
e.g. A = atmospheric pressure change data
![Page 8: Information Entropy Monte Carlo Simulation · 2009-04-09 · Monte Carlo Simulation Ayato Kato March, 27th, 2009. Rock Physics 2 Outline Shannon’s Information Entropy Monte Carlo](https://reader034.vdocuments.us/reader034/viewer/2022042808/5f86bf4814f71d0ec8660d2a/html5/thumbnails/8.jpg)
88Rock Physics
Our Case : Porosity Prediction
X : Porosity (continuous variable)
H(X|A) : Information Entropy for Porosity’s PDF at given seismic attributeH(X) : Information Entropy for Porosity’s PDF
PDFsAdding more seismic attributes
PDF shape → Narrow, steepUncertainty → Decrease In. Entropy → Decrease
H(φ)=3.44
H(φ|Vp)=3.06
H(φ|Vp,Vs)=2.87
Mukerji et al. (2001)
A : Seismic Attribute
Porosity Prediction by Seismic Attribute
![Page 9: Information Entropy Monte Carlo Simulation · 2009-04-09 · Monte Carlo Simulation Ayato Kato March, 27th, 2009. Rock Physics 2 Outline Shannon’s Information Entropy Monte Carlo](https://reader034.vdocuments.us/reader034/viewer/2022042808/5f86bf4814f71d0ec8660d2a/html5/thumbnails/9.jpg)
99Rock Physics
Information carried by Seismic Attributes
( ) ( ) ( )AXHXHAXI || −=Mutual Information
Conditional entropy at given A
Mutual Information can be regarded as the reduced uncertainty by the seismic attributes. Thus, we should choose the one which will maximize the mutual information.
Information entropy
For example, X = PorosityA = Seismic Attribute
![Page 10: Information Entropy Monte Carlo Simulation · 2009-04-09 · Monte Carlo Simulation Ayato Kato March, 27th, 2009. Rock Physics 2 Outline Shannon’s Information Entropy Monte Carlo](https://reader034.vdocuments.us/reader034/viewer/2022042808/5f86bf4814f71d0ec8660d2a/html5/thumbnails/10.jpg)
1010Rock Physics
Case Studies
( )attributesfaciesH |
North Sea Tertiary Turbiditic Reservoir
Case I- Facies Identification
Case II- Pore Fluid Identification
References- Tapan et al. (2001)- Takahashi et al (1999)
( )attributesfluidH |
![Page 11: Information Entropy Monte Carlo Simulation · 2009-04-09 · Monte Carlo Simulation Ayato Kato March, 27th, 2009. Rock Physics 2 Outline Shannon’s Information Entropy Monte Carlo](https://reader034.vdocuments.us/reader034/viewer/2022042808/5f86bf4814f71d0ec8660d2a/html5/thumbnails/11.jpg)
1111Rock Physics
Well Log Data
Facies Classification in Well Log Data
IIa: Cemented SandIIb: Uncololidated SandIIc: Laminated SandIII: Interbedded Sand-ShaleIV: Silty ShaleV: Pure Shale
Takahashi et al. (1999)
![Page 12: Information Entropy Monte Carlo Simulation · 2009-04-09 · Monte Carlo Simulation Ayato Kato March, 27th, 2009. Rock Physics 2 Outline Shannon’s Information Entropy Monte Carlo](https://reader034.vdocuments.us/reader034/viewer/2022042808/5f86bf4814f71d0ec8660d2a/html5/thumbnails/12.jpg)
1212Rock Physics
CrossplotIIa: Cemented SandIIb: Uncololidated SandIIc: Laminated SandIII: Interbedded Sand-ShaleIV: Silty ShaleV: Pure Shale
Takahashi et al. (1999)
![Page 13: Information Entropy Monte Carlo Simulation · 2009-04-09 · Monte Carlo Simulation Ayato Kato March, 27th, 2009. Rock Physics 2 Outline Shannon’s Information Entropy Monte Carlo](https://reader034.vdocuments.us/reader034/viewer/2022042808/5f86bf4814f71d0ec8660d2a/html5/thumbnails/13.jpg)
1313Rock Physics
Conditional probability distributions
Takahashi et al. (1999)
Univariate
Bivariate
![Page 14: Information Entropy Monte Carlo Simulation · 2009-04-09 · Monte Carlo Simulation Ayato Kato March, 27th, 2009. Rock Physics 2 Outline Shannon’s Information Entropy Monte Carlo](https://reader034.vdocuments.us/reader034/viewer/2022042808/5f86bf4814f71d0ec8660d2a/html5/thumbnails/14.jpg)
1414Rock Physics
Mutual Information
Takahashi et al. (1999)
Information about lithofacies carried by Seismic attributes
![Page 15: Information Entropy Monte Carlo Simulation · 2009-04-09 · Monte Carlo Simulation Ayato Kato March, 27th, 2009. Rock Physics 2 Outline Shannon’s Information Entropy Monte Carlo](https://reader034.vdocuments.us/reader034/viewer/2022042808/5f86bf4814f71d0ec8660d2a/html5/thumbnails/15.jpg)
1515Rock Physics
Case Study 2 (Pore Fluid)
Extended PDFs
Takahashi et al. (1999)
Univariate
Bivariate
Gassmann eq. was applied for fluid substitution in only sand reservoir
Vp
![Page 16: Information Entropy Monte Carlo Simulation · 2009-04-09 · Monte Carlo Simulation Ayato Kato March, 27th, 2009. Rock Physics 2 Outline Shannon’s Information Entropy Monte Carlo](https://reader034.vdocuments.us/reader034/viewer/2022042808/5f86bf4814f71d0ec8660d2a/html5/thumbnails/16.jpg)
1616Rock Physics
Mutual Information
Takahashi et al. (1999)
Information about pore fluid carried by Seismic attributes
![Page 17: Information Entropy Monte Carlo Simulation · 2009-04-09 · Monte Carlo Simulation Ayato Kato March, 27th, 2009. Rock Physics 2 Outline Shannon’s Information Entropy Monte Carlo](https://reader034.vdocuments.us/reader034/viewer/2022042808/5f86bf4814f71d0ec8660d2a/html5/thumbnails/17.jpg)
1717Rock Physics
Discussions
Variance (covariance) can work only at limited situation- Parametric PDFs, such as Gaussian distribution- Continuous variable
Linear measures of uncertainty, such as variance (σ2) and covariance (σ12), can be used instead of the entropy (H) ?
Information Entropy can work more flexibly- Nonparametric PDFs- Categorical variables (Shale, Sand)
The Entropy offers a more flexible representation of the state of information about the rock.
![Page 18: Information Entropy Monte Carlo Simulation · 2009-04-09 · Monte Carlo Simulation Ayato Kato March, 27th, 2009. Rock Physics 2 Outline Shannon’s Information Entropy Monte Carlo](https://reader034.vdocuments.us/reader034/viewer/2022042808/5f86bf4814f71d0ec8660d2a/html5/thumbnails/18.jpg)
1818Rock Physics
Monte Carlo Simulation
A technique using random numbers for probabilistic solution of a model
http://www.vertex42.com/ExcelArticles/mc/MonteCarloSimulation.html
• Model is nonlinear system• Input parameters with uncertainty• Uncertainty analysis instead of
deterministic method
![Page 19: Information Entropy Monte Carlo Simulation · 2009-04-09 · Monte Carlo Simulation Ayato Kato March, 27th, 2009. Rock Physics 2 Outline Shannon’s Information Entropy Monte Carlo](https://reader034.vdocuments.us/reader034/viewer/2022042808/5f86bf4814f71d0ec8660d2a/html5/thumbnails/19.jpg)
1919Rock Physics
Monte Carlo Simulation
Step 1: Create a model, y = f(x1, x2, ..., xq).
Step 2: Generate a set of random inputs, xi1, xi2, ..., xiq.
Step 3: Use the model to obtain outputs.
Step 4: Repeat steps 2 and 3 for i = 1 to n.
Step 5: Analyze the results using histograms, summary statistics, confidence intervals, etc.
LognormalNormalConstant etc
Data-deriven nonparametric PDF
http://www.vertex42.com/ExcelArticles/mc/MonteCarloSimulation.html
![Page 20: Information Entropy Monte Carlo Simulation · 2009-04-09 · Monte Carlo Simulation Ayato Kato March, 27th, 2009. Rock Physics 2 Outline Shannon’s Information Entropy Monte Carlo](https://reader034.vdocuments.us/reader034/viewer/2022042808/5f86bf4814f71d0ec8660d2a/html5/thumbnails/20.jpg)
2020Rock Physics
Thank you for attentions