Geophysical Insights hosting the 2018 Oil & Gas Machine Learning Symposium in Houston on September 27, 2018
Introduction to Machine Learning for Multi–Attribute Interpretation and AASPI attributes - A 1-day, DGS Continuing Education course in Denver, CO on September 18th
Dr. Tom Smith presenting on Machine Learning at the 3D Seismic Symposium on March 6th in Denver
What is the "holy grail" of Machine Learning in seismic interpretation? by Dr. Tom Smith, GSH Luncheon 2018
Using Attributes to Interpret the Environment of Deposition - A Video Course. Taught by Kurt Marfurt, Rocky Roden, and ChingWen Chen
Workflow Page Banner V3.png
Workflow Page Banner V3.png

Seismic Interpretation l Machine Learning for Attribute Analysis


 

 

 

SCROLL DOWN

Seismic Interpretation l Machine Learning for Attribute Analysis


 

 

 

Guided Workflows: A Step-by-Step Process

Scroll through the steps below to gain a sense of how straightforward Paradise is to use, starting from screens that show how data is loaded in Paradise, and proceeding through the analysis and interpretation process.  While Paradise is a class of learning machines, it has been designed to be used by any interpreter who seeks to obtain greater information from the seismic response.  

Click here to start the tour

Loading SEGY data for Seismic Interpretation


Loading SEGY data for Seismic Interpretation


Load Data & Generate Attributes

 

 
  

Use the new Attribute Generator in Paradise to calculate attributes within Paradise for convenient loading. Leveraging the world-class attributes routines from the AASPI Consortium at the University of Oklahoma, the Attribute Generator produces over 80 attributes types, including amplitude, geometric, and spectral decomposition. Attributes are straightforward to generate and apply using guided thought-flows in Paradise.

Paradise has been designed to make data import efficient and straightforward. Once a few parameters are set for the seismic and attribute volumes within a survey, multiple SEGY files can be loaded with a single command. As the loading proceeds, Paradise creates bricking files called PDM that are designed for fastest processing efficiency. This is just one example of how data management in Paradise has been optimized to work with all interpretation software packages.  

 

Principal Component Analysis for Analyzing Seismic Attributes


Principal Component Analysis for Analyzing Seismic Attributes


Create a PCA

 

Principal Component Analysis (PCA) is a linear process that helps to determine those attributes that contribute most of the information to the study area.  Use PCA on a superset of attributes to quantify the relative contribution of each attribute based on its variance. 

The attributes within a region analyzed using PCA can be set according to inline, crossline, time range, between horizons, or above and below a horizon.   Constrained around production zones, for instance, PCA can be used to obtain a better understanding of the attributes that highlight geobodies.   Apply PCA before running a Self-Organizing Map (SOM) to distill those attributes that are most meaningful, then run a SOM on those attributes to highlight geobodies.  

 
 

 

Principal Component Analysis for Seismic Interpretation


Principal Component Analysis for Seismic Interpretation


View Results of a PCA

 

PCA produces an Eigenvalue per inline across the user-defined area in the form of a bar chart (top) and a set of Eigenvectors (bottom left) that indicate the direction and magnitude of greatest variance among the set of attributes.  Each Eigenvalue reflects the extent of variance of the set of attributes in the PCA and can be selected at any inline to reveal its corresponding set of Eigenvectors.  The chart at the top can be used to understand how the variance among the selected set of attributes changes across the survey. 

Select each Eigenvector to expose the relative contribution of each attribute for that Eigenvector, as shown in the chart on the lower right.  In this way, use the PCA process to quantify those attributes that are most significant to the region prior to running a SOM.  Often, the first and second Eigenvectors are most significant to the geology.  

 

 
  

 

Self-Organizing Maps for Seismic Interpretation


Self-Organizing Maps for Seismic Interpretation


 

Create a SOM

 
 

The SOM process is a powerful, non-linear data classification process that is sensitive to subtle changes in the volume.   Using the straightforward, left-to-right guided workflows, run multiple SOM’s with different configurations to analyze the region in detail. 

A good practice is to run a SOM having a 5X5, 8X8 and 10X10 neural map to see how classification results change at different resolutions, i.e., 25, 64, and 100 classes of data respectively.  Since Paradise is both a batching and interactive technology, multiple SOM’s can be queued and run while the interpreter is away.   Use the Task Manager with Paradise to set job priorities and the number of processors to be dedicated to Paradise. 

 

Interpret Geobodies


Geobodies is a dramatic new machine learning tool within Paradise that enables geoscientists to isolate and investigate areas of interest in the SOM results. Using Paradise to identify geobodies empowers interpreters with the ability to:

Investigate geobodies at the sample level of each neuron

Capture details on areas of interest, including volumetrics and statistics

Edit/clean-up selected geobodies by filling in areas or pruning extraneous samples

Export areas of interest to an interpretation system

Interpret Geobodies


Geobodies is a dramatic new machine learning tool within Paradise that enables geoscientists to isolate and investigate areas of interest in the SOM results. Using Paradise to identify geobodies empowers interpreters with the ability to:

Investigate geobodies at the sample level of each neuron

Capture details on areas of interest, including volumetrics and statistics

Edit/clean-up selected geobodies by filling in areas or pruning extraneous samples

Export areas of interest to an interpretation system

 

Interpret Geobodies

 

Geobodies is a dramatic new machine learning tool within Paradise that enables geoscientists to isolate and investigate areas of interest in the SOM results. Using Paradise to identify geobodies empowers interpreters with the ability to:

  • Investigate geobodies at the sample level of each neuron
  • Capture details on areas of interest, including volumetrics and statistics
  • Edit/clean-up selected geobodies by filling in areas or pruning extraneous samples
  • Export areas of interest to an interpretation system
 

Free Trial of Seismic Interpretation Software


Free Trial of Seismic Interpretation Software


 

Paradise Evaluation

Contact us to explore how the machine learning workflows in Paradise can reduce exploration risk through an evaluation.