A Rich Set of Data Analytic and AI Tools for Geoscience
Paradise uses robust, unsupervised machine learning and supervised deep learning technology to accelerate interpretation and enable geoscientists to discover more from the data. Apply the guided ThoughtFlows™ in the Paradise AI workbench to…
- Identify thin beds
- See detailed stratigraphy/facies
- Analyze reservoir properties
- Reveal DHIs
- Detect faults and fracture trends
- Isolate pressure-related facies
Generate attributes to extract meaningful geological information and is input to machine learning analysis
Highlight geologic features in 3D by blending up to three attribute volumes
Identify attributes having the greatest contribution among a set of attributes in a geologic region
Classify multi-attribute data to map the distribution of seismic facies that correlate with geology using the SOM machine learning process
Analyze the classified volume to corroborate the presence of stratigraphic facies in 3D with or without well control
Calculate volumetrics and estimate recoverable hydrocarbons using machine learning geobodies
Capture facies based on distinctive seismic patterns using deep learning technology
Produce fault labels and probability volumes automatically with the option to pick examples using deep learning technology
The Paradise Attribute Generator places best-in-class post-stack seismic attribute calculations in the hands of general interpreters and specialists alike using easy-to-follow ThoughtFlows (TM), dependency detection, and geoscience-focused visualization capabilities. The Paradise Attribute Library includes algorithms and workflows backed by the highest standards in academic research by the Attribute Assisted Seismic Processing and Interpretation (AASPI) group, a first-rate research team from the University of Oklahoma led by Dr. Kurt Marfurt.
Families of Attributes
• Spectral Decomposition
Attributes differ in their relative contribution to information (energy) in a given volume. In Paradise, Principal Component Analysis (PCA) is employed to identify those attributes that have the greatest contribution to the data and quantify the relative contribution of each attribute to a volume. Using an Eigen spectrum chart, the relative contribution is easily seen both graphically and numerically, taking the guesswork out of selecting the right attributes for the region.
Multi-attribute Classification is accomplished using the Self-Organizing Map (SOM), a form of unsupervised machine learning. Applied in Paradise, the SOM improves the resolution of stratigraphic facies and their distributions. This methodology has proven to define potential fluid effects on seismic data given the appropriate seismic attributes. The technology also has demonstrated the ability to detect thin beds that are below classic seismic tuning thickness. The multi-attribute classification results are corroborated with existing well control, and volumetrics for selected geobodies can be calculated from designated classes that result from the SOM using the Geobody Detection application.
A SOM analysis with a corresponding 8x8 neural hexagonal 2D color bar that are interactive with the Universal Viewer
Interactive 2D Colormap
Paradise is the only software product in the industry with a 2D Colormap representing neuron classifications of attributes. The 2D Colormap is used interactively with the Paradise Universal Viewer to select and isolate specific neurons which have classified a set of seismic attributes according to where data is concentrated. The Paradise 2D Colormap focuses on specific anomalies, discovering geologic features, stratigraphic features, high pore pressure, or direct hydrocarbons indicators (DHI’s).
The blending of attributes in an RGB display can greatly enhance the visibility of geologic features within the seismic data. The RGB Blending application is a 3D visualization tool that combines three separate attribute volumes into a single volume for interpretation. When applied to seismic data RGB blending or co-rendering can highlight geologic and geophysical features, thus improving interpretability.
Machine learning geobodies are isolated at the bin level, enabling volumetric estimates of hydrocarbons
Machine Learning Geobodies
Geobody generation and editing are unique to the Paradise AI workbench. Geobodies are created from neuron groups selected from a Self-Organizing Map (SOM), and are used to generate a three-dimensional visualization of geological features, such as channels, sand bars, deltaic characteristics, or any other area of interest. The number of Geobodies generated depend upon the SOM topology and the number of neurons originally selected.
Once generated, industry-standard volumetric calculations are available in the tool to estimate the amount of in-place and/or recoverable oil or gas. 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
Deep Learning (DL) Fault Detection
The DL Fault Detection application in Paradise 3.3 is equipped with general pre-trained deep learning models, enabling their application to a wide range of seismic data without the need of user-provided fault examples for training. Using Graphic Processing Unit (GPU) technology, the tool dramatically reduces the time to identify faults in a volume.
Fault detection result on a seismic survey from the Great South Basin, offshore New Zealand. a) Seismic amplitude data on which fault detection is performed. b) Fault probability from the CNN-based fault detection. c) A state-of-the-art seismic coherence attribute. Note the greatly improved fault continuity, resolution, and reduced noise from the CNN-based fault detection method.
Deep Learning (DL) Seismic Facies Classification
The new DL Seismic Facies classification tool in Paradise 3.3 enables the development of a structural and stratigraphic seismic interpretation based on deep learning technology. Seismic facies and any other patterns present on seismic data such as Direct Hydrocarbons Indicators, multiples, etc., can be identified in a seismic volume given the appropriate training. The 3D extent of these features can provide significant and valuable information to the interpretation process.
High Performance Computing (HPC)
With the increase in the size of data volumes, traditional servers and workstations can limit the speed of interpretation. The new Paradise HPC capability enables the generation of geometric and spectral decomposition attributes on clusters of machines, dramatically reducing computation time.
The Petrel connector allows Paradise users to import and export data from Paradise into and out of Petrel.
The Petrel Connector to Paradise is installed on the same machine with Petrel and appears as an icon on the Petrel ribbon. Simply open the Paradise Connector, select the data to be moved to Paradise, drag and drop the data items onto the Petrel Connector dialog, and select Export. Import from Paradise to Petrel is equally straightforward. Multiple items can be exported at once by selecting the folder to export, such as all wells. The products of machine learning and deep learning in Paradise, such as Deep Learning Seismic Facies, Faults, and Multi-Attribute Classification volumes, can be imported easily into Petrel for further analysis.
*Petrel is a registered trademark of Schlumberger
Below is a list of the minimum technical hardware and software requirements for running a single instance of the Paradise client and server on a dedicated workstation.
- Windows 10 Pro
- MS SQL Server 2012
- 8+ processing cores
- 32GB RAM
- 4GB NVIDIA graphics card
- File storage capacity depends on several different variables: See note below*
- 10Gb ethernet connection for networked data storage or USB 3.0 connection for external storage devices
Contact us for more information regarding technical specifications on client/server installations.
* File Storage capacity is dependent on the size of the surveys and attributes volumes, plus interpretation files, e.g., faults, horizons, wells, etc. Local storage (on the server running Paradise) is encouraged for evaluation purposes; however, if a centralized network storage facility is to be used for project files, a 10 Gb LAN connection is recommended between the target server and project file storage facility.