Thomas A. Smith

• The University of Houston College of Natural Sciences and Mathematics, Distinguished Alumni Award, 2017.[1]
• Houston Geological Society, Geophysicist who has made important contributions to the field of Geology, 2016
• Iowa State University Distinguished Alumni Award, 2015
• Iowa State University Citation of Merit for National and International Recognition, 2002
• Geophysical Society of Houston Honorary Membership, 2010
• Society of Exploration Geophysicists Enterprise Award, 2000,[2]
• Iowa State University Distinguished Alumnus Lecturer Award, 1996

Geophysical Insights was founded in 2008 in Houston, Texas, by Dr. Thomas A. Smith with the vision of applying machine learning to seismic interpretation to reduce the risk of oil and gas exploration and the cost of field development. Shortly after launching the company, Dr. Smith assembled a team of geoscientists and interpretation software specialists to develop the next generation of seismic interpretation software. The group’s mission was to apply machine learning to the problem of seismic interpretation and deliver the new capability in an off-the-shelf, commercial software product that could be used by any geoscientist. The new AI workbench, branded as Paradise®[3], would feature an intuitive interface that guided geoscientists in applying machine learning to many different types of geologic and stratigraphic investigations. Over three years in development, including a year of field trials, Paradise was launched as a commercial product at the Society of Exploration Geophysicists (SEG) Annual Convention in 2013 and has since seen steady adoption worldwide, including by national oil companies (NOCs) and international oil companies.

Geophysical Insights develops and markets the Paradise AI workbench for seismic interpretation, which applies machine learning technologies on seismic and well data to reveal more significant insights in the data versus traditional seismic interpretation tools.   Paradise uses combinations of supervised and unsupervised machine learning algorithms that are enabled by GPU computing. Applications in the current (version 3.3) release of Paradise include seismic Attribute Generation, Attribute Selection, Multi-Attribute Classification, automatic Fault Detection, seismic Facies Classification, and Machine Learning Geobodies.

Seismic interpreters apply the different tools in Paradise either singularly or in combinations according to the geologic or stratigraphic investigation performed. Built for large datasets, the software scales from a single workstation to an enterprise.  The typical deployment is a client-server configuration where multiple geoscientists collaborate on a shared set of seismic surveys and well data.

Paradise is the only commercial software product that incorporates the spectral decomposition and geometric seismic attributes from the AASPI Consortium at the University of Oklahoma.  Including single-trace attributes, Paradise has a library of over 100 seismic attributes that can be generated in the software and used by machine learning algorithms.  The library of spectral decomposition and geometric attributes are being migrated to GPU processing in Paradise, enabling these compute-intensive attributes to generate at 50X speed compared to multi-core CPU computing.  

• Smith, T., Marfurt, K., "Machine Learning Revolutionizing Seismic Interpretation" The American Oil & Gas Reporter (1 July 2017)
• Smith, T., "Geobody Interpretation Through Multi-Attribute Surveys, Natural Clusters and Machine Learning" (5 June 2017)
• Smith, T., Roden, R., Santogrossi, P., Sacrey, D., Jones, G. "Seismic interpretation below tuning with multiattribute analysis" The Leading Edge, Vol. 36 No. 4 (April 2017); p. 330-339.
• Roden, R., Smith, T., and Sacrey, D., 2015,"Geologic pattern recognition from seismic attributes: Principal component analysis and self-organizing maps." Interpretation, Vol. 3, No. 4 (November 2015); p. SAE59-SAE83. doi: 10.1190/INT-2015-0037.1.
• Smith, T., Treitel, S., "Self-Organizing Neural Nets for Automatic Anomaly Identification" (5 November 2015)
• Smith, T. "Delighting in Geophysics" GeoExPro Magazine (September 2014) Volume 11, No. 4
• Smith, T., Sacrey, D. "Seismic Attribute Analysis Can Benefit From Unsupervised Neural Network" Offshore Magazine (September 2011)
• Smith, T., Treitel, S., "Introduction to Self-Organizing Maps in Multi-Attribute Seismic Data" Geophysical Society of Houston (25 January 2011)
• Smith, T., "Unsupervised Neural Networks – Disruptive Technology for Seismic Interpretation" Oil & Gas Journal (7 October 2010)
• Smith, T., Neidell, N. "Improved seismic resolution of stratigraphically complex reservoirs through modeling and color displays." SEG Technical Program Expanded Abstracts (1990); p. 357. doi: 10.1190/1.1890196.
• Smith, T., Sendlein, L. "An Algorithm for the Best Fit Solution of a System of Linear Regression Equations for Seismic Refraction Data." GEOPHYSICS, Vol. 38, No. 6 (December 1973); p. 1062-1069. doi: 10.1190/1.1440396.