E-Lecture Webinar: Automated top and base salt interpretation using machine learning

Webinar details
Instructor:   Oddgeir Gramstad
Duration:   20-30 min + 15 min Q&A
Discipline:   Data Science
Main topics:   Top salt interpretation
Language:   English

Request this Webinar   Overview of all scheduled webinars    Read Paper in EarthDoc

Attending webinars and access to recent EarthDoc material is free of charge for EAGE members, join here.


We present a new automated workflow based on machine learning which can significantly reduce the amount of manual interpretation of the top salt boundary. Manual interpretation of top salt on large seismic surveys with complex salt geometry is a time-consuming task. The interpreters typically need to scan through the seismic volume and pick control points line-by-line. It can take more than a month to complete a top salt interpretation. In this new method, a convolutional neural net is designed to detect the top of salt boundaries and the training data are picked as 2D images on a manual top salt interpretation in a specific seismic survey. The trained network is then evaluated both on the seismic data used in the training and on another seismic data not used in the training. In both cases we can produce a top salt interpretation that covers the main parts of the corresponding manual interpretations. The results can be further improved by adding more training data. This new automated workflow has the potential to reduce the interpretation turnaround time of top of salt from approximately a month or more and down to hours.

Participants' Profile

Geoscientists, geologists, geophysicists, salt interpreters and data scientists.

About the Lecturer

Oddgeir Gramstad

Oddgeir Gramstad has a MSc. in cybernetics from the University of Stavanger in 2005 and he is working as a Senior Research Scientist at Schlumberger Stavanger Research (SSR). He has worked with developing different tools for automated interpretation of seismic data and attributes including classification and pattern recognition. The two main technologies are the Seismic DNA and the Extrema interpretation method, which have been tested and applied on different real data sets world-wide together with different oil companies. Oddgeir is currently working on developing new machine learning applications for G&G applications, including automated interpretation of top and base salt.