Reveal is a periodic newsletter that aims to highlight recent changes and additions to Reveal. In this issue we discuss new tools added, a few updates to tools and we introduce three new team members.
VelocityMLPicker is a guided velocity auto-picker that uses a modified semblance-based approach in conjunction with machine learning to create new velocity picks. Figure 1 shows two CMP shot gathers with NMO applied with a) hand-picked velocities and b) machine learning auto-picked velocities. The auto picked velocity used a single guide function picked at a CMP 10 km away while the hand-picked velocities were picked every km. The two shot gathers are of similar quality, displaying the solid performance of the auto-picker.
Figure 1: CMP gathers with NMO applied for a) left, hand-picked velocities and b) right, machine learning auto-picked velocities.
New tool that creates a WEIGHT which is stored in a DB-file. The WEIGHT is computed using image dilation and smoothing algorithms. It can be merged on traces as a header and used to spatially blend and transition between traces from two surveys, even when the surveys have irregular shapes.
Figure 2: Crossplot view of DB created by MergeWeights. The map is colored by the WEIGHT column. The weights smoothly transition between surveys, also for irregularly shaped surveys.
Shapefiles (.shp) are now supported in Reveal. The shapefile format is a vector data format for storing geometric location and associated attribute information. These files are viewable in crossplot complete with drag-and-drop functionality.
Figure 3: Shapefile display in Reveal
Importing Project Grids
A new option has been added to import a project grid from an existing project. This allows users to simply copy over an existing grid rather than having to re-enter the grid parameters.
The GatherFiltering tool has been ported from Python to C++: it now uses multiple cores and runs significantly faster.
DBSetup, DBSetupMPI, and DB2Trace
DBSetup had a bug that caused failures when smoothing interpolated floating datums. This has now been fixed. We have also added a new DBSetupMPI tool which allows DBSetup to parallelize over multiple MPI ranks. This is useful both for reducing memory usage on very large surveys and allows flows containing DBSetupMPI to be sped up by splitting over more nodes. The DBSetupMPI logic will be moved into DBSetup within the next month or two.
Figure 4. DBSetupMPI with 10 ranks running at 2280 MB/sec.
DB2Trace now has support for MPI, so flows processing db files using DB2Trace and DBWrite can now run on multiple nodes.
Since late 2016, we have made a number of performance improvements to PreStackDepth. In December, we modified the migration logic to avoid wasted work outside of the aperture ellipse. This should give up to a 20% speed boost in some cases. More recently, we have optimized memory-bound portions of the code to provide a speedup of 2.5% with hyperthreading on or 6.5% with hyperthreading off.
Welcome to the team
Senior Software Programmer
Eric Geordi has over 15 years of experience in software development in a range of industries, spanning aerospace, finance and oil/gas. He holds a bachelors in computer science from The University of Texas in Austin and a masters degree in software engineering from The University of Houston.
The bulk of his experience includes working with geoscientists to implement seismic processing tools for the purpose of manipulating subsurface velocity models and he looks forward to putting this experience to use at Shearwater Reveal.
Senior Software Architect
John Pixton has over 25 years of experience developing software for biomedical, astronomical, and oil/gas sectors. He holds a bachelor's degree in computer & electrical engineering and a masters degree in computer science, both from the University of Illinois at Urbana-Champaign.
Since leaving the university, he has been developing software for exploration and production, focusing primarily on developing high-performance interactive and processing tools. Most recent work focused on developing architecture, technologies and tools for velocity model building and well qc workflows. He looks forward to applying his experience as well as continued growth at Reveal.
Software Support - Geophysicist
Cliff Mack has nearly 4 years of experience working as a Processing Geophysicist for both Dolphin Geophysical and Shearwater. He received his bachelor’s degree in Geology from Texas A&M University.
The majority of his young career has taken him offshore - working first for the Integrated Ocean Drilling Program (2010-2014), then moving on to Dolphin Geophysical as an offshore processing geophysicist in 2014. For the last 18 months he has been at work on the processing team in Houston as well as adding support to Reveal. He is excited to make use of his processing experience in Reveal when providing support to clients.
Reveal US User Group meeting
The annual Reveal US user group meeting took place in Houston on April 20th, the biggest to date. It featured team members discussing tools that have been developed over the last year along with tools that will be added in the near term.