Skip to main content
Share this page
Share on Facebook
Share on Twitter
Share on LinkedIn
May 16th-West Tech Lunch: Wolfspar an "FWI-friendly" Ultra-Low-Frequency Marine Seismic Source*
Norris Conference Center
816 Town & Country Blvd.
Houston, TX 77024
(Free parking off Beltway-8 northbound feeder or Town & Country Blvd.)
Meeting Time: 11:00 to 1:00 pm
Registration Begins at 11:00
Lunch Served at 11:30
Presentation starts at Noon
You Must Be Logged In to Register.
Speaker: Joe Dellinger, BP America
Co-Authors: Allan Ross, David Meaux, Andrew Brenders, Glenn Gesoff, John T. Etgen, and John Naranjo, BP America
Graham Openshaw, TecPM and Mark Harper, Cambridge Applied Physics
Over the past decade, BP designed, built, and field tested Wolfspar®, a full-scale, ultra-low-frequency seismic source capable of producing frequencies of 1.4 – 8 Hz. This was quite an ambitious project to undertake, and was the culmination of years of thinking differently about the challenge of seismic imaging in deep water under complex salt.
First, we realized that our wide-azimuth seismic data have become good enough that they are no longer the primary impediment to imaging under complex salt. The limiting factor is now our velocity models. Our “Garden Banks” model study showed that in areas of complex salt, our velocity models likely include “interpretation busts”, where cubic kilometers of the velocity model may have salt where there should be sediment, or vice versa. These mistakes are very hard to identify and fix using any form of manual interpretation --- if that strategy worked, the problem would have been solved by now. Model studies showed that FWI can fix these problems, but requires seismic data with ultra-low frequencies (below 2 Hz) recorded at wide offsets (20-30 km) to do so. The problem then becomes, how can we get the required data at a reasonable cost? How do we design an entirely new acquisition strategy around velocity-model-building using FWI?
Acquiring frequencies below 2 Hz at wide offsets introduces a whole new set of challenges. Marine seismic sources naturally roll off at low frequencies at about 18 dB / Octave, and below 2 Hz the natural microseismic background noise also rapidly increases, resulting in a very steep S/N “wall” to climb. It is very difficult to penetrate this noise wall. We need a source that increases the signal at low frequencies that does not also create undesirable additional high frequencies. Conventional broadband sources are not well suited to the challenge. These considerations drove us to controlled sources that could tailor their output to concentrate their power on just the required frequency range. Conventional marine vibrator designs optimized to produce 10-100 Hz are problematic to scale up to produce lower frequencies, however. To maintain a constant far-field amplitude, the volume of water to displace scales as inverse frequency cubed, so a 1 Hz source would need to displace 1000 times the volume of a 10 Hz source of the same design. The power required to move such a large volume of water using vibrator technologies can become impractically large. Our solution was an energy-efficient, resonating piston design. The engineering challenge of how to seal a 1.6-meter diameter piston moving back and forth over a range of motion of half a meter proved to be solvable. Field-testing the source under tow at 4 knots, recording into ocean-bottom sensors, we achieved an excellent signal-to-noise ratio in the deep water Gulf of Mexico at offsets of over 30 km and at frequencies as low as 1.6 Hz despite the significant ambient noise at these frequencies.
Now that a prototype source exists, the remaining problem is acquisition design. Streamers are noisy at low frequencies and don’t easily accommodate the necessary offsets and azimuths. This drives us to a low-noise nodal acquisition. Conventional node arrays spaced ~400 meters apart recording offsets out to 30 km or more at all azimuths would be quite expensive. The key is to realize that we are designing the survey to record low-frequencies for velocity-model-building, not for conventional imaging purposes. Sampling considerations allow us to consider a cost-efficient sparse 3D acquisition geometry, reminiscent of land acquisition strategies of the late 1990’s, but performed at sea. We believe that by rethinking acquisition from end to end, we can meet the challenge of imaging in areas of deep water and complex salt.
Speaker Biography: Joe Dellinger, BP America
Joe Dellinger received a PhD in 1991 from Jon Claerbout’s Stanford Exploration Project, then did a 3-year post-doc at the University of Hawai`i before joining Amoco’s Tulsa Research Center in 1994. He moved to BP in Houston in 1999 and has worked there since. In his career he has specialized in Anisotropy, multi-component algorithms and processing, and most recently rethinking marine seismic acquisition. As part of that work he has spent considerable time closely examining seismic data trying to understand all the signals present in it. He calls this “Forensic data processing”, which was the title and subject of his 2016 Spring SEG distinguished lecture series (41 talks in 11 countries in 3 months!). Since 2006 Joe has been the lead BP geophysicist supporting BP’s “Wolfspar” project, with the goal to design, build, deploy and commercialize an ultra-low-frequency marine seismic source.
Joe was awarded SEG life membership in 2001 and honorary membership in 2016 for his services in helping the SEG to adapt to the internet age. Joe’s hobbies include attending the Houston Symphony, photographing birds, recording frog calls in the swamps around Houston, and public outreach astronomy at the George Observatory. Asteroid “78392 Dellinger” was named in his honor.
5/16/2017 11:00 AM - 1:00 PM
Central Daylight Time
Create a new account