Innovation Geophysics

Huw James

12th October 2012

My Secondary School

Some creative former pupils Ernest Shackleton Phil Manzanera Raymond Chandler C.S. Forester Michael Ondaatje P.G. Wodehouse Peter Oosterhuis Eddy George

• Analogue Seismic Computing

• Stacking of marine seismic & low energy sources

• Analog to digital transition

• 2D seismic to 3D seismic transition

• Navigation & surveying for 3D seismic

• Interpretation workstations

• Equity

• GPS navigation and positioning

• Multi-boat, multi-source, multi-streamer marine seismic

• 3D graphics workstations

Technology TI makes commercial silicon transistors 1954 TI’s Jack Kilby invents integrated circuit 1958 TI’s Harvey Cragon & Joe Watson build IC computer 1961

Converter

FM Techno AM 28 trace Scope

Camera

Bias Oscillator – Junior Geophysical Engineer

Drum & Amps

Drum & Amps

Drum & Amps

Data & Sync

@$10 per transcription 400x2 copies per 8 hour shift Parallel tape load & unload Sync -> Servo [relied on transistor logic]

FM Techno AM 60x2 copies per 8 hour shift Serial tape load & unload

No Sync needed

Single Axle

Analogue processing

• MAE - multiple analysis and elimination • Normal Move Out correction • Muting • [stacking] • Section Plotting • Base Map generation – navigation & survey processing • Horizontal mis-tie estimation & correction • Horizon [unconformity] picking & posting • Hand contouring • Copying results to paper • Final QC

• Land Noise Analysis – 3D spread • Seismic velocity survey – 2 boats • Check Shot correction • Transcription

• Engineers • Processors • Analogue Operators • Interpreters • Draftsmen

J. Fred Bucy at controls of seisMAC hybrid digital/analog computer

©Texas Instruments

MAE See: water Reverberations – their nature and elimination, Milo M. Backus, Geophysics XXIV, April 1959 – equation (20)

surface

water bottom

reflector

Primaries Multiples

• Compute water depth • Estimate periodicity of multiple by trial subtractions based on water depth • Estimate the RC multipliers • Eliminate multiples by delaying recorded trace by 1x & 2x the multiple period and

adding to the recorded trace after scaling by the rc & r𝑐2, respectively Delay was via electronic delay line We, in Bromley, changed the delay peg-board to an automatic counter; added multiple staggered heads to the source drum; added parallel processing. So: Automation, Pipelining & Parallelization

TIAC 827 digital computer

• Consortium with Mobil & Texaco

• Transistor based

• Installed in GSI Dallas computer center in Fall of 1961

• DFS 9000 – digital recording in 1960

• Office Playback system based on DFS 9000

• True amplitude recording [not AGC]

• Greater dynamic range than Analog

• Unlimited record length

• About 25 systems delivered

• Storage from 1K words to 4K words

• One inch multiplexed tape format

• Asynchronous instruction execution, NOR gate logic

• Filter Box/Convolver 250 nanosec fixed point MPA

Anti-aliasing, Stationarity, Linearity, Minimum phase &

Zero phase Emphasized

©Texas Instruments

©Texas Instruments

TIAC 827 digital computer Software explosion from Fall of 1961-1966

• Wiener Deconvolution, Gapped Deconvolution, Section Multiple Attenuation

• Velocity analysis, Statics & NMO correction

• Trace muting & Editing

• Pie Slice & Butterfly velocity filter

• Minimum & zero phase filtering, matched filtering, correlograms

• Vertical & Horizontal stack

• Markov Power Spectra, Maximum Entropy Spectral Analysis - Burg

• True Amplitude Gain recovery, spherical divergence, exponentiation

• Optimum Wide Band Stack – De-ghosting – all the suspects – Backus, Burg, Larner, Schneider

• Dynamic Correlation Analysis, Multiple Attenuation by differential move out

• Refraction data processing

• Diversity Stack

• Cross Equalization

• Common Tangent Stack – Don Rockwell’ s migration

• 200 Package trace processing system, M4 Marine system & much, much more …

TIAC 827 digital computer 1967-1968

• Various static programs from Blum and others

• Surface Consistent statics from Hileman et al

• 400 Package Trace Processing system – 15 seconds/Shot Point

TIAC 870 & 870A digital computers 1965-1970

• Used 7000 & 7400 series integrated circuits, , NAND gate logic

• Synchronous with 200 nanosecond clock

• CPU Vector Dot Product -> 200 nsec MPA

• Floating Point Arithmetic Unit

• Array Processor from shared memory

• TIADD - CRT film recorder

• Massive Disk

• Basic FG/BG OS & Fortran compiler

• Interactive work stations

©Texas Instruments

©Texas Instruments

TIAC 870 & 870A digital computers 1966-1970

• Apick – section horizon picking

• Adaptive deconvolution – see TI TMS 320 digital signal processor in your mobile phone

• Air Gun trace processing package – 28 seconds per SP

• 300 trace processing package – 4.5 seconds per SP

• 600/700 package

• Diffraction Post stack Migration – 1970

• Diversity & Inversity stack for Vibroseis acquisition in Inglewood & Watts, LA area

600 package move out scan and automatic horizon picks Schneider Geophysics December 1976

TIAC 870 & 870A digital computers 1966-1970 Diffraction Post Stack 2D Migration - 1970

Schneider Geophysics December 1971

Advanced Seismic Computer

TIPEX 3D system 1970 – 199X

• Changed from Shot Point data entry to X-Y data entry

• Spatial Attribute Files

• Land Survey data input

• Marine Navigation data input

• Changed from stacking schedule to CDP binning – described by Ed Tegland SEG 1974

Tipex migration on ASC

Superposition principle Dutch Connection – Huygen’s Principle – every point in the sub-surface is a diffractor

Source X

Point diffractor X

Receiver X

Velocity in Common Tangent Stack – velocity input oriented and un-migrated

Velocity in Diffraction Migration – output oriented but velocity field not migrated

Velocity in Kirchhoff Migration – output oriented and velocity field dip corrected & migrated

Carter & Frazier JGR November 1982

Bruce Secrest in TI CRL – analyzed 15 degree finite difference, Kirchhoff & constant velocity FK migration -1974-1975 Bruce provided “correct” mathematics for Diffraction -> Kirchhoff Migration

Tipex 2D post stack migration on ASC - 1975

Schneider SEG 1976, Geophysics February 1978

GSI promoted Kirchhoff Digicon promoted finite difference Western – Ken Larner – offered both [Houston or London?]

Tipex 3D post stack migration on ASC

Equations Time Hyperbola t0 = [0

2 + (2x/v)2]; 0 is the output time, v is measured at 0 Differentiate time hyperbola with respect to x x = t0v2(dt/dx)|t=t0/4 = t0v2(D0/4G) where D0/G is the observed input dip 0 = t0[1 – (vD0/2G)2] & 0d = t0D where d is the output dip In 3D: In 1975 Jerry Saxton the author of the prototype 3D diffraction migration on the 870 asked me to see if I could speed up the 3D program by a different approach for the ASC. Time Hyperboloid t0 = [0

2 + (2x/v)2+ (2y/v)2] For a line at cross line distance Y we can decompose the hyperboloid to a hyperbola where t0’ = [0

2 + (2x/v)2] and another where t0 = [t0’2 + (2y/v)2]. Note the lack of specificity on where and when v is measured. For Bell Lake 3D 180x96 traces – full 3D -> 17280 input traces/output trace 2x2D -> 62x reduction more typically 600*600/(600+600) -> 300xreduction See: Schneider TLE March 1998

Tipex 3D post stack migration on ASC

In 1975 Bruce Secrest asked me to see if I could use a very fast constant velocity FK migration & Richard Uden asked me why we did not simply perform migration in depth. Equations x = t0v2(dt/dx)|t=t0/4 for a dipping layer to migrate the same distance with another velocity vref we need to transform time and thus dt/dx such that t0v2(dt/dx)|t=t0/4 = tnvref

2(dtn/dx)|t=tn/4 so t0v2dt = tnvref

2dtn -> we can scale t such that tn = kt0 & dtn = kdt0 once migrated the output section can be scaled in time to the correct migration times. Cam Wason, head of GSI R&D ex CRL, was also trying to solve this problem. Cam was very focussed [determined] so it took a while before we could discuss a solution. Cam drove many of the efforts for modern processing methods with great enthusiasm – inversion, wave equation processes and velocity filtering instead of vertical stacking. See: Stolt Geophysics February 1978

TIPEX Dewater 2 & Designature on ASC

In 1976 Davy Einarsson a geophysicist in Calgary reported that he had sold a large Dewater job on the basis of R&D slides for an unreleased product that would be very expensive at that time. Nat Natarajan from London had visited Pete Embree in our AG group and suggested that we could do something with FK filtering for multiples. Cam & I discussed these two items at the coffee bar’s atrium rail. Thus was born Dewater 2, which was implemented very quickly. Digicon, Western and Geco offered similar software also very quickly with some producing brochures before GSI. The AirGun package on the 870 was initially offered with only a non-whitening adaptive deconvolution process. We added a conventional wiener deconvolution in London in order to match processing on the 827. Milo was disappointed when we discussed this. So I added a non whitening wavelet estimation in Designature to accompany the multiple elimination of Dewater2. This included a minimum or zero phase filter to shape the estimated wavelet’s spectrum. In recent discussions, Milo had no memory of the adaptive versus wiener deconvolution so his disappointment was probably just that the AirGun package needed more work. So part of this successful product was likely the result of transatlantic miscommunication. See: Potts et all SEG abstracts 1982.

Arco Workstation #1 – assembled by Aftab Alam

Arco Workstation #2 Maps-3 • Traverses • Flattening • 3D trace brick file • Transparency • Dual image overlay • RGB blended images for AVO • 3D block display • Paint brush interpretation • Integrated plotting • Integrated mapping • Arco Spec for GeoQuest’s IESX

• Synthetic benchmark with all multiples; extended my colleagues for a suite of P traces for an offset synthetic

• Log response deconvolution

• Structural mapping test – 400 well tops, 7 tops held back -> 16 mapping packages • Purchase of BluePack mapping software

• Volumetrics test

• Maps - 1 & Implementing Bok Byun’s velocity inversion [VISAS] as an interactive

package

Some ARCO tasks

GPS

• Seismic crews were early users of GPS, in differential mode it improved positioning to @ 2 meters accuracy.

• TI & Western’s Aero Service were also producers of early commercial receivers

• TI’s 4100 Navstar receiver is shown below with developer Phil Ward.

• Aero’s Macrometer – developed by Charles C. Counsellman III of MIT is also shown. • Phase tracking • GPS antenna • Datum shift function • Ekofisk subsidence • Japan Earthquake forecasting

TI GPS receiver

Macrometer Antenna

GeoQuest’s GeoViz/Voxels

Courtesy Schlumberger SIS

James, First Break, March 2003 James, GCAGS 2008 3D Propagator

James, First Break, March 2009 Co-rendered Amplitude with TerraSpark’s AFE Lineament

Computer Assisted Interpretation in VoxelGeo

Manual Blue, Tracked Purple

James & Klein, AAPG 2007

Computer Assisted Interpretation in VoxelGeo

FaultTrak

VoxelGeo courtesy Paradigm Geophysical

Volume Rendering on the GPU

James et al, First Break, March 2011

Anti-aliasing

India original India with 25% reduction with anti-aliasing + 400% zoom

India with 25% reduction without anti-aliasing + 400% zoom

Anti-aliasing

SeisEarth 3D Canvas courtesy Paradigm Geophysical

• Anti-aliasing applied to seismic, grids, contours, maps, wells and interpretation in 3D, Section and Base Map views

Milo Backus – Research Oversaw: MAE Analog -> Digital transition 2D -> 3D transition Pre-stack transition with SEER @ UT

Bob Graebner – Oversaw the Area Geophysicists

Cecil Green – GSI Founder with Erik Jonsson, Gene McDermott & H.B. Peacock

Thanks for 50 years of fun