An Integrated Approach to 3D y Seismic Interpretation ce] Nee NAUTILUS Rem g a ecay @ NAUTILUS An Integrated Approach to 3D Seismic Interpretation © 2012 by Nautilus Limited and Mark Sawyers All Rights Reserved. This manual, or any portion thereof, may only be reproduced, for non-commercial purposes, by any member(s) of the Geoscience Training Alliance, for use by the employees of such member(s) in the conduct of the business of such member(s). Notwithstanding the foregoing, no member(s) of the Geoscience Training Alliance may copy, reproduce, disseminate, use, publish, store in a retrieval system, or transmit any portion of this manual as part of any training course that such member may conduct for its own employees, its independent contractors and/or others. In all other circumstances this manual, either in paper or digital form, may not be copied, reproduced, disseminated, used, published, stored in a retrieval system, or transmitted, in any form or by any means, whether written, electronic, mechanical, photocopying, recording or otherwise, without the prior written permission of the copyright owners except as permitted by the Copyright, Designs and Patents Act 1988. Nautilus Ltd. Ashfields Farm, Priors Court Road, Hermitage, Berkshire, UK. RG18 9XY Tel: +44 (0) 1635 248910, Fax: +44 (0) 1635 247324, E-Mail: info@nautilusworld.com www.nautilusworld.comMark Sawyers Exploration Geosciences Limited, Director Shear Farm Bar North Road Goudhurst Kent TN17 1JR Telephone: 01580 212080 Website: www.expgeo.co.uk Courses Taught: M255: An Integrated Approach to 3D Seismic Interpretation Mark Sawyers joined the oil industry in 1981, working initially in potential fields processing, interpretation and modelling. In 1985 he moved to seismic processing with Phillips in London and was let loose on his first 3D interpretation in 1987 (flown to Bartlesville 3 weeks before Christmas and told he couldn't come home till he'd finished!) Numerous 3D projects followed in the following 15 years, for Conoco, Sun, Chevron and Enterprise, amongst others. During this time he trained and mentored a succession of young geoscientists, and gave courses in geophysical mapping and physics-based risking. Whilst at Enterprise he was a key contributor to the Paleocene chapter of the Millennium Atlas. In 2000 he chaired the UKOOA (now Oil & Gas UK) Exploration Committee. During a secondment to knowledge management he trained as a facilitator and researched the transfer of technical experiences within a large organisation. He left Enterprise in 2002 to become a director of Exploration Geosciences, a UK based geoscience consultancy. Here he has specialised in prospect generation for various clients, with projects in North West Europe, North and West Africa, India, Eastern Europe, and has worked closely with several major seismic contractors in collaborative projects.An Integrated Approach to 3D Seismic interpretation By Mark Sawyers 1 4 saris Mark Sawyers, Director, Exploration Geosciences Ltd. www.expgeo.co.uk 2ae ae Section No.1 IntroductionIntroduction Introduction agenda Introductions — who am I = who are you — what do you want to get out of this course An integrated approach to 3D seismic interpretation - introducing a 7 step workflow Thinking creatively. The necessity of PowerpointQualifications for this job fee + Mark Sawyers ~ 30 years as a geophysicist — 1 contractor, 6 oil companies, 9 years as an independent consultant at Exploration Geosciences - Interpreting 3Ds since 1987, using 6 different interpretation systems - Interpreted 50 to 100? 3D surveys + regional, prospect generation, well planning, development and production + reviewed 50 to 100? farm-in/partner projects + had exposure to 100s colleagues/client projects — Trained or mentored ~15 junior geoscientists — Hovered between technical and management + chaired UKOOA’s Exploration Committee in 2000 + Co-authored the Paleocene chapter of the Millennium Atlas + frequent member of peer review team + Why? ie: Exploration Geosciences + Small independent UK consultancy — 4 directors, 8 employees + Products — UK and Norway Discovery Digests WHAT Map various vintage technical reports + Consultancy - Working as outsourced exploration team for oil companies + UKCS, Norway, West Africa, North’ africa, Mediterranean, IndiaIntroductions YOUR TURN Who are you? Likes and dislikes What do you want to learn about? = Maximum 6 words, block capitals = Add your initials what sort of company do you work for? + Oil company, contractor, other what are you working on? + Exploration, appraisal, development, A&P, marketing, technical support, other What software do you use? Geologist or geophysicist? How many years experience? Flip chart Hexagons Did anyone bring any examples to discuss? Hexagons 1 YOUR TURN Geophysical likes What's easy? What's hard? i 10Hexagons 2 YOUR TURN a2 + What do you want me to cover over the next 4 days? — How to recognise multiples — How to account for anisotropy - How to structure a peer review presentation — How to autotrack more effectively = (hint ~ none of the above is explicitly covered...) ny Aim of this course @ The course assumes a basic understanding of seismic interpretation procedures The principal aim is to help you improve as interpreters by — seeing the bigger picture (cliché bingo!) = arming you with tools to work better and faster - helping you get your message across How? = by learning from the experience of others (not all giants!) 2Steps in an integrated approach to 3D interpretation 1. Define and understand objectives 2. Organise and understand your data 5. Assess implications 6. Investigate alternatives and risks eel a alaale alee . rc cer YOURTURN %& In teams of 3 or 4, spend 5 to 10 minutes brainstorming ideas for a new and improved technique for seismic interpretation Some examples I wish I had a “goodness” of fit measure when deciding if an amplitude anomaly is coincident with closure I wish my interpretation package gave an audible warning of a mistie = Lwish I was paid a Euro for every pixel picked 19 “Death by PowerPoint” q Phrase introduced by Angela Garber in 2001 and describes the state of boredom and fatigue induced by information overload during presentations created in PowerPoint This is a PowerPoint based course... 20Section No.2 Interpretation fundamentals Introduction to the session + The aim of this session is to recap on some aspects of geophysics that it is hoped that most attendees are already familiar with. + An understanding of these are not pre-requisites for the remainder of the course, but they will help. + The following slides have been taken from Nautilus course F085: Introduction to Seismic Interpretation + Who are the experts in the room?M255: An integrated approach to 3D Seismic interpretation — Waves, diffraction, refraction * c Seismic waves @ Seismic energy propagates through the Earth in the form of acoustic and elastic WAVES SURFACE WAVES BODY WAVESRecording Seismic Reflections a Reflections Receivers Source Direct Arrivals 25m Time “Ee ee Refractions Reflections Seismic Recording Field Record Ray Bending a For V: = 5000,8:=30°, and V2 = 7000, What is 62? incident energy V, interface If V2 is increased to 10000, What is 02? transmitted energy The angle at which the ray leaves the interface depends on the velocities in the two rock layers (note the density isn’t a factor). The law that governs the behaviour is Snell's Law: sin(@,)V, = sin(8,V; The parameter sin(9)/V is constant for a ray at each interface as it passes through multiple layers and is known as the Ray (or Raypath) parameterReflection Coefficient at ‘normal incidence*’ Density p. incident reflected " Velocity V, °ne"ay energy : . Densit MeBaHN transmitted energy Amplitude of Reflected Wave = R x Amplitude of Incident Wave V2 - PrV. Al, -Al At normal incidence : = Pte Pils Aa PV. + PVi Al, + Al, * Normal Incidence means at “right angles” to the reflecting surface Seismic wave propagation Shear wave Compressional wave = P-wave ssos 8 #S-Waves (Elastic) i? Wave (Shear , transverse particle motion) direction © —e— +~e— ~+o— S-waves *S-waves are becoming more important in seismic reflection prospecting. - Fractured and Unconventional reservoirs (due to polarization effects) - Challenging imaging conditions (through shallow gas clouds, etc.) *Special 3-component receivers are required to capture the shear energy. +S-waves cannot propagate in water, so receivers must be placed on the sea bed. -When processing P waves, S-waves can appear as noise. +S-waves mainly respond to the properties of the rock matrix, not pore fluids, so if we can record both P and S we can sometimes differentiate between the rock and fluid properties. +S-waves travel at 25%-65% of P-wave velocities through the same rock matrix. P Wave Arrivals Direct arrival Reflection Refraction AA«Ground Roll 10Primaries & Multiples Source . Primary > Surface Multiple Receiver + Multiples are caused by reflected energy that has followed a different path through the earth than normal reflections = Multiples always arrive later than the primary event they are associated with © Multiples spend proportionally more time in shallow formations and therefore usually appear to have slower velocities than the primaries: Ground Roll 1280 ft/s 390 mis Reflection Shot Record ci a “til i 7 : } 12,000 ft (3660 m) Direct Arrival 4890 ft/s 1490 mis Refraction 7760 fs 2360 misWhat do we record at the surface? «@ Al RC Seismic Low High - + Energy Strength of reflection Time 3 The Real World is Not a Series of Spikes @ Individual Velocity Density RC = Responses wove xX o=» > ‘The seismic “wavelet” is dependent on selsmic acquisition and data processing parameter, and will be discussed in more detal in the following tte processing section. owM255: An integrated approach to 3D Seismic interpretation _ Fold, wavelets, phase crearain IF -oiene fie 4 saurios 15 ae Importance of “Fold” on data quality 15 fold 30 fold 60 fold “Fold” refers to the number of traces that are summed together at each CMP to make a “stacked” trace. Increasing fold during field recording can improve processed data quality significantly by improving random noise = ees cancellation, and also by providing Dec. more control points for an improved "statics" solution. Fold can be increased during the field recording stage by adding more source and receiver points to the recording template. AN 163D Acquisition “Footprint” ‘“@ 3D seismic data often show amplitude and other anomalies that develop because of the way the data are acquired and processed. These are collectively known as the “acquisition footprint” . Proper survey design & good data processing can help to minimize these effects. Processing can minimize but generally not completely overcome deficiencies in acquisition design/execution Good survey design balances acquisition needs with processing capabilities so that fit-for-purpose images obtained at minimal cost 7 + Butterworth and Ormsby wavelets are commonly used examples of zero phase band-pass niters. + The top example shows a relatively narrow band-pass filter with significant side lobes. + The 2" example shows a broad spectrum filter containing higher frequencies yielding a higher resolution wavelet. Side lobe amplitudes have been reduced. + The bottom example has poor side lobe attenuation due to the loss of bandwidth ~ lower frequencies in this case. 18Frequency spectra plots ae ‘eva: tna arttgrateryscent MIG ay Star rca ‘Sr ae Co eas ws a Gea = Fourier analysis may be used to see the bandwidth of the seismic data in a given portion of the data + Frequency Spectra plots can be generated on most seismic workstations + Amplitude spectrum on left shows a broad bandwidth, rich in high frequencies necessary to generate a compact wavelet. + Spectrum on right is irregular and abruptly truncated at about 45 Hz. Data were filtered post-stack as a noise reduction exercise. The resulting wavelet is likely to have significant “side lobe” distortion. Would reprocessing help? 19 a Wavelet phase - Summary « + Zero phase wavelets are generally preferred since they ~ can improve seismic resolution and reduce ambiguity during the interpretation process. + Phase can be modified using Fourier Transforms during data processing and also when calibrating seismic data on the seismic workstation, Wavelet Phase influences the character of reflections + It is important to have a good idea of the phase of the data when performing stratigraphic interpretation 20M255: An integrated approach to 3D Seismic interpretation Processing, velocity C saris 2 Data Processing - Key Concepts @ + Amplitude recovery — Coherent noise filters (optional) + Datum statics corrections * The seismic wavelet & deconvolution + Sort data to common mid-point gathers + Velocity analysis & NMO + Residual statics + Stack + MigrationNormal move-out The NMO equation defines the relationship between the time delay for a reflector as a function of offset distance and overlying velocity. Also known as RMS _ CMP GATHER NMO STACK velocities, the stacking offset,x velocities for the events in a CMP can also be useful as predictors of lithology and abnormal pressure. Time (s) 23 Seismic Velocity Definitions a Normal Move Out (NMO) - Velocity required to “flatten” an event ina CMP gather. t,2 = ty? + x2/v2 RMS Velocity - Approximately equal to NMO Velocity. This equation can be used to predict NMO velocities from well velocity data by summing interval velocities and interval times down to the objective horizon. Vams? = Zvjat/Zat, Dix Equation - Provides an estimate of interval velocities from NMO (or RMS) velocity time pairs. Useful for velocity model building and also gross lithologic predictions from NMO velocity data. Vine? = (W2?+to-Vi?t)/(tarti)Stacking Velocity Analysis - Review Static errors on CMP gathers ft 000 2.000 ‘7 + Typical stacking velocity analysis, = Designed to “flatten” traces in CMP gather. = Performed every KM+ = Analyzed vertically based on coherency indicator. - Adversely affected by multiples and noise. ~ Resulting velocity model not consistent from geological perspective. - Limited by NMO assumptions. + Improved velocity analyses = High density velocity analyses performed at every CMP. = Horizon consistent picking results in better fit to geology = Tomographic (ray-trace modeling) approaches yield superior results 25 These gathers have had NMO and datum statics applied. Residual static errors are present. If uncorrected, static errors will degrade stack Static errors can alter structure, Static errors can affect dip of events changing apparent NMO velocities 26Pre-Stack Depth Migration Summary a + Images (positions) seismic data correctly in the presence of lateral velocity changes. + Better conformance of velocity model with geology. However, more precision in the velocity model is required than is the case for time migration. + Improved imagery than time migration in areas of.. - complex surface topography — steep structural dips — significant changes in lateral velocity within layers — high velocity changes between layers. + Computationally more demanding than time migration. + Active participation by the interpreter often necessary for optimum results. 27 M255: An integrated approach to 3D Seismic interpretation Synthetic seismograms C saris,Synthetic seismograms — phase correction ay (Phase correct the seismic data for a best fit to the zero phase synthetic data) C The synthetic has been shifted in time to tie the seismic data. Note that that the black autocorrelation trace is not symmetric. The interpreter should now consider phase shifting the 7 seismic data so av ezutesy RNOTC, Geasyn exampe Courtesy ~ Bi ckeraoe that it matches After time shifting the synthetic to ties the seismic traces, there may _ the zero-phase remain some differences in reflection shape due to phase differences synthetic between the seismic data and the synthetic traces. seismogram. Problems/Pitfalls with synthetics ae + Wavelet extraction, stretching of synthetics and other ~ capabilities can make a synthetic tie to the seismic nicely when it shouldn’t. — Unreasonable velocities - Unrealistic extracted wavelets + Some other problems: — Sonic & density logs image a limited amount of rock, seismic images much larger area — Borehole conditions may affect log values — Sonic logs uses much higher frequencies than seismic and oftentimes indicate higher velocities - Multiple content in synthetics is difficult to model correctly + Near-surface sonic information usually absent + Normal incidence vs CMP recording 30Calibrate using seismic well tie work flow @ a Use several well control points if available Mis-aligned events 31 M255: An integrated approach to 3D Seismic interpretation Horizons, faults, sequences and attributes wn ne GD ( aur 323-D horizon interpretation “’@ Most interpreters begin by picking a course grid of “in-lines & cross- lines”. This procedure is sufficient for many time structure maps. If the interpreter is planning to generate any horizon attribute maps such as amplitude, dip/azimuth, edge detection, etc., picks will be required for every trace in the volume. Autotrackers can speed up the interpretation process considerably and also provide the detail necessary to generate horizon attribute maps. Generally, the steps are: = Identify horizon at wells — Start by manually picking horizon on a grid of lines - Make sure horizon pick is consistent (loop ties) within fault blocks = Auto-track horizon — Iterate Fault and Horizon picks 33 @ *Reflection terminations that align +Offset reflections +Local amplitude changes “Dip changes -Fault plane reflectors, in certain circumstances ction 2, p *Be aware of expected fault plane angles for your play area! 34Importance of fault surfaces ae > 8 3 $ «Interpreters should also utilize other 0.0 Os indicators such as “conformity to 5 Gas structure” when evaluating prospects Saturation with AVO anomalies. 51 Elastic Moduli: Bulk modulus “K” a “K” is a measure of a rock's resistance to eee a uniform compression, ..its resistance to ee changing volume under pressure, ....its “incompressibility”. K = -V * AP AV Stiffer, more cemented rocks have higher values of “K” than softer, more compressible rocks. Water- filled reservoirs will have often have higher “K” than similar reservoirs containing small amounts of gas. 52Elastic Moduli Shear modulus “N” (or “S”, “G” or “p’”) Shear modulus N (rigidity) is the stress- strain ratio for a simple shear force “F” F applied to area “A”. N= FFI AAx rung, Wikipedia "N” is larger for more rigid materials and is sometimes referred to as the “rigidity modulus”. No change is volume occurs for this type of rock deformation. Fluid content in 23 reservoirs has little or no effect on “N”. Highly cemented rocks have larger values for "N”. 53 A Moduli, density and seismic velocities Bulk modulus K x+a™) V= __\3) P Shear modulus N p =DensityV,/V, and Poisson’s Ratio (0) a (An importa nt parameter for amplitude-versus-offset studies) wf p/Vs)?- 2) = Os sae renee 035 [2(vp/Vs) 2-2) 03 {2 02s 202 Low Vp/Vs means low Poisson’s fos Ratio rock - (like gas sand) o1 ons 0 12 14 18 18222 vpivs is a measure of transverse strain when a stress is applied. Values range from 0 (cork) to .5 (rubber, water balloon). o in a gas-filled sand can range between .10-.20, o increases with water saturation in a reservoir. 55 Vp/Vs measurements allow lithology predictions a Lithology Discrimination 22 20 VplVs 56Poisson's ratio and the a Shuey AVO approximation “ R(O)= R, cos?(8)+ PRsin?(9) Shuey approximation (@ < 30°) where Zero Offset Reflectivity (Section 2, page 40) Poisson's Ratio Term, Poisson's Ratio - measures the degree to which a ‘media bulges as it shortens, or thins as itis. extended. 0< 0 5.5 © = 0 for non-deformable solids, o = .5 for water (uncompressible). For a gas sand, o = .1- .18 97 Session conclusion Any questions? Don’t worry if you struggle to remember the details — as an interpreter you will almost certainly never need to calculate, say, Poisson’s Ratio. Knowing enough to know what to ask is what is important. 584 swan Section No.3 Workflows An Introduction to Interpretation Workflows =ia? Session learning outcomes + After this session the participant will be able to: — Recognise the key steps in a 3D interpretation project = Examine the interpretation process and judge where best to focus their efforts in order to achieve their objectives exploration, eosciencesa Why worry about workflows? « We should be aiming at the most effective way of getting from the starting point to the objective The actual workflow followed for a given objective depends on whether we want: — the best possible results in a fixed amount of time = the shortest time to the required result, or — the best result in as long as it takes to get there + NB this never happens outside of academia! A common geoscience workflow M The workflow shown on the right is a common approach to 3D seismic Get data interpretation « This approach is sometimes good enough, and many geophysicists (particularly consultants) use this pet) exclusively Many managers are very happy with this = visible results can be produced quickly and regularly ae erat This may be the best approach.. Dr ncCesa icy ttad « «What might go wrong with this approach? + Aworkflow that is too basic might lead to some of the following outcomes = Missed deadlines - Recommendations turned down by management = Need to repeat work = Missed promotions = Well comes in low to prognosis — Overbidding or underbidding in licensing rounds = Failure to farm-out = Dry holes — Low levels of job satisfaction — Redundancy or getting fired! The step wise approach (2) @ Here is another recommended interactive 3D interpretation procedure (from Brown, 2004) 1.Preview of data on composite displays and movies 2.Horizon identification at wells. Assessment of data phase and polarity 3.Recognition of major faults on widely-spaced vertical sections. 4.Fault framework by tying together with horizontal sections. 5.Initial horizon control using vertical and horizontal sections 6.Automatic spatial tracking to complete horizon on every point - get to this point as quickly as possible 7.Scrutiny of intermediate horizon products for new features and for validation of tracking 4.Colour-posted time structure 2.Colour-posted horizon slice 3.High spatial frequency residual 4.Dip magnitude and azimuth, difference, edge detection and illumination 8.Revision of horizons and faults, and rerun of auto-tracking. 9.Final time structure maps and horizon slices with chosen amounts of gridding or smoothing 10.Isochron, isopach and depth maps. 11.Detailed stratigraphic and reservoir studiesey But there is more to think about! @ Other things to con: ler include. 1.What are the project objectives? 2.How much time do I have? 3.When is the deadline? 4.Who will be using my results? Future workers, management, engineers? 5.What software do I have available? 6.What is my frame of reference, or starting point? 7.What do I need to do to get up to speed with the area? 8.Do I understand the dataset I have been given? 9.Has it been loaded correctly? 110.Who is providing my well control, and is it reliable? 11.What has been published on the area? 12.What are my real objectives? 13.How will I document my work? 14.How do I need to deliver my results? 15.Do I trust my own abilities, or do I need help? 16.What prior knowledge is there that will affect the way I interpret? 17.What decisions will be based on the results of my work? This course advocates a seven-step approach 1, Objectives defining and understanding your objectives should be your first priority 2. Data organise you database, work out what data you have and what data you need 3. Geometry establish the subsurface geometry with an open mind 4. Geology Infer geology from the pattern of reflector you have mapped 5. Implications Once you have the geological picture, decide what implications are for prospectivity, reserves, well location 6. Alternatives Consider other possible interpretations for what you see in your data 7. Communication Communicate your results clearly and effectively 10The seven steps to... 1. Define and understand objectives rganise and understand your data 5. Assess implications MTN eect eeleoieeiecclaarel encore lace ac Peele sland eral (ese mel Ly The seven steps to... Pid In most cases it will be an iterative process: 1. Define and understand objectives 2. Organise and understand your data 5. Assess implications 6. Investigate alternatives and risks 7. Communicate resultsExamples of the need to iterate @ In a poor data area where interpretation is difficult it is often necessary to adopt a geological model to base the interpretation on ~ but don’t assume that the interpretation then proves the model ‘Once you have defined a trap, go back and check your picks carefully Your seismic anomaly could be: a DHI (direct hydrocarbon indicator), tuning, a lithology effect, an acquisition or processing artefact - go back and interrogate your datasets in greater detail — You may need to revisit your database as well Communicating your results often lead to a new set of objectives 3 e Using the seven steps The workflow describes a seven step approach to geophysical interpretation, developed from over 30 years experience Following these steps is recommended} Referred to from now on as the seven steps Keeping the seven steps in the 3D interpretation process in your mind throughout should help you achieve your objectives more quickly and more effectively During the remainder of the course we will look at each step in detail 14The seven steps Seven Steps to Heaven is the eighth studio album by jazz musician Miles Davis, released in 1963 by Columbia Records. M255: An integrated approach to 3D Seismic interpretation Some cautionary tales ( ey % 2 ay 16The Isle of Wight @ And I’ve cycled round it, sailed round it and flown round it Pre drill seismic “interpretation” case 2Post drill “interpretation” ceo. 271.06, i a : - : 7 ccan a 4s) i sa 3) i Sea 100.9) eeseaty Gtr Te) em-etaresy’ cayenne) supe ata 19 The moral? “Don’t smoke your own dope!” The Isle of Wight Inversion 20What went wrong? + Brabant focused on just three of the seven steps, and in the order shown to the left. Pena noe ‘Almost no time was spent in steps 2 (understanding your data) and in step 6 (looking at alternatives) Penta se In this case time more time spent understanding the data would have saved a dry hole + (even if someone else was paying) a Some more examples, taken from the Central North Sea ie 22Fleming discovery 5 1986 a certain geophysicist’s first seismic interpretation project Objective ~ primary target, Lower Cretaceous turbidite sands had been drilled and proved to be dry. Phillips believed there was limited further potential but still had a commitment to drill an exploration well. + The objective therefore was to convince the UK government to let Phillips off final well commitment + Outcome - objective not met, well drilled, Fleming fleld discovered + Lesson - the conclusion (which was wrong) came first 23 Fleming discovery Field Extent 2415/25b-3 Brenda discovery ie In 1989 Conoco had 2 geophysicists working on newly acquired seismic, A working 15/25, B working 16/21 Their objectives were to find drillable prospects B, on comparing picks, realised that Ahad mis-tied the well control B extended his work into 15/25, and convinced management to drill 15/25b-3 A failed because the subsurface geometry he defined was the wrong surface 25 Macculloch im + Inthe UK 12 Licensing Round a seismic contractor acquired a seismic data set over open block 15/24 + Conoco bought this data (along with an interpretation report) wn) ‘objective of determining whether the block was prospective, and if so, hig preparing the bid document. | vy © Tighe achieves sng the [obit technical work presented in the report. Several of her colleagues were horrified at her plagiarism. From Gunn (2003) + BUT, she achieved her objectives in the most efficient way ~ no further technical work was required. Conoco were awarded the block and to date 26The moral of the stories i: You can do steps 2, 3 and 4 perfectly, but if you fail in steps 1, 5, 6 or 7, you fail. + You will spend the minimum amount of time in steps 2, 3 and 4, if you've performed steps 1, 5, 6 and 7 properly. 5. Assess implications RITES elece hese \e =e ae ed 7. Communicate results 27 = =a Work flow comparison Bid «@ Recommended interactive 30 interpretation ‘procedure (hom Brown, 2004) 1. Define and understand objectives 2. Organise and understand your data 5. Assess implications 6. Investigate alternatives and risks Pertti meee This is a comparison between the workflow described in Alistair Brown’s 2004 book on seismic interpretation land the seven steps presented in this course. Brown focussed his attention almost entirely on steps 2 to 4 (with most time spent in step 3). One of the key objectives of this course is to help the interpreter to recognise the importance of each stage of the complete workflow 28Exercise @ In teams of 3 or 4 — The dataset you are about to be given represents a project in microcosm - In the next 10 to 20 minutes, try to go through all seven steps + Define your objective (hint: this is to go through all seven steps!) + Describe you dataset + Determine the subsurface geometries + Infer your geology + Decide what the implications are + Suggest alternative interpretations (however unlikely) + Communicate your results Workflows: session summary ed E i 5 We have broken down the interpretation process into seven distinct steps We have examined some examples of projects that have failed largely as a result of an incomplete process Session learning outcomes = After this session the participant will be able to: + Recognise the key steps in a 3D interpretation project + Examine the interpretation process and judge where best to focus their efforts in order to achieve their objectives 30Nauths patal ee RPS Ga Section No.4 Objectives cnn cosciencesSession learning outcomes After this session the participant will be able to: — establish project objectives — recognise the project “audience” — define an appropriate set of deliverables Project objectives Defining, and understanding, the objectives is a prerequisite of a successful project Questions to ask (your manager, your team & yourself): — What are the key deliverables? — What are the deadlines? + Are they targets or drop-dead dates? — What type of project is it? — Who is the target audience? — How much time do I have? — Who else is in the team, and what is my role in the team? — What decision rests on the outcome of the work: IK i i i iay Types of project 1 In terms of exploration/appraisal/development phase + Prospect generation + Prospect refinement = Regional evaluation + New area prior to new ventures + Regional petroleum geology to support on ging programme = Prospect generation = Licence evaluation ~ Well location + Exploration, appraisal, producer, injector - Farm-in evaluation = QC or audit Types of project 2 « + Categorize by starting point In terms of data + New 3D + Reprocessed 3D + Re-acquired 3D + Re-interpretation in the light of new well data = In terms of well control + No wells + Re-interpretation in the light of new well data + Many wells to tie inPrimary and secondary objectives Whilst your project will likely have no main objective or purpose, secondary objectives need to be recognised Example Primary objective: Evaluate block to determine whether to make a licence application Secondary objectives, without which the primary objective is unlikely to be achieved = complete the project on time and within budget — ensure results are presented clearly for managers and potential partners = record work so as to make bid preparation easy — focus on prospectivity that fits the company risk profile and is likely to prove commercial = ensure integrity of technical work Which is the more useful image? Who is the interpretation for? Whatever the main objective, the exact scope of the project and especially the. presentation of the results will vary depending on the intended audience. The secondary objectives and the agreed list of deliverables can be very different if the decisions based upon the results are taken by an exploration manager (2 geologist by background) or a field supervisor (an engineer). In the following slides we look at some examples > rN 2 ey iManagement ae If the final outcome of the project will be a decision made by management ~ such as a farm-in approval, or a licensing round application - your objectives will be a set of clear and easy to understand recommendations, backed up by a sound technical presentation and maybe a report. You will need to have a good quantative assessment of volumetrics, risks and costs. Geologist % If you are going to hand your results over to a geologist, who will calculate reserves or design a drilling programme, your primary concern will be to deliver a reliable set of depth grids - working under the assumption that you will work together on steps 4, 5 and 6 together. A geologist will care about your levels of uncertainty 10An engineer «@ + Areservoir engineer will want reservoir depth grids. A driller will want a target location with tolerances, and a prognosis, with +- error bar. + Once engineers get involved, money is likely to be spent, so although the deliverables may be limited in extent, confidence in this must be high. u Aclient ie + Aclient won't know what he wants but he will want it now. Be prepared - documenting as you go along will be the key. I want all my groceries in went the bog want to be heavy. 2A publisher «@ + If you are producing work for a non- exclusive study you will most likely need to cover each of the seven steps in some detail. The reader will want to know what your objectives were, what data you used (or didn’t use), a comprehensive set of maps and illustrative cross sections and the geological interpretations that follow from the geophysical maps. You will want to try and consider all options (you won't be around to make changes to your maps in response to well results) You will need your presentation to look immaculate. 3 A buyer or farminee a @ A farminee may not have time to do their own work, so they will thoroughly QC yours. Your depth grids may not need to tie offset wells with great accuracy for your prospect to remain robust, but you will have to be able to demonstrate that. Bottom line figures must be clear and robust. Incorporating analogues into your work will increase buyer confidence. 14A prospect inventory The objective is prospect generation ~ be creative {Soe The technical audit committee + This might be a peer review or a team of corporate experts - either way possibly the most critical audience. + With experience you will know what to expect and therefore how to prepare. + Ifyou have company standards and procedures then one of your key objectives will be to follow them! 15 16Deliverables + Thinking about (and agreeing) the end product before you start will save huge amounts of time + Actual delivery varies widely from project to project, from - UTM coordinates and a well prognosis to = A full technical report of 10,000 words and 100 figures and — every thing in between + But remember, the final deliverables will almost never match the originally agreed deliverables... 1 mer eal acon a om a © grinning ih ee ee Could you just... Once you have delivered or presented your results, a common phrase to hear from managers, clients and partners is “Could you just...” Here are some common requests: could you just.. —_..send me your seen picks = ..send me your velocity grids ~ .. make an unconformity subcrop map — .. supply depth maps for the overburden For an easy (well, easier) life, add a list of possible deliverables ~ Have these in mind during the project work 18Exercise YOUR TURN a ie You have a one month viewing licence over a multi-client 3D covering around 1000 km? of the central North Sea, what might your “deliverables” be if: = you are the publisher of regional reports = your company is valuing a portfolio of licences within the area = you are looking for drillable prospects to farm-in to — your company has encountered an unexpected pay section in a well = you are a PhD student Work in teams, brainstorm, discuss, prepare a 1 flip chart page listing Use the seven steps Use the seven steps 1. Define and understand objectives anise and understand your data 5. Assess implications 6. Investigate alternatives and risks PAC oyu alate ales \ Mae 20eee YOURTURN %& Predict some “could you justs” a Fine tuned objectives + Project objectives are not all there is to think about: every horizon may have a different set of objectives within the overall context of the project. + It is useful to get into the habit of thinking about the objectives of each stage of the project + Consider the objectives for each horizon —_ Is it prospective? Is it useful for isopach? — Is it just for structural control? — Will it be used for depth conversion only? = Is it too early to decide? 2Fine tuned workflow Your objectives, workflow and deliverables will vary from horizon to horizon Prospective horizon most likely fully autotracked misties resolved fault planes defined fault polygons derived/drawn tied to wells position on event understood Isopach use autotracked or finely picked then gridded Structural control coarsely picked fault planes defined fault polygons defined Depth conversion coarsely picked Objectives: session learning summary In this session we have looked at the importance of understanding and agreeing the project objectives before you start work We have seen how your objectives can vary depending on the audience and how each phase of the project will have a different set of “secondary” objectives. Session learning outcomes = After this session the participant will be able to: + establish project objectives + recognise the project “audience” + define an appropriate set of deliverables )) « 23 @ 24nn GF © ( saris Naar parthe APS omy Section No.5 Understanding your data Session learning outcomes M + After this session the participant will be able to: = assess the quality of 3D seismic data sets - develop strategies for handling large multi-volume 3D data sets in an effective manner = assess (or establish tests to assess) which are the optimal datasets to use in order to achieve the project objectives — compare attributes and focus on the useful onesSession topics Recommended approaches to organising and understanding your data Looking at data types Attributes Using velocity data Loading and viewing seismic data Datasets - good practice (1) Establish your area of interest = ina small area, expect to use all of the available data (or at least look at it) — ina larger area, define a subset (bearing in mind your objectives and time frame) Define your stratigraphic limits — be aware of deep structural controls . on deposition = Work out in advance which horizons will be required for depth conversion = Do not neglect the overburden (as we shall see later) @Datasets - good practice (2) + Talk to the geologist(s) — which wells in the area should be tied into the interpretation? = Which wells in the region have the target reservoirs? + Talk to the processors (if you can) — Were there any issues such as multiple problems, statics, hard sea bed? + Talk to previous workers - Some of the mundane but useful stuff doesn’t make it into reports Datasets - good practice (3) it + Find out as much about your seismic as you can — What is the phase and polarity of the data, the bandwidth, resolution and tuning thickness? - What is the datum, how does it tie with: overlapping surveys and wells? — What is the signal to noise, does it need conditioning, if it's depth data what was the velocity model?Datasets - good practice (4) «@ + Auto-pick the water bottom (or a flat lying near surface reflector) + it's (normally) quick and easy + it gets your eye in + it'should be unaffected by multiples, diffractions: + amplitude extraction can reveal acquisition footprint, data merges + if water is deep enough it’s a first look Seabed reflector amplitude at the seismic wavelet (although it won't be the same phase as at the target!) + you'll need it to pseudo-depth correct seismic displays in areas of variable water depth + you'll need it to depth convert (more often than not) Datasets - good practice (5) im + Extract a statistical wavelet + Create an amplitude spectrum + Calculate a tuning curve a | -Datasets — good practice (6) A + Run some filter tests - Check if the loaded data is raw or filtered = Use workstation filtering to investigate the frequency content of the data + Filtering low frequency and high frequency noise can improve autotracking efficiency exploration eosciencesWhere are we? a 1. Define and understand objectives a 5. Assess implications CRIN ESET staar- lene) Peel ralanle ale leom aes) Topic introduction Starting point: conventionally acquired, pre or post stack migrated 3D marine seismic data, delivered as seq-y files on disk, or loaded as a data volume If that’s all you have: = you're very lucky, or - you're in trouble! 30 years ago the final migrated stack was the just about only 3D data type used by the interpreter In 2012 a typical data set will comprise 5 or more data types Aim of this session: = to develop an understanding of LL SE multi-cube data sets and to have Lif fe strategies for dealing with them 3 FEE fipsmy The memory eaters Depth volume Shot cDP records Gathers \ Migrated eens ll stack stacks Data types used in 3D interpretation e + Pre-stack = Common-mid point gathers + Partial stacks Angle stacks (nears, mids, fars) + Partial stack derivatives ~ Elastic impedance inversion — AVO attributes (offset, gradient, fluid factor etc) + Full-stack — pre and post stack migrations Full stack derivatives = Acoustic impedance inversion = Instantaneous frequency, amplitude ete Related data — Velocity fields 14Some data «“ + These might be quotes: - “Data costs money”. - “Managers are mean”. — “Data vendors* are sharks”, = "So are lawyers” = "Seismic processors are so slow”. + All of which means: = If you are working on a development or producing field, you will probably have access to all the data ever acquired across your area of interest. Your problems will be in managing the data. = If you are working exploration for a small independent oil company, your problems will be deciding how best to gain access to data: 1993 released 3D for copying costs or $5mm for the latest long-offset data acquired last year. With uplifs for nears, mid, fars, ultra-fars, El impedance. Not forgetting partner uplifts. = Most of us work somewhere in between these two end members. Which means managing our data and worrying about which additional products to purchase and who can use which data for what... + In other words, you'll have too much data, or not enough + bata vendors are sctually marvelous people inthe fortron of techlogy asvancement 15 Rules for handling large datasets 1. Compile a database map showing all the datasets available together with derivative volumes, well control, regional lines etc. Include in this stage data access rights. 2. Decide which is your primary dataset. This will normally, but not always, be the most recently acquired or reprocessed 3D. Don’t always assume that newer is better. ‘An older survey, where langle-stacks are available, may make a better primary dataset. Similarly, the depth volume might seem the obvious primary dataset, but your gathers and angle-stacks may be in time. 3. Divide your remaining data into interpretation aids and anomaly analysis, bearing in mind that some data may fall into both categories. 1. Interpretation aids ~ Al inversion, coherence, curvature 2. Petroleum geology analysis ~ AVO attributes 4, Decide if you have enough data, or are likely to need more. Start the ball rolling on accessing additional data if required. 5. Use your primary dataset to reach your first pass at your primary objective. Do not get buried in data manipulation looking for the quick answer. 6. Check you first pass results using your interpretation aid datasets, infiling and correcting as required. Analyze your plays, leads, etc. using your analysis datasets. 1. But try to look at everything, in case you have missed something! 16M255: An integrated approach to 3D Seismic interpretation Seismic Data Volumes Stacks, angle-stacks and gathers err GF essen i 4 wauritus v7 Data types @ In the following section we look briefly at some of the common data types What proves useful varies from area to area, and with the maturity of the play, and, of course, with the project objectives The more you know about your data the better - read the processing report, or at least the SEG-Y headers + make use of your own processing knowledge, if you have any 18Data types: Stack Assumption (in the 21% century) that data will have been migrated (normally pre-stack) = very, very, rare now to see an un-migrated stack Can be in time or depth domain Will normally be the primary data set for most interpretation projects By definition, the final stack contains virtually all, of the useful seismic data, and, as a good rule of thumb, if you can't see it on the stack, it might not be real. But also, just because you see it on the stack it doesn’t mean it is real. Why do we stack seismic data? To cancel out random noise and boost signal to noise To attenuate multiples To reduce the data volume to a manageable size - for display — for understanding 19 20What about angle stacks? + Generation on angle-stacks didn't become a standard product until the mid 90s ? ere I + Initially gathers were stacked according t to offset distance, with the angle of offset varying down the trace. + Standard practice now is to stack i | eu according to a calculated angle of “ 2 incidence it + Angle ranges typically 0-12, 12-24, 24-36, These are often referred to as nears, mids, i i and fers, following from early days of using - ea LL offsets rather than angles. Ultra-fars, such Ml " wn as 36 - 48 degrees are also routinely From Bacon 207 produced. + Some of the AVO approximations break down at wider angles, but that doesn’t mean ultra~ fars don’t contain useful information. a1 Angle stacks (2) @ + Angle stacks are likely to have: - different final processing steps to the final migrations = Increased noise (both a result of low fold stacking and the final processing steps) = Significant multiple energy in the nears (but not always) = increased stretch/decreased frequency on fars — complex geometries appearing very different across the angle range + Angle stacks are used for = AVO (or more correctly AVA) effects - multiple discrimination/identification - thin bed interpretation on the nears = a better understanding of what makes up the full stack — input into attribute/elastic impedance calculations + Certain events may be more easily interpretable on angle stacks than on the full-fold data 2Characteristics of angle stacks «@ Characteristics — Nears are higher frequency, but with increased multiple content = Mids generally look most like final stack — Fars are lowest frequency and often noisy, but they are often where the interesting amplitude anomalies can be found — Fold of coverage varies Offset stacks are a compromise: most interpretation systems do not allow access directly to migrated gathers, nor would most interpreters have time (or the brain power) to work with them The following slides use an example dataset from the UK North Sea to show the characteristics of the different stacks 23 24Full-fold ie Nears ae Increased high frequency noise, more multiple energyMids similar to full-fold, but note improved imaging of steeply dipping reflectors in lower % of section Fars limited depth range, amplitude anomalies associated with class Ill sands 28Shallow section Full-fold aeNears Mids 32.Fars @ Deep section e 34Full-fold Nears 36Mids Maga f if ( syeoy oF NM) ON 38Orthogonal line Full-fold 40Nears Mids 42a Fars 43 Observations ae + The relative amplitudes and positions of reflections can vary significantly between angle-stacks = AVO effects — complex geometries - migration and focussing effects — undershooting shallow features — poor processing... + Angle-stacks allow the components of the stack to be broken down and better understood + Some events will be easier to pick on a particular angle stack than on the full-stack 44Why use angle stacks? « + Most obviously to allow investigation of variations on amplitude with offset (avo) + To generate avo attributes such as gradient, intercept and fluid factor But they are not just for AVO! The stack process itself is a compromise, and clever use of angle stacks can really add to an interpretat = The high bandwidth of the near-offsets can mean much better resolution than the full stack, ‘so mapping of thin beds, onlaps, pitchouts etc. can be much better carried out on the nears. - Multiples should show rapid dimming across the offsets, so comparing an event across the angle range will make multiples easier to recognize. ~ Dipping beds beneath a flat lying unconformity can be obscured by multiples (and could aliasing be a factor here?) on the full-stack. Some events are imaged much better on the mids or fars. + Early in the interpretation (as soon as you have identified the zone(s) of interest) take a close look at the angle stacks and decide how best to pick your events. Don't wait until you have finished the interpretation and then just look for avo anomalies. + PowerPoint can be incredibly useful here as it provides a means of flipping backwards between datasets in a way not always seen in interpretation systems. 45 Recommended angle-stack procedures ae + Inspect far-offsets for high amplitude "soft" reflectors - the simplest form of anomaly hunting. In the right circumstances a class 3 sand will be the brightest thing on the section + Generate avo attributes and identify avo anomalies + Extract amplitudes from horizon based windows - Look for depth controlled anomalies on these maps and avo derivatives + Personal favourite: interpret a seed grid on the full stack. Snap this onto nears, mids, fars. QC snapped horizons, then autotrack. Before calculating near - far etc., often a good idea to correct amplitudes against a background pick, or windowed amplitude extraction above the zone of interest. Obviously near/far or other ratios do not need background correction = Don't forget to also look at time differences between the picks as well as amplitudes as these will reflect phase changes 46Handling gathers « + Gathers contain everything the interpreter needs, but just aren‘t easy to handle! Interpretation software increasingly allows gathers to be easily incorporated into the interpretation process, but most interpreters do not have the facilities to handle gathers on the desktop + But you should inspect sample gathers early in the project — Are they noisy? = Is the mute appropriate? — Are they flat? + Ifyou have established potential drilling — Are they balanced? targets/locations based on some AVO — are they suitable for your project response, always carry out a detailed objectives? visual inspection of the gathers in the area to confirm that there are no data issues. 47 Example AVO anomaly on a gather “i Postion of Position of Gather Gather am mm y mo mm moe ved Seismic Gather Near Stack Far Stack NB ~ this is a good response in a good data area! 48M255: An integrated approach to 3D Seismic interpretation Seismic Data Volumes Seismic Attributes cartoon IF exsene a ( rete 49 =a Introduction «@ An attribute is any measure (or display) of the seismic data that might be used by the geophysicist to improve the interpretation or to better display certain aspects = ugly definition but this can be an ugly field ‘There are a huge number of attributes and more seem to get invented each year Some are excellent, some are redundant, some are useless, some are misleading (worse than useless). Many attributes can be applied to seismic data volumes, to interpreted horizons (grids), or to sub-volumes of seismic data generated from a horizon based window. — make sure you know what you are looking at! 50Single-trace attributes @ + Instantaneous attributes: - Phase - Envelope ~ Frequency — Complex attributes (Hilbert transform) + These are normally calculated “on-the-fly” and not stored as individual cubes + Used less than they used to be as better displays have been developed and multi-trace and angle-stacks etc. have become available + Still can be useful in helping to map certain features consistently si + Hilbert transform converts seismic trace into complex trace z(t) + x(t) = x(t) + y(t) where = x(t) is real seismic trace = y(t) is imaginary seismic trace + can be visualized as helix in xyt space + instantaneous attributes have the benefit of being almost instantaneous displays on most work stations mw — em Hage 010 52Instantaneous trace attributes @ The Hilbert transform can be used to calculate instantaneous amplitude, phase and frequency of a seismic trace Interpretation aid - a different way of viewing data. Instantaneous phase is perhaps most useful for defining terminations, onlaps etc. A good reference (if available) Most interpreters rarely use these. Most likely to be of use were there is uncertainty as to where a pick should go as they give another perspective. 53 Instantaneous attributes @ sed fouanbeia edojenug, Veg Nw! Vy nnn 34Normal amplitude data ia 56Instantaneous phase Instantaneous frequencyInstantaneous phase example Final migration Noise cancelled Instantaneous phase This is a detail of a thrust fault, deepwater Niger Delta, taken from the Virtual Seismic Atlas, Author: David Iacopini The instantaneous phase display makes picking the fault plane easy ee nat al pm ge Dt eal Eames Adtran 59 Multi-trace attributes ee + These work by comparing adjacent traces or groups of traces and include — Dip magnitude and azimuth — Coherence + Semblance - Curvatures - Texture + energy + entropy + contrast + homogeneity + ete, + Google “Chopra Marfurt" for a for a whole list of references 60Dip magnitude and dip azimuth Mi These have been around for a long time as grid horizon derivatives, used for highlighting faults, discontinuities and sedimentary features More recent algorithms have been developed to generate dip information directly from seismic data Used as an interpretation aid, Particularly as horizon slices ~ convert and divergent package mapping 6 Closely related to continuity and semblance Essentially a calculation of the similarity (or difference) between adjacent traces Can be calculated for an entire volume, or on data slabs (usually horizon driven) Useful for faults, channels etc. 62Comparison between coherence slice and time slice, ay offshore Trinidad @ From Gersaterkorn (1298) Curvature Started as a horizon/grid derivative but now increasingly generated from seismic data directly Useful for highlighting: = faults and folds - fractures - channels - differential compaction I, IR CNSpectral decomposi + As beds thin, their top and base reflections will constructively interfere at progressively higher frequencies. Spectral decomposition makes use of this by splitting the seismic data into discrete frequency bands (in effect Seer rnmaney =e very limited band-pass filtering) + Spectral decomposition volumes can be very good at showing complex channel geometries 65 a Recommendations for the use of attributes + Find out what, if anything, has proved successful in the area - If an attribute has a proven track record, include it in your work flow — If no track record, by all means spend a short time trialling attributes at the data review stage, but don’t expect success + Assign your attribute datasets to your “Interpretation Aids” = Use them if you have problems interpreting your primary dataset + At prospect/discovery/field definition stage use attributes to enhance your understanding as far as possible + Always check your observations on your primary dataset + Don't get carried away interpreting.... 66Redundant and useless si @ Arthur E, Barnes GEOPHYSICS, VOL. 72, NO. 3 MAY-JUNE 2007; P. P33-P38, provides the following words of wisdom + Seismic attributes should be unique. You only need one attribute to measure a seismic property. Discard duplicate attributes. Where multiple attributes measure the same property, choose the one that works best. If you can’t tell which one works best, then it doesn’t matter which one you choose. + Seismic attributes should have clear and useful meanings. If you don’t know what an attribute means, don’t use it. If you know what it means but it isn’t useful, discard it. Prefer attributes with geologic or geophysical meaning; avoid attributes with purely mathematical meaning. + Seismic attributes represent subsets of the information in seismic data. Quantities that are not subsets of the data are not attributes. + Attributes that differ only in resolution are the same attribute; treat them that way. + Seismic attributes should not vary greatly in response to small data changes. Avoid overly sensitive attributes. + Not all seismic attributes are created equal. Details of implementation can be important. Avoid poorly designed attributes. + The utility of a seismic attribute is readily judged through visual inspection aided by crossplots, histograms, correlation, rank correlation, principal components analysis, and spectral analysis. + The following slides use examples supplied by Arthur Barnes or List of seismic attributes, by property measured ae ampitude Phase Frequency Discontinuity Miscellaneous jection strength startaneous phase instantaneous correlation lscontinuitre length, frequency " ace envelope ‘sine of phase weighted average femblance dscontinuitylenergy halftime instantaneous frequency rstantaneous ‘apparent polarity weighted average fovariance discontinutyavadrature trace amplitude | Fourier spectral requency Jiwsampitude fesponse phase sponse frequency figeratructure [aip-animuth iscontinuity erage ortotal weroge phase ecral bandwidth PCL ade velit oeolvte amplitude maximum peak unwrapped phase spectral frequency |PC2 Karhunen-Loeve ~ frinimum trough lance signa complexity mplitudes erage energy total Keectrliurose «PCS mth = ip lnerey - andard amplitude jostantaneous Iamoco © Lipwrince riance frequency slope ormalized amplitude beraleism .and this is not a comprehensive list!_ How does the interpreter avoid getting lost? 68Comparison of amplitude attributes ‘Avg. Reflection Strengt RMS Amplitude ‘Amplitude Variance Max. Peak Amplitude Max. Trough Amplitude ‘Average Energy FT Max. Abs. Amplitude ‘Avg. Peak Amplitude Total Abs. Amplitude In this example, one amplitude attribute is enough Comparison of discontinuity attributes ~~ Covariance Weighted Correlation In this example, one discontinuity attribute is enough @Comparison of frequency attributes Instantaneous f Response f ‘Weighted average f In this example, one discontinuity attribute is (more than) enough Removing amplitude variation « Cosine of phase A short AGC (or very high gain) is as good as cosine of phase, and easier to understandInstantaneous phase vs. amplitude gain ie Onlaps and subcrops can be seen equally well on both sections Time slice vs. coherence slice i Radial faulting around this salt diapir can be readily identified on the coherence slice (right). Once you know they are there you can easily pick them on the time slice.@ Shaded relief vs. coherence slice This isn’t an exact positional match, but the shaded relief view of a well-autotracked horizon is easy to interpret and relates directly to the horizon of interest 8 Dip azimuth vs. shaded relief fe This dip azimuth display (left) is harder to interpret than the simple shaded relief display on the right. a6Dip azimuth vs. Jimi Hendrix @ The dip a: my bedroom wall in the 1970s. I know which I would rather interpret. List of attributes after ruthless clean up This is Art Barnes's selection of useful seismic attributes juth display (left) bears an uncanny resemblance to a poster that hung on 7 a» ie nergy halftime fon te change) allelsm mplitude Phase Frequency Discontinuity IMiscellaneous fetlection ewrengh [response phase frequency liscontinuity yraded rele? hormalized amplitude lbanawiath = zimuth, dip 7 pesicin Develop your own list. The useful attributes will depend on seismic data quality, structural and stratigraphic setting and, of course, the software you have available. 78Attributes versus the interpreter “@ + The experienced interpreter’s eye is as good at recognising subtle changes in the seismic data as most (any?) attributes if you see an interesting pattern or anomaly in your attribute you should be able to see where it is coming from in the original data - If you can’t, treat with extreme caution + Use attributes in one of two ways = look for patterns in the data, go back to the stack(s) to see what is causing the response — observe a feature in the stack(s) that is time-consuming to map, find an appropriate attribute that allows it to be mapped quickly, cross check the results against the original data 79 Attributes discussion YOU R TU RN fe + In teams of three or four, what attributes have you used (or seen), and which have you found useful? + Consider - what was the attribute, what property of the seismic data caused the response, what was the underlying geology, 80M7255: An integrated approach to 3D Seismic interpretation Seismic Data Volumes: Using velocity data ( suru Velocities Mi Types of velocity data = stacking velocities well velocities average velocity interval velocity = Instantaneous velocity Most obviously - converting time to depth — horizons and grids = seismic data Modelling = ray-trace = rock physics As an interpretation tool = regional velocity breaks generally correspond to regional sequence boundaries = lithology prediction = confirmation of pull-up/push-down - pore pressure prediction 82Handling veloci How you handle velocities depends on your software. And the age of your survey Recommended procedure - obtain or generate an interval velocity cube (for example, using stacking velocity cube and key horizons) = depends if velocities are well understood and controlled — generally migration velocities are of more use than stacking velocities — load into your interpretation project and use (with caution) + to guide your depth conversion + tocreate a “depth” volume + to use as an interpretation tool — gas chimneys and other shallow anomalies — structural inversion = major changes in sedimentation Remember: stacking velocities are a measure of the horizontal move-out velocity which stacks the data optimally and is not the same thing as vertical interval velocities. Raw Velocities The raw stacking velocity field shows a general increase in velocity with depth, as would be expected, {generally smoother in the near surface where data quality (less noise) is better and the geology is simpler. But it would be difficult to pick velocity boundaries on these data 4 a 3 4 f| LiLiiii 83 a 84Dix Display « The Dix display (internal velocity in between picks) allows a better Visualisation of the velocity field with depth. From this display velocity boundaries can be interpreted This process should be done in Conjunction with well data , if available. The velocities for each layer can then be extracted , using either single interval velocities for each interval or VOK profiles. 85, Seismic velocities: use with caution + Velocity maps derived from stacking velocities should be considered as a guide and the results must be treated with caution. + If you have wells in the survey which show a velocity trend and this trend is not seen in the stacking velocities ~ be very careful. + Typically well velocities will be ~95% of the stacking velocities. + However in the absence of all other data and to determine intra-well velocity changes , stacking velocities may be very useful. 86M255: An integrated approach to 3D Seismic interpretation Seismic Data Volumes At the work station ( Data loading issues aY « 2 The way seismic data is loaded onto the workstation can have a significant effect on the success of the interpretation project + loading more data than is needed will slow down the speed of your interpretation but loading too little data can mean throwing away important information! When loading data there are a range of things to consider, including: 1. In-line and cross-line spacing Dynamic range Scaling and clipping depth Filtering . Time range 87 ay ae 88Some data loading suggestion a As a general principal you should reduce your file sizes to a minimum by only loading what you need to work with You can, in many instances, without affecting your results load every other line and cross line (sometimes less) + load the data as 8 bit load data only down to the base of your zone of interest + load a limited range of XY ‘The aim is to speed up your data display, autotracking, random track generation, volume attribute calculations and so on you don't need to load the full length of the data ~ processing length is long to ensure all useful data (diffractions, steeply dipping events etc) are fully recorded don’t need to load 12.5 x 12.5m data if you don’t have steep dips 89 Structural aliasing fe 20Aliasing spacing « Frequency 10 20 30 40 50 60 70 80 90 5 714 357 238 179 143 119 102 89 79 10 354 177 118 89 71 59 51 44 39, 15 233-117 78 $8 47 39 33 29 26 @ 2 172 96 57 43 34 29 25 2 19 Se ee ee Bs 89 45 30 22 18 15 13 1 10 Qa 63 31 21 16 13 10 9 8 7 Vf) oy Oe ae 0 Use this chart to determine the required cell size e.g. if you have dips of up to 30 degrees, and frequencies up to 60 Hz, then your cell size needs to be less than 18m to avoid spatial aliasing. Bear in mind you may want to be picking on random lines that are at an angle to the grid 2 SY) ws Example: aliasing vs. frequency s I i 5 92Effect on random lines @ 93 Effect on random lines a The jitter seen on the seismic display is not aliasing, it is the result of the software program selecting the closest trace to each point on the random line, rather than interpolating the trace from the surrounding data. You are actually looking at a zig-zag line 94Effect on random lines (2) Data loading Bit depth = Supplied SEGY data is likely to be 32 bit — Your screen's resolution is 8 bit Should you load at 8, 16, or 32 bit? Advantages of 8 bit: — Smaller file sizes — Faster line displays — Speedier interpretation Downside to using 8 bit: = Worrying that you have missed some important amplitude information! 95 962 The effect of bit size 2 Sy s 7 The effect of bit size a»EY S bit vs. 32 bit ie One of these sections shows data that has been rescaled to 8-bit, the other shows data that has been loaded at 32 bit. The colour bar has been manipulated to show only the strongest amplitudes. They look the same because the interpretation software can only use an 8 bit colour bar (256 colours). 99 a The effect of decreasing resolution « Original size 1518 x 1094 Original size 300 x 216 100The effect of changing the colour depth T-bit c Sit & Data loading: how many bits? a 6 bt 101 ay + 3D seismic data is normally processed at 32 bit - Amplitudes will be in the range -2147483648 to +2147483647 + Seismic data is normally loaded to the workstation at 16 bit or 8 bit - Data must be clipped, scaled or resampled — How this is done depends on software and project objectives 102Full 32 bit amplitude distribution M This shows a smooth distribution of amplitude in the range -2147483648 to +2147483647 103 ay Clipped distribution In order to load the data at 16 or 8 bit we need either to scale or clip the amplitudes (or both) in order to preserve a good spread of amplitudes in the middle of the range where most samples sit 104Rescaled ee —— (eee — — — wig oe The alternative is to rescale. The amplitude range is preserved, but the amplitude distribution is now stepped rather than being smooth. 105 What does this mean? i: We have already seen that there may be no visible difference between the seismic data viewed at 8, 16 or 32 bit But what happens when we map amplitudes or other attributes? 106,32 bit Ae This is part of a velocity map, with the velocity values stored at 32 bit 107 16 bit @ Here the velocity values stored at 16 bit - there is no visible difference 1088 bit At 8 bit, the map becomes noisy, with a banded appearance 8-16-32 bit in profile 109 im 11016-32 bit in profile (zoom) At 8-bit, there are not enough available numbers to accurately define the profile Bit depth conclusions The bit depth of your data will have little effect on your structural interpretation ~ the times (or depths) at each CDP will be within one sample of each other Your amplitude maps may however look different - the 8 bit data will appear “noisier”. This may not matter, particularly if your data is already noisy, but it may be important: subtle depositional patterns or hydrocarbon effects may be masked. Pay attention to your data loading Make sure it is optimized for your project objectives Create subsets of your data (either in term of bit depth or in terms of xyz limits) to make your workstation as speedy as possible = interpreting on an 8 bit volume and extracting amplitudes on a 16 bit volume may represent the best of both worlds If possible keep seismic volumes on fast local disks rather than trying to run across the network. Your project can stay on the network so that it can be shared and backed up. e 12iy ep nanpe e + Modern 3D interpretation data sets can contain a huge variety of data types ~ Understanding and organising your data early in your project will save time and lead to improved results later on + Prioritize your data, but don’t view each dataset in isolation - A final stack, 3 angle stacks, an impedance volume, a discontinuity tool, and the ability to view shaded relief is all you need + Be ruthless/rigorous in identifying redundant and duplicate data + Don't believe anything you see on attribute maps and volumes without going back to the stack data to find out where the response is coming from + Use whatever visualisation software that you have available + AVO anomalies identified on derivative or partial volumes should be checked on gathers 113 Data types: Session learning outcomes + After this session the participant will be able to: = assess the quality of 3D seismic data sets - develop strategies for handling large multi-volume 3D data sets in an effective manner assess (or establish tests to assess) which are the optimal datasets to use in order to achieve the project objectives — compare attributes and focus on the useful onesexploration AM cosciences Section No.6-1 Subsurface Geometries Session learning outcomes After this session the participant will be able to: ~ Analyse their data set to decide on the most effective picking strategy ~ Practice good habits to enhance data and knowledge management Assess the value of different ways of extracting information from 3D data sets (surface picking, extracting volume attributes, geobody propagation, etc.) x : 5 5The seven steps to... 7. Communicate results 2 Defining Subsurface Geometries Detail of this covered extensively in the rest of this course Here we will look at: simplified workflow some good habits to adopt finely tuning objectives and deliverables some QC techniques a@ A simplified approach... 1 Peel ta on comps pee and mos aaa 1. Identify horizons to be picked, using ‘2Horizon identitcation at wells Assessment of data phase and well ties as necessary poiarty {Recognition of major auton widely spaced vertical secon, «4 Fautamenerkty tying together with horizontal sections. 2. Interpret horizons and faults Sn horizon contol using vera andhorsona sections ‘© Automate spatial tacking to complete horzon an evry point = get tothis poi as qucky as posse 3. Make maps 7 Seratny of ntrmesatshorzon product or new features and for valida of racing {8 Revision of horizons and fat, and rrun of auto-rackng . 4. QC and check well ties ®8.ina Ye structure maps and horizon sices wth chosen amounts of grasing oremoainng 1{soctvon lspach anc dept maps, 5. Produce final displays 1 1.Detaled stratigraphic and reservoir stusios a Picking strategies « + How you go about interpreting your horizons depends, as we have seen, on your objectives, but other factors of course play a part + The biggest factor is data quality, or more accurately signal to noise ~ the strength and continuity of the reflection(s) caused by the geological surface(s) you are trying to map in relation to other reflections and noise + The second biggest factor is how much time you haveSignal to noise + Try not to think about the signal to noise of your data, but the signal to noise of the reflector of interest = a good reflector in a poor data set is easier to pick than a poor reflector in a good data area + Signal to noise depends on — fold of data — the overburden = vintage of acquisition and processing + which affect a whole range of factors such as positioning, source type, etc. — structural style = depositional environment = the choices made by the processors These displays are from the same province, same play, same contractor, similar processing, similar water depth, acquired one year part and they are displayed at exactly the same scale+ Coarse seed grid, easy autotracking and volume picking. + Multiple horizons, isopachs, amplitude extractions, faults easily picked from maps + Ina word - easy, but beware: you may not see wood for the trees + Focus on objectives Low signal to noise @ Fine seed grid, semi-manual picking. Fewer horizons, amplitude extractions may be noise dominated, faults require manual picking In a word - hard work Focus on objectives, but may need to pick strong(er) events to build @ geological model 10@ + Allinterpretation projects should include the following (unless the information can be gained from previous work in the area: 1, The seabed or a near surface reflector 2. A background pick - above your target but below near surface effects. A volume attribute between bounding surfaces may work as well or better 3. Any strong regional reflectors - they supply a framework for your interpretation even if they are not prospective and they should be quick to produce 4, Significant unconformities 5. A deep structural pick - this may just be a form map 6. Basin defining faults + Ideally you should have the above before you pick your target (but most of you won't!) u Work station session ay + AUK North Sea data set has been loaded to an interpretation system + Areview of the project will show — pick intensity/regularity depends on the quality of the pick and what you intend to do with it — it pays to spend some time, in a new area or with a new data set, investigating what works and what doesn’t ~ pick as coarse a seed grid as you dare, then work your way down to something that produces satisfactory results 2és! 100 x 100 lines 50 x 50 lines i : = a vA af GELS Sg sb sp. ]25 x 50 lines ae 45 Good habits i? + When you start interpreting it is good to establish habits that will help you out in the long run + These include: = Using a rigorous and meaningful naming convention — Keeping a real time record of work, such as in OpenJournal, Powerpoint or just Jpegs — Working on standardised base maps — Keeping consistent horizon colours = Exporting and archiving horizons at regular stages in the project + Don’t just copy and rename within the project ~ horizon proliferation is the enemy of the efficient interpreter = Pick seed grids on regular lines (2, 4, 8, 16, 32, 64 or 5, 10, 20, 40) + Do it EVERY time, or you won't do it at alll... 16Naming conventions «“ A good name will be - easy to understand = self explanatory = searchable and sortable (is that a word?) - Memorable A bad name will be — complex = confusing = meaningless to anyone else a) Memorable? @ A former colleague used to name his horizons after songs by the 80s weap Manchester band The Smiths —__ ra Easy for him to remember, meaningless to anyone else 18Some more poor names «@ + Here are some examples of some very poor horizon names that are commonly encountered: = Top Chalk — New top chalk — New new top chalk - TC - Te1 — MJS - 167 = temp1 — Latest = Yellow - Interesting 19 Proposed naming convention + This a very personal thing, but: | Horizon identifier | | | T 0468 TCret TWT Nears Repick June11 MJS + The sort code allows arranging in stratigraphic or depth order rather than alphabetic, if the horizons aren't hard wired into a strat column + Asa whole the horizon name, although complex, is easy to understand + Initials can go at the front to allow sorting by author + In theory the TWT shouldn't be needed in most systems for horizons, but will be most likely required for grids, and it’s a good idea to match grids to the horizon (although unique descriptor may change) + Ina project covering more than one area, include the block or prospect name 20Exploration Geosciences’ convention 800 90.0 = | 99.0 = i, a M255: An integrated approach to 3D Seismic interpretation Voxel picking ( auritosVoxel picking + Googling the words “voxel picking geobody” produces a string of patent applications in the first few results = there is no doubt that interpretation systems are being pushed this way +5 more patents in the next 10 results Voxel picking + So what is a geobody? + In 3D visualisation software a zone of interest is isolated (sculpted) based on seismic properties (possibly combining several 3D volumes), This zone of interest is extracted as a geobody. + The technique is very powerful and can produce results (and results that are instantly useable by engineers), but + Be careful how you use these - once you sculpt you lose the context +The technique is generally only reliable when the body would be relatively easy to pick conventionally Per rane ages by Sticke eet th new ecto exact +5 more patents in the next 10 results 23 ia 24Co-blending etc @ + Displays are generated in ffa's SVI software 35 Hz data extraction The following sequence of slides compares spectral decomposition with coherency and amplitude extraction from a meandering channel system, onshore Eastern Europe 25 a 2645 Hz data extraction 55 Hz data extraction 28Colour blend using 35-45-55 Hz data Semblance/coherency data extraction from 3D 2 29 y ae 30Body labelling from 3D Attribute envelope extraction from 3D ay @ 4 oy Wy) aRMS amplitude 40 ms extraction window And the conclusion is? This slide deliberately left blank! 34Noise +The presence of coherent and incoherent noise complicates the icking procedure — you may need to pick a strong event above or below your reservoir and isopach down (or up) if auto-tracking tools break down you may need to resort to manual picking seed lines, then gridding and back calculating the horizon = try band-pass filters, trace mixes and whatever else your software has to offer but don’t be surprised if you end up hand picking + The following slides use some simple sine waves to demonstrate the effects of noise on wavelet shape Simple 2 interface model synthetic (90° phase) original wavelet C74 35 M 36Adding noise original wavelet noise wavelet Amplitude summing for a synthetic trace original wavelet composite wavelet noise wavelet 38Looking at lower frequency noise And lowerAnd higher Zero phase 4asyWith a 90°wavelet. lef, the zero crossing, with this noise train, stays in much the same place, whilst the second trough is reduced to a side lobe of a noise peak With a zero phase wavelet, right, the peaks/troughs stay aligned, but the second zero crossing has moves significantly and the preceding slides mean? In noisy data, the exact position of the primary reflections is uncertain ~ so if using amplitudes, volume extractions such as RMS may be more reliable.. Knowing the phase of your data (if that is ever really possible in the real world) affects where you pick = even though a zero crossing contains no amplitude information, it may be a more accurate pick Lets look at some real data — rotating the phase can significantly change how the data can be interpreted 47 48Pick comparison in the real world 1 Original data simoath event, | itacs easily ‘bt token of mts Pick comparison in the real world 2 90 degree phase rotated Bath of these picks til ook ak, leone secon BES leaking pk 50scary? + The change in the appearance of the upper event in the preceding slide with just a change in phase is a little disconcerting + It is hoped that the experienced interpreter, picking a series of lines in a data volume would be able to establish the correct relationship. + It is also hoped that an interpreter equipped with a full suite of analysis and visualisation tools would quickly arrive at the solution as well. + But maybe they wouldn't? + Always try to capture and document significant uncertainty Guidelines for picking 1 + As always start with some questions: = What do we want to measure and record when we interpret a particular horizon? — What will the horizon be used for? + Look at the well data first. Decide which picks are required for: = Potential reservoirs, seals and sources = Structural and stratigraphic control = Velo breaks for depth conversion = QC (often neglected) a2 E 51 52ie Guidelines for + Using synthetics (if available) decide which part of the reflection package is the optimal for achieving horizon objectives + First preference ~ pick the largest peak or trough — amplitude available immediately, no snapping or extractions required — likely to be least affected by noise and multiples — even if data known not to be zero phase + Second choice - pick the zero-crossing snap or RMS amplitude extraction required + Factors that affect this choice = signal to noise = event separation = auto-tracker behaviour = visibility + Remember you are not stuck with the phase and polarity you are supplied with! 53 Session learning outcomes ie + After this session the participant will be able to: — Analyse their data set to decide on the most effective picking strategy ~ Practice good habits to enhance data and knowledge management — Assess the value of different ways of extracting information from 3D data sets (surface picking, extracting volume attributes, geobody propagation, etc) 54Session 6-2. Seismic “Noise” Introduction Session learning outcomes 4 After this session the participant will be able to: = recognise “noise” - develop strategies for mitigating the effects of noise1. Define and understand objectives 2. Organise and understand your data 5. Assess implications Paes pein les le eee Objectives of the session @ To look at the things (for want of a better word) in the data that do their best to make unravelling the petroleum geology more difficult and which if not recognised can lead to erroneous results 1. Acquisition noise 2. Processing noise 3. Geological noise2 The Leading Edge (2008), Special section on Seismic Noise 2 = E What is seismic noise? - any recorded energy that interferes with the recorded energy utilised by the application from which the seismic data were acquired (!) From an interpreters point of view: - anything that makes interpreting the primary reflectors difficult and which masks their geological attributes stimulation exercise. = YOUR TURN @ In teams of four - Brainstorm a list of aif the things you have seen that have made interpreting the primary reflectors difficult and which may have masked their geological/geophysical attributes Asa group = Concatenate the list — Vote on which are the biggest problems in 2012 - Each person gets 3 votesSources of seismic n + Ambient or background noise = wind, swell, nearby seismic acquisition, production + Source-generated noise — direct and scattered surface waves, air waves, multiples + Instrument noise A processor is interested in? + Coherent and random noise attenuation using the curvelet transform + Application of multichannel Wiener filters to the suppression of ambient seismic noise in passive seismic arrays +The impact of random noise on seismic wavelet estimation + Twin recording method for basic noise subtraction from seismic land data: Power lines case + Seismic surface waves in a suburban environment: Active and passive interferometric methods + Ground-roll attenuation based on Wiener filtering and benefits of time-frequency imaging + Paradigm shift in 3D multicomponent seismic imaging in the Arabian Gulf + Optimized deghosting of over/under towed-streamer data in the presence of noise + Noise suppression for enhanced imaging: A subsalt case history from the Gulf of Mexico + Two different algorithms for seismic interference noise attenuation + Optimizing the removal of seismic interference noise+ Surely you just need to be able to mentally or programmicaly filter it out? + Noise, once it has been through the processing mill, and is displayed in beautiful colours on a workstation often looks like signal Reconstructing the Earth Spectrum With Inversion Frequency (H2) Figure 1. This illustration shes the econstruction ofthe Earths spectrum by rserooir property inoersion. The extent ofthe Earth spectrum that car be recovered (hue curoe) is Himited bythe fennency ‘ange ie which he seismic signal amplitude (green) is above the noise ‘amipltude (red). The 2er0-frequency information can be estimated from imaging velocities, and te higher-fejuency information ca be bua from the seismic reflection data. There ia gop (purple) in the Seismic information atthe lover frequencies. Obtaining lor-frequency information fs particulary challenging because of increase ise at oto frequencies. Krohn et al 2008, This example is looking at the ends of the frequency spectrum ‘Towards the limits, the signal to noise ratio falls Any processing or attribute work which seeks to boost the data at its limits runs the risk of boosting noise as well (or instead) 10exploration eosciences Acquisition footprint ie + Any acquisition or processing technique that results in fold of stack to vary between bins will produce an acquisition footprint — at shallow depths fold is low, so bin to bin variations will be significant + this is why seabed amplitude maps in shallow water frequently show a significant footprint — assimilar problem can affect far and ultra-far angle stacks ~ hence these should be used with care 2Any 3-0 seismic survey can have an acquisition footprint. Our prablem is to determine whether we have one ‘and if so, whether we can recognize it, how severe it is and, most importantly, what we can do about it. What is an acquisition footprint? Tt is an expression of the surface geometry (most common on land data) that leaves an imprint on the stack of our 3D seismic data, Often we recognise it as amplitude and phase variations on time slices, which of course display the ‘amplitudes within our data set at a specified two-way time, More seriously, on horizon slices, footprint can interfere with and confuse stratigraphic patterns Many different contributions to the generation of acquisition footprint are possible: Line spacings. Fold variations. Wide vs. narrow patch geometry. Source generated noise. ‘Topography. Culture. Weather. Surface conditions. Processing artefacts. These can be divided into two main categories of geometry effects (above the line) and non-geometry effects (below the line). Geometry Effects Most of the time the acquisition footprint is based on source and receiver line spacings and orientations; the larger the line spacing, the more severe the footprint In land situations where access is very open and, therefore, the lines are very regularly spaced, we may be able to recognize the footprint very clearly. Since the geometry is regular the footprint also will have the same periodicity. Fold variations themselves are the simplest form of an acquisition footprint, Fold changes with offset (or rather mute distance from the source point); each offset range therefore has differing fold contributions. 3 Since each individual bin of a 3D survey has changing offset distributions, the CMP stack of all traces in a bin will display bin-to-bin amplitude variations. This variation in itself can produce an acquisition footprint (figure D. Some processors compensate for this with simple trace borrowing from surrounding bins to fil in the missing offsets and to provide smooth offset distributions in all bins. Although this may be successful in reducing the footprint, it also may reduce the resolution by degrading the high frequency content. Generally it has been thought that acquisition footprint is far worse in the shallow part of the seismic and therefore, of course, the geological section, mainly because the fold is lower and therefore amplitude variations are far more dramatic. Offset limited fold variations alone may produce a recognizable footprint. ‘The higher the fold, the better the signal-to-noise ratio and therefore less footprint is evident, Wide recording patch geometries are far more accepted these days than narrow patch geometries. The reasons are numerous and range from reduction in acquisition footprint (particularly that due to back scattered shot noise), to improved statics solutions and the availabilty of large channel capacities on seismic recording crews (also leading to higher fold). In adaition to the impact of the fold variations, acquisition footprints are made worse by source generated Noise trains that penetrate our data sets (figure 2). The lower the signal-to-noise ratio Is, the worse the footprint will be. Unfortunately, the noise typically has a low frequency content that is much less affected by attenuation, ‘Therefore the noise becomes more prominent relative to the signal content deeper in the section. Our experiences have shown that acquisition footprint problems can be just as prevalent in the deep section as they are up shallower. Non-Geometry Effects If surface access is poor because of topography variations, tree cover, towns, etc., we irregularise the ‘geometry by moving source points to locations of easier access, and therefore mask the acquisition footprint. {tis stil present, however. The footprint is just so much harder to identity. Weather and surface conditions may also impact the recorded amplitudes. A swamp in the middle of a 30 survey can have a significant impact on the amplitude of a stack volume, The interpreter reviewing the 3D volume has to decide what is a geological anomaly and what is acquisition footprint. Not always an easy task! 14vvutninmtnrnincnnn Modeling We can model an acquisition footprint by creating a stack response on either synthetic or real data. Starting out with a geological model of the sub-surface, any source-generated noise can and should be included if the noise velocities and frequencies are well known. ‘We will stack the data in a 3D cube (figure 3) and display the resulting seismic data over a small time window, e-g. 60 ms (Figure 4, depending on the frequency of the data). This process can be repeated using real selemic data as an input, The best input is a single NMO and static corrected, offset-sorted 2D (or 3D) CMP gather, These traces will be applied to each CMP bin in the recording geometry. The correct offsets for each bin are then stacked in that bin to create NMO-corrected CMP gathers and the time interval of interest studied. This process Is repeated for any acquisition geometry under consideration for the recording of the seismic data ‘The geometry with the least variation in this modelled stack response (acquisition footprint) should be chosen, Processing artefacts also can leave an imprint on stacked seismic date, for example, by applying wrong NMO Yelocities. Choosing incorrect velocities will Ieave remnant moveout on the horizons that should have been stacked as flat data ‘This affects the primaries as well as multiples and source generated noise; now possibly all of them leave Undesirable amplitude and phase distortions in our data. Conclusions Interpreters have lived with footprint since the advent of 3-D. In order to advance our interpretations further, we need to understand and recognize footprint, and make every effort to distinguish it from geology. ‘Acquisition footprint has many different sources. It should be minimized as much as possible, preferably at the recording stage. Therefore one should always model the acquisition footprint for different recording geometries under consideration. Generally, wider acquisition patches are better. Increasing the fold will help reduce the footprint. Moving source points (and receiver stations) in the field produces an irregular acquisition geometry and therefore the footprint may not be as severe. Removal of an acquisition footprint is possible to some degree in the sophisticated selsmic date processing centres, but is not performed on a routine basis. 15 Acquisition Footprint ie + An acquisition footprint is an expression of the survey geometry in the final stack + Causes include — variation in fold (often seen in the shallow section where fold is low) = variations in streamer feathering water depth and/or source or cable depth variation weather, wind and waves tcanstoorer 200. oot sequen ott st 16