Realising the Benefits of Long Offset Data: A Case History from the Utgard High Area Richard Morgan, Richard Wombell, Dave Went Veritas DGC Ltd, Crawley

Location 2D line & two well tie

LONG OFFSET ACQUISITION Objectives: Improve imaging through use of longer offsets Wide Angle AVO analysis Challenges: Severe multiple contamination Accurate imaging of long offsets Angles up to 45 degrees High order moveout Accurate amplitude preservation The primary aims of the acquisition program was to improve the ability to image the targets of interest through the use of longer offsets along with the potential use of wide angle AVO analysis, I.e. the use of angles greater than 30 degrees. Realising this in practice required dealing with a number of issues. First on the list was the significant multiple content of the data, which had to be attenuated without incurring primary damage, particularly at the longer offsets. It was also necessary to be able to accurately image the longer offsets. As will be seen in the following slides, the sequence used a combination of surface related and Radon multiple attenuation along with a ray-traced Kirchhoff pre-stack time migration for imaging.

Surface Related Multiples Raw Data Surface Related Multiples This slide shows the raw data as acquired. On the right we see a limited angle stack, using angles from 15 to 30 degrees and on the left three gathers with NMO applied as for the stack. The location of the three gathers are indicated by the white markers on the stack section. The first step in the sequence is to attack the surface related multiples. We can see that these are very strong, swamping the underlying primary data in the pre-stack data and showing strong leakage through the stack. CMP gathers 15o-30o stack

Surface Related Multiple Attenuation Linear Noise contaminating Far-offsets This was done using a 2D surface related multiple attenuation. Here the raw data are used to predict the multiples in the data in a 2D sense and the model of the multiples are then adaptive subtracted from the data. Toggling back to the input sections, we see that the multiple is significantly attenuated and we can now start to see the underlying primary in the gathers. However, the multiple is not fully removed, so some additional de-multiple will be required later in the sequence. We see that there is some linear noise contamination of the longer offsets, most obviously at shallower times. The next stage was to attenuate this using a high resolution linear noise attenuation algorithm. CMP gathers 15o-30o stack

Linear Noise Attenuation Residual Multiple And as required the linear noise attenuation suppresses these events. As we noted before, although the surface related multiple attenuation removes a lot of the multiple, there is still significant remnant. We attack this is in two stages, a mild Radon de-multiple before migration and a final pass of Radon after migration where it can be driven by a more accurate velocity field and used attenuate the faster multiples. CMP gathers 15o-30o stack

Radon Multiple Attenuation Higher order moveout And these are the data after applying a targeted version of Radon de-multiple. We have now significantly attenuated the multiple, while still preserving the primary energy. The next stage in the processing is the imaging. CMP gathers 15o-30o stack

Ray-traced Kirchhoff Pre-STM Ray-tracing automatically incorporates 4th and higher order moveout terms Faster residual multiple For the imaging, we use a ray-traced Kirchhoff pre-stack time migration. The ray tracing allows the migration to automatically incorporate all the v(z) higher order moveout terms, not just 4th order. We see that the gathers have now been we imaged, but some residual multiple, not attenuated by the first pass of Radon, still remains. This is attacked by a second pass of Radon. CMP gathers 15o-30o stack

Post-Migration Radon De-multiple And this slide shows the final processed gathers. CMP gathers 15o-30o stack

Post-Migration Radon De-multiple A comparison of a 30 degree and a 45 degree stacking mute. Use of the longer offsets significantly improves the signal continuity and the signal-noise ratio. 15o-45o stack 15o-30o stack

Realising the Benefits of Long Offset Data: A Case History from the Utgard High Area Richard Morgan, Richard Wombell, Dave Went Veritas DGC Ltd, Crawley

Example Cross Line N S Superior fault definition in Cretaceous section revealing tilted fault blocks. Deep reflection events revealed (Jurassic?). Intrusives cut across bedding but exploit fault planes.

Nyk High 6707/10-1 well logs Poisson’s ratio Gamma Ray Acoustic impedance

VP/VS vs DepthBSB: by VCL Rock physics basis ACOUSTIC IMPEDANCE POISSON’S RATIO AI vs DepthBSB: by Vcl VP/VS vs DepthBSB: by VCL Brine sand Shale This slide shows the physical basis for my assertion that lithology can be detected using AVO On the left we have a plot showing acoustic impedance versus depth coded by lithology – sand is in red shale is in blue – we can see AI increases with depth steadily but that sands are sometimes harder – sometimes softer than shales there is no consistent pattern that could be used for lithology detection On the right we have a plot of Vp/Vs vs depth – same colour coding – now sands are consistently to the left of shales Ie for a given depth sands show a lower Vp/Vs than shales poor lithology discrimination much better lithology (& fluid) discrimination From: Seismic Rock Properties, Judd Basin, WOSI. Non- proprietary study Data from logs – multiple wells

Improved hydrocarbon detection G + R0 + - I NB: fluid anomalies can be still be prescribed but must be outboard of the brine sand trend – NOT just the mudrock line IIp II Brine sand line Mudrock line Hence with this knowledge of litholgy effect in mind we in mind we can look for fluid anomalies that are outboard of the clean brine sand line – not just the mudrock line Of course it can work the other way - some gas filled silts or muddy sands might be buried in the clean brine line III IV YOU CANT BE PRESCRIPTIVE ABOUT FLUID WITHOUT KNOWLEDGE OF THE LITHOLOGY!

Theory - Poisson’s reflectivity AI Forward model - half space plot PR R0 R90 0.2 Vp/Vs Shale/sand low Vp/Vs Shale/shale high Vp/Vs Sand softer than shale Sand harder than shale Shale Shale Sand Sand This slide shows a forward reflectivity model summarising the key issues shown on the log data scatter plots but for a single half space - i.e. Sands can be harder or softer than shales, hence give +ve or –ve R0 but are invariably softer than shales at R90 – brine (or gas) Shale on shale contacts may show significant R0 (due to different compressibility) – but negligible R90 due to similar PR -0.2 Sin2Ø 1 Rc0= AI/AIav Rc90  PR/PRav Ø

Seismic facies or lithology interpretation Calibrated thin resolvable sand thin resolvable shale thick shale thicker sand shale sand shale thick shale or thick sand or thinly interbedded sand and shale We need a facies / lithology detection scheme Thin but resolvable shales are blue, thin but resolvable sands are yellow (upper pair of pictures) Thicker intervals of unbroken shale or sand (or unresolvable beds of sands and shale) are expected to be red-brown dominated in the central parts – (bottom right) Shales are expected to have kick to blue and out via a blue where bounded sharply by sand (bottom left) Gradational contacts will modify motif according to the above constraints

Seismic stack with 6707/10-1 well AI GWC GWC Gas

P-impedance with 6707/10-1 well AI GWC GWC Gas

Vp/Vs with 6707/10-1 well GR GWC GWC Gas

P-impedance with well AI Weak near offset reflectivity Gas

S-impedance with well AI Stronger shear impedance response

Vp/Vs with well GR Strong Vp/Vs response with opposite polarity to Is = sand Strong far offset gas response Stronger far offset contrasts in Vp/Vs: sand-shale

Utgard 2D Facies Finder inversion 10-40 degrees 2 term AVO partition from 6Km data Good correlation with well lithologies Interwell areas also show AVO Blue-yellow-blue sand signatures are present suggesting the presence of sand deposits The potential exists for delimiting sand fairways if applied to 3D Strong near offset impedance with decreasing energy with offset correlate with and suggest the presence of other intrusions

Acknowledgements Veritas would like to thank Norsk Agip and Gaz de France for supporting the Utgard Long Offset 3D programme