Maximizing the value of 3D seismic vintage through state of art land ... · A state of art land...

Maximizing the value of 3D seismic vintage through state of art land depth imaging and

quantitative quality controls - A case study at Yucal Placer field, Venezuela Yann Montico, Jean Patrick Mascomère; TOTAL Exploration & Production

Xavier Duwattez, Milad Bader; CGG

Summary

The paper presents a land seismic depth imaging case history during which an advanced TTI PSDM velocity model building approach was applied on data acquired in 2003. The method includes an attempt to compensate for the lack of fold coverage in the shallowest part of the gathers along which seismic events cannot be used in reflection tomography. A state of art land time processing, including the building of an accurate near surface statics model, was applied to generate the input data to a TTI PSDM workflow. Then a detailed near surface velocity model derived from refraction tomography was combined with deeper RMS velocity to create an initial PSDM hybrid model. The depth imaging results, obtained from true topography, were validated through quantitative quality controls to ensure that PSDM angle substack volumes were reliable for future reservoir characterization. The study area, located onshore Venezuela, presents a relatively mild surface topography but has a complex subsurface geology which includes large lateral velocity variations induced by a NW-SE oriented thrust lying right above the main producing reservoirs. The applied workflow provided a better image quality against the legacy processing. It was seen by the operator as a last-chance to maximize the value of a twelve years old vintage data-set and in a lower oil price context it should eventually have a major impact on the field static reservoir modeling update. Geological setting and seismic data description

The study area is centered on the Yucal Placer gas field located onshore Venezuela in the South-East of Caracas along the northern edge of the Eastern Guarico foreland sub-basin (Figure 1a). It is limited to the North by the thrust of the Northern Mountain System present along the Caribbean coast and to the South by sandstones reservoir pinch-outs. The main reservoirs belong to the autochthonous tight sandstones of the Roblecito and La Pascua formations occurring at depths of 1900 to 2600m below Mean Sea Level. The geological formations above these reservoirs have been strongly structured by the regional tectonic activity inducing massive Andean stacked thrust sheets in the shallowest part of the field (Figure 1b). The reprocessed 3D survey was acquired in 2003 and covers 330 km2 with narrow azimuth distribution and nominal fold of ranging from 56 to 64. Surface topography

varies from 190 to 320m above MSL (figure 1c). Explosive seismic sources were used every 56m, source and receiver lines separation distance was 300m. The entire survey was laid-out in pseudo-orthogonal slanted mode with about 63° angle between the source and receiver lines (Figure 1d). This resulted in a nominal bin size of 12.5x25m and a largest maximum offsets of 4350m.

Figure 1: a) Location map of the Yucal Placer area, b) schematic description of the local geology and different reservoirs, c) surface topography d) fold map of the seismic survey (330km2).

Introduction

In complex structure imaging problems, near-surface static corrections often play a critical role. Various approaches have been tested to produce deep targets depth imaging from true surface topography using the most accurate near surface velocity information (Rajasekaran & McMehan 1995; Hu & Kim 2002,). Several authors have also proposed static tomography as an efficient way to incorporate accurate near surface velocity heterogeneities to enhance depth imaging results of deep targets (Zhang et al. 2005, Tanis at al 2006, Han et al. 2014, Ji et. al 2015). The proposed workflow applied similar near surface velocity model techniques in combination with state of art land processing, reflection tomography and TTI PSDM. The objective was to obtain the best pre-stack primary amplitudes of deep reservoirs covered by complex geology.

Maximizing the value of 3D seismic vintage through state of art land depth imaging and quality controls

Applied reprocessing workflow

Time processing The reprocessing main goal was to deliver a depth volume and strictly amplitude preserved Angle sub-stack volumes for seismic elastic inversion, reservoir porosity and Poisson ratio prediction. A state of art land time-processing sequence was tested to obtain the best input data to the depth imaging workflow. First breaks picking was performed up to 2700m offset on conditioned shot gathers. Refraction statics Linear Model Inversion was applied including the velocity information of 36 up-holes. The resulting weathering model had two layers: the velocity of the first refractor was around 890m/s (+/-180m/s) for a thickness varying from 6 to 35m. The velocity of the second refractor was around 2790m/s (+/- 300m/s) for a thickness varying from 14 to 270m. Total statics values ranged from 170 to 260ms. The reference Flat Datum was the Mean Sea Level and a replacement velocity of 2500 m/sec was used. Along the sequence, shot gathers underwent several noise attenuation steps to attenuate spikes, ground roll, guided waves, scattered and random noise. The surface topography is incised by several local rios causing large amplitudes variations from shot to shot. Therefore simultaneous joint inversion for surface consistent amplitude and deconvolution was computed on shot gathers with strong noise attenuation conditioning (Garceran et Le Meur, 2012). Operators and scalar were applied thereafter on mildly de-noised shots. A double iteration of velocity picking and residual reflection statics was done followed by extra passes of surface consistent amplitude corrections. After five dimension regularization the data were sorted in 65 Common Offset Vectors and extra noise attenuation was applied to further enhance the weakest primary signal at targets. Output gathers were on a 12.5m square bin. Depth processing Preconditioned regularized gathers were referenced to surface topography, the PSDM hybrid velocity model building was performed in the three following phases using Kirchhoff migrations from true surface Topography. Phase 1, called near surface model building differed from the conventional path to land PSDM during which all statics are kept in the input gathers and the depth converted RMS velocity field is directly used as the initial PSDM model (Figure 2). On the one hand, this phase consisted in removing from the input data, the primary Long/Medium wavelengths refraction (LW) and elevation statics while only the short wavelengths residual sources and receivers

statics (SW) remained applied to the gathers traces. On the other hand, a near surface velocity model was derived from the gridded refraction static model and traveltime diving-wave tomography (ref. Tanis et al. 2006). Figure 3 compares both initial and final Near Surface inverted velocity fields. This accurate model was taking into account the details of the weathering layer. It was smoothed and vertically merged with a depth-converted and filtered Dix version of an existing PSTM velocity field in order to create the initial hybrid PSDM velocity model. The trickiest step was to spatially determine the best merging location between both models around the Ray Penetration Base (RPB). In practice several tapering, scaling and smoothing attempts were required and quality control was done checking the shallow events continuity. Figure 4 depicts the impact of the method on the shallow depth imaging quality by comparing a PSDM stack and gathers using the conventional and advanced models.

Figure 2: Comparison between conventional and advanced PSDM initial model building. Phase 2 consisted in a single PSDM isotropic model building iteration using Kirchhoff migration on a 50x50m sparse grid. High density volumetric dips and RMO curvatures were measured on CIP gathers and stack sections. Velocity model update was obtained using kinematic invariant and non-linear slope tomography (Guillaume et al. 2008). Despite a high preconditioning effort, due to weak shallow pre-stack quality (figure 4), RMO and dip events picking remained quite challenging down to roughly 1km depth, this unavoidably lead to a weakly constrained tomography.

Phase 3 included two iterations of tilted transverse isotropic PSDM model building. Initial anisotropy fields Delta (δ) and Epsilon (ε) were built using well markers, geological horizons as guides. Specific iterative efforts aimed at keeping δ and ε fields structurally conformable, simple and spatially following a Gaussian shape. New


kinematic invariants were recomputed, velocity and epsilon were jointly updated.

Figure 3: a) Input: linearized inversion near surface velocity model b) Output: refraction tomography model c) perturbation (input – output) following the local geology.

Fifteen different velocity models were tested, focus was on getting a good balance between the depth accuracy, image quality at targets and matching to reliable wells velocity profiles. The depth control was done through an analytical calibration workflow using markers misfits of ten vertical wells. Very large δ and ε values had to be used to honour the well markers indicating that the area is highly anisotropic. Geophysically and geologically meaningful TTI PSDM Vp, Dips, δ and ε fields were used in the final Kirchhoff migration (input/output bin 12.5x12.5m, steps 5m).

Figure 4: a) PSDM stack and gathers using conventional initial model b) Same stack migrated with advanced initial hybrid velocity model. Note the weak quality of prestack event for model building, noticable primary improvement around the black arrows.

Quantitative Quality Controls

As our primary objective was to deliver amplitude preserved Angle Substack volumes, data processing was monitored using a procedure of Quantitative Quality Controls, QQC (Paternoster et al. 2009, Araman et al. 2014). For each substack and for every bin, up to ten quantifiable seismic attributes were spatially computed along time windows centered on the top reservoir P7 featuring a robust seismic reflector everywhere in the volume. In addition, well-to-seismic tie quality attributes such as phase shift, wavelet amplitudes and correlation coefficients were monitored at six wells. Critical steps such as deconvolution, statics and de-noising, regularization, migration and Post-RMO processing were selected to perform up to eight QQC’s along the project. It was to validate that the reprocessing sequence enhanced the seismic primary signal at targets on each substack volume. The Figure 5 illustrates the QQC outcomes on a four dimensions spider plot. It quantifies the improvements from pre-migration to intermediate PSTM and finally PSDM stages. The reference at 100% being the processing stage 1 (i.e. prior Deconvolution). An increase in percentage means an increase in quality. Each dimension of the spider plot is given as percentage and results from the weighted average of several specific attributes (e.g. resolution, SNR, bandwidth, Near-to-Far correlation, phase & time shift).

Figure 5: Four dimensions spider-plot summarizing QQC results along the processing flow, units are all in Percentage . Particular attention concentrated on the sector located below the thrust zone where depth imaging was expected to have the largest impact. Figure 6 shows the correlation map between Final RNMO corrected Near and Far angle sub-stacks computed 300msec around the Top P7 (yellow arrow around blue horizon). The analyzed data did not have specific alignment but were filtered in a common


bandwidth. TTI PSDM average correlation is clearly superior to PSTM and also much more stable laterally. This attribute, among others, is indicative of pre-stack data usability in elastic inversion. It confirms that in such context TTI PSDM is the best choice to go toward reservoir characterization. The black dashed line materializes the limit of the thrust zone. Kirchhoff PSTM and PSDM Full stack sections right below are respectively shown in Time and Depth domains and clearly illustrate the impact of sub-thrust depth imaging at target.

Figure 6: a) Final Kirchhoff PSTM Near-Far correlation map, b) same map for Kirchhoff TTI PSDM c) PSTM section in TWT d) PSDM section in depth domain. Note, fake fault (black arrow) is no longer present on PSDM image.

Results

The new TTI PSDM results were compared against the legacy volume, a full stack POSTM of 2003, to confirm the overall benefit of the project (see figure 7): 1- In the overburden the complex tectonic features have

been dramatically improved and the continuity of the shallowest reflectors is also superior on PSDM.

2- Signal to noise ratio, frequency bandwidth and reservoir continuity have been enhanced. Coherency, resolution and RMS energy top reservoir maps have superior lateral stability on the PSDM volume.

3- Most of the fake large structural undulations present in the POSTM and PSTM images at the reservoir level are attenuated on the PSDM image, the velocity model (figure 7b) show that tomography was able to detect lateral velocity variations in the overburden.

4- Residual depth misties at wells were minimized using in-house workflow prior to reservoir interpretation.

Conclusions

This case study presented an advanced onshore TTI PSDM hybrid model building approach. It should be reminded that depth imaging took full advantage of a detailed time processing work. This state of art amplitude preserved workflow allowed a considerable improvement in structural image and deep reservoir modeling. Finally the QQC monitoring of data quality along the project is a guarantee that the best pre-stack data are handed-over to the reservoir characterization teams. Acknowledgments

We acknowledge the management of Total, Repsol, Inepetrol, Otepi and CGG for permission to publish this work. We also thank in particular Jean-Luc Piazza and Alain Riou for their contribution on this project.

Figure 7: Random section across the field a) of the legacy POSTM processing AGC applied (in TWT, ref. to MSL) b) of the final TTI PSDM in depth domain with the PSDM velocity field overlaid.

EDITED REFERENCES

REFERENCES

Araman A., Paternoster B., 2014, Seismic quality monitoring during processing, First Break, vol.32, Sept. p 69-78. Garceran, K., and D. Le Meur, 2012, Simultaneous joint inversion for surface consistent amplitude and deconvolution: 74th Conference & Exhibition, EAGE, Extended Abstracts, C015. Guillaume, P., G. Lambaré, O. Leblanc, P. Mitouard, J. Le Moigne, J.-P. Montel, T. Prescott, R. Siliqi, N. Vidal, X. Zhang, S. Zimine, 2008, Kinematic invariants: an efficient and flexible approach for velocity model building: 78th Annual International Meeting, SEG, Advanced velocity model building techniques for depth imaging workshop. Han X., Wang D., Yang S., Hinz E.C, Song M., Liu X., 2014, Combining diving wave tomography and pre-stack reflection tomography for complex depth imaging – A case study from mountainous west China. SEG International Exhibition and 84th Annual Meeting, Expanded Abstracts. Hu L. and Kim J.J., 2002, Imaging complex shallow structures by pre-stack depth migration. SEG International Exhibition and 72nd Annual Meeting, Expanded Abstracts. Ji X. et al., 2015, Imaging complex geology through challenging surface terrain – a case study from West China. First Break, vol. 33, March, p 55-69. Paternoster B., Lys P.O., Crouzy E. and Pagliccia B., 2009, Seismic quality monitoring during processing for reservoir characterization. SEG International Exhibition and 79th Annual Meeting, Expanded Abstracts. Rajasekaran S. and McMechan G.A., 1995, Pre-stack processing of land data with complex topography. Geophysics, 60, p.1875-1886. Tanis M.C., Shah. H., Watson P.A., Harrison M., Yang S., Lu L., Carvill C. 2006, Diving-wave refraction tomography and reflection tomography for velocity model building. SEG International Exhibition and 76th Annual Meeting, Expanded Abstracts. Zhang J. and Oz Yilmaz, 2005, Near-Surface corrections for complex structure imaging. SEG International Exhibition and 75th Annual Meeting, Expanded Abstracts.

MAXIMIZING THE VALUE OF 3D SEISMIC VINTAGE THROUGH STATE OF ART LAND DEPTH IMAGING AND QUANTITATIVE QUALITY CONTROLS: A CASE STUDY AT YUCAL PLACER FIELD, VENEZUELA

Yann Montico, Jean Patrick Mascomère (Total Exploration Production) Xavier Duwattez, Milad Bader (CGG) NEW ORLEANS, LA OCTOBER 19, 2015

1 SEG 85th Annual Meeting New-Orleans 2015

TALK OUTLINE

• Motivations and challenges

• Time and depth processing strategy

• Results and quality controls

• Final thoughts


YUCAL PLACER GEOLOGICAL FRAME

100 km

A

A’

CARACAS

YUCAL PLACER KEY FACTS

Dry gas field discovered in 1946 Cumulated production ~ 500BCF Reservoirs: autochthonous Oligo-Miocene tight sandstone

North

CARIBBEAN SEA

SOUTH AMERICAN PLATE


VENEZUELA

2000

m/M

SL

0

70

FOLD MAP ~330 KM2

Yucal Placer Guarico Basin,

+320m

+200m

NARROW AZIMUTH ACTIVE SPREAD

20 KM2

4090m

TOPOGRAPHY

3D SURVEY ACQUISITION PARAMETERS


TWT

(ms)

FULL POSTM 2004

LIMITATIONS WITH 2004 LEGACY PROCESSING

• High noise level • Narrow frequency bandwidth signal

Primary signal too

narrow-band and

high frequency noise


Artificial

Push-downs

• False structures at target

Strong lateral variability in quality at targets (highly unstable amplitudes, phase and frequency)

• Misleading amplitudes

KIRCHHOFF PSTM

COV SORTING AND 5D REGULARIZATION

3D RADON DE-MULTIPLE

JOINT DECONVOLUTION & AMPLITUDE COMPENSATION

NOISE ATTENUATION

SP WITH UPDATED GEOMETRY

RESIDUAL MOVE-OUT

NOISE ATTENUATION

RESIDUAL STATICS &

VELOCITY PICKING

RESIDUAL SCAC

POST-PROCESSING

Pre-decon.

Post-decon.

Post-denoise

Post-Regul.

Post-PSTM

FINAL PSTM

REFRACTION STATICS CALCULATION

FIRST BREAK PICKING + UPHOLE

LINEAR MULTI-LAYER INVERSION (LMI)

PRIMARY STATICS

MODEL M0 (REF. TOPOGRAPHY)

TIME RMS VELOCITY

Conversion to depth & spatial smoothing

Regularized gathers with LW & SW statics applied

(ref. topography)

INITIAL KIRCHHOFF

PSDM

MODEL M1 (REF. TOPOGRAPHY)

REFRACTION TOMOGRAPHY

NEAR SURFACE MODEL IN DEPTH

Vertical merge with tapering

& scaling at RPB

Regularized gathers with only SW statics applied

(ref. topography)

PSDM INITIAL VELOCITY MODEL BUILDING FLOW


CONVENTIONAL APPROACH ADVANCED APPROACH

Total Statics = Se + LW + SW LW = Total Statics - Se - SW

Se = Elevation statics = LW = long wavelength statics = derived by Linear Multi-layer Inversion (LMI)

recoverable by velocity model tomography SW = short wavelength statics = residual statics derived with stochastic method

non recoverable by velocity model tomography

Elevation

Replacement Velocity

LONG WAVELENGTH STATICS DEFINITION


-4ms

+20ms

LW statics map N

LMI INPUT MODEL

-500m/sec

+500m/sec

REFRACTION TOMOGRAPHY OUTPUT MODEL

DIFFERENCE

Dep

th (

m/D

PA)

4000m/sec

PSDM SHALLOW MODEL BUILDING


TOPOGRAPHY 1500m/sec

400m depth slices comparison

SE NW

IMPACT OF ADVANCED INITIAL VELOCITY MODEL

CONVENTIONAL INITIAL PSDM MODEL USED ADVANCED INITIAL PSDM MODEL USED


ADVANCED INITIAL PSDM MODEL BUILDING

PSDM VELOCITY MODEL BUILDING FLOW


ITERATION 1 - ISOTROPIC PSDM MODEL BUILDING

INITIAL PSDM

RMO & DIP PICKING

KINEMATIC De-migration

KINEMATIC INVARIANTS

Kinematic Migration, Predicted RMO

RMO MINIMIZED ? YES NO

New Inversion Velocity Update

Non-linear Reflection

Tomographic Update Isotropic

Velocity model

ITERATION 2 - TTI PSDM MODEL BUILDING Non-linear slope reflection tomography joint-inversion V, δ, ε

TTI V, δ, ε models

ITERATION 3 - TTI PSDM MODEL BUILDING Analytical well calibration

TTI V, δ, ε models Well-tie at

Geological markers

TALK OUTLINE

• Motivations and challenges

• Time and depth processing strategy

• Results and quality controls

• Final thoughts


RAW PSDM CIG MIGRATED WITH INITIAL VELOCITY MODEL

No RMO and conditioning applied

RAW PSDM CIG MIGRATED WITH FINAL VELOCITY MODEL

No RMO and conditioning applied

TARGET RESERVOIRS


FINAL MIGRATED GATHERS BELOW THRUST ZONE

CIP GATHER BELOW THRUST

SEG 85th Annual Meeting New-Orleans 2015 13

Low

High

FINAL ANISOTROPY and VELOCITY FIELDS

Low

High

DELTA δ

2700

4300

VELO

CIT

Y (m

/sec

)

EPSILON ε

VELOCITY

TOPOGRAPHY

Km

/M

SL


TVD WELL A

(THRUST ZONE) WELL B

(FLAT ZONE) TVD

PSDM velocity

VSP velocity

SONIC velocity

WELLS CONTROL AND GATHERS COMPARISON FINAL

PSTM GATHER FINAL

PSDM GATHER FINAL

PSTM GATHER FINAL

PSDM GATHER

TWT

(mse

c re

f. da

tum

)


LEGACY 2004 POSTM - 2015 TTI PSDM

TIM

E (m

sec)

Well A Well B Well C Well D

+

_


LEGACY 2004 POSTM - 2015 TTI PSDM Well A Well B Well C Well D

DEP

TH (m

)

+

_

TOPOGRAPHY


FINAL FULL STACK COMPARISON 2004 LEGACY POSTM 2014 KIRCHHOFF PSTM 2015 KIRCHHOFF TTI PSDM

TIME TIME DEPTH

+

_


RMS AMPLITUDE MAP AT TOP RESERVOIR

LEGACY POSTM PROCESSING NEW TTI PSDM PROCESSING


NEAR SUB-STACK

2014 PSTM

2015 PSDM

FULL STACK

12ms 19ms

N

SEISMIC RESOLUTION ANALYSIS

FAR SUB-STACK

15ms 27ms


NEAR NEAR

MID MID

FAR FAR

PSTM 2014 PSDM 2015

95%

10%

75%

NEAR-FAR Cross-correlation

95%

10%

NEAR-MID Cross-correlation

FINAL PSTM 2014 FINAL PSDM 2015

A

A’

89%

ANGLE STACK CORRELATION QUALITY CONTROL


QUANTITATIVE QUALITY CONTROL RESULTS

FREQUENCY BANDWIDTH

SIGNAL/NOISE RATIO

LATERAL CONSISTENCY

WELL CALIBRATION

PHASE STABILITY

LEGACY 2004 PSDM 2015

SIGNAL CONSISTENCY

PRE-STACK CONSISTENCY

LATERAL CONSISTENCY

PSTM 2014 PSDM 2015

PRIOR DECON

WELL CALIBRATION

• Accurate near surface velocity field derived from refraction tomography is

a good starting point in such land depth imaging context


FINAL THOUGHTS

• The TTI PSDM workflow has improved sub-thrust image quality at target - Much more realistic structures

- Enhanced Signal / Noise Ratio

- Improved pre-stack quality

- Better stability of primary signal amplitude, phase and frequency

• Quantitative quality control is a powerful tool to qualify PSDM data for future

reservoir characterization studies


We acknowledge the management of Total, Repsol, Inepetrol, Otepi and CGG

for permission to publish this work.

AKNOWLEDGMENT

Maximizing the value of 3D seismic vintage through state of art land ... · A state of art land...

Documents

Transcript of Maximizing the value of 3D seismic vintage through state of art land ... · A state of art land...