Structural interpretation and faults mapping

1
win-win partnership STRUCTURAL INTERPRETATION AND FAULTS MAPPING USING SEISMIC ATTRIBUTES AND NEURAL NETWORK OF A 3D POST-STACK SEISMIC SURVEY IN KABOUDIA PERMIT General setting Fault Enhancement Workflow Applying The Filters & QC The Kaboudia permit lies in eastern offshore Tunisia and covers an area of about 3104 km². It is bounded to the West by the Monastir-Mahdia shoreline and to the North east by the Halk el Menzel Oil Field, the water depth is generally less than 200 meter. The Kaboudia block is located in the Pelagian platform in the NW-SE “Mahdia-Isis” paleohigh which separate the Gabes basin from the Hammamet basin. It lies in a highly complex structural settings since Late Triassic to Early Cretaceous period which is marked by extensive cycle and is relayed by Early Albian to Early Quaternary compressive cycles which are mainly related to the subsidence and collision between African and European plates. The compressive cycle is recognized by four major tectonic unconformities related to Early Albian-Austrian phase (top Serj unconformity), to Late Cretaceous-Eocene Pyrenean phase (top Souar-base Fortuna unconformity), to the Middle-Late Miocene Atlassic phase (top Fortuna-base Ain Grab unconformity) and to the Early Quaternary phase (base Pliocene-base RafRaf unconformity). These structural activities were associated with significant movements of pre-existing faults and folds leading to the generation of many local unconformities, structural inversion, erosion and lateral variation of facies of different stratigraphic series. We set out to develop a seismic-fault detection method which recombines multiple attributes into a new attribute that gives the optimal view of the targeted object .The seismic volume has been subject to several stages of pre-conditionning to enhance discontinuities and calculate similarity and curvature attributes These final attributes show detailed geometry of the major fault system and numerous subtle lineaments in the study area. METHODOLOGY: Seismic attributes can provide essential details from seismic data, but noise present in the data- set often distort the outcome. To optimally remove noise, data conditioning was carried out in two steps, in the first step, a dip volume was created using Fourier transform based algorithm in inline and cross-line directions. This dip volume was then used as a reference, and a median filter was applied on the seismic data guided by prepared dip volume. It reduces the structurally oriented random and coherent noise and increases the continuity and visibility of faults and fractures by preserving and at the same time sharpening the edges. Definitions: Similarity is a coherence attribute that expresses how similar two or more traces are to each other either in the crossline or inline directions Trace discontinuities of the seismic data may be the result of the faults or stratigraphic features. In the other hand, curvature attributes are being increasingly used to characterize faults and fractures, based on the fact that certain areas with curved surfaces are related to discontinuity zones, which can be represented by faults and fractures. The curvature attribute emphasizes the positive (antiforms) and negative (sinforms) curvatures, whereas on a flat surface or in inflection zones the curvature is zero. Inline 1249: Fault Enhanced Volume Small Scaled Fault Interpretation Inline 1249: Residual= Initial Seismic Volume- Fault Enhanced Volume Noise spikes and bands of noise around low amplitude levels are removed Noise is localized mainly around Major faults The enhanced Seismic will help to better image weaker faults and enhance their Continuity Inline 1249: Initial Seismic Volume Fault Probability Cube Seismic volume Calculate Dip Volume Fault Enhanced Volume TWT SURFACES Curvature Mapping Seismic Geomorphology TVSW Spectral Bluing Seismic Attribtes Grids Detailed Small scale faults Interpretation Similarity Seismic Attribute Grids METHODOLOGY: In order to improve imaging of small scaled faults two techniques that shape the seismic spectrum to optimize the vertical resolution were used. The first is TIME VARIANT SPECTRAL WHITENING (TVSW). This method involves passing the input data through a number of narrow band-pass filters and determining the decay rates for each frequency band. The inverse of these decay functions for each frequency band is applied and the results are summed. The second is SEISMIC SPECTRAL BLUEING where an operator was designed by matching the spectrum of seismic data with well data and then convolved with input data to create enhanced volume. 3D View of Similarity Cube, Full Stack Seismic and faults Interpretation Serdj Horizon: Similarity attribute co-rendered with crossline Direction Energy gradient to better Image fault Continuity Serdj Horizon: Maximum curvature attribute co-rendred with Energy it shows minor E-W Faults E-W Minor Faults Fault Probability Cube computed Using Neural Network Extracted On Ain Ghrab Horizon Fault Probability Cube computed Using Neural Network Extracted On Serdj Horizon Blue: Intial Volume Red : Fault Enhanced Green: TVSW Cyan: SSB Inline 1252: initial Seismic Volume With Fault interpretation (black) Inline 1252: TVSW Seismic Volume With Minor Fault interpretation (white) the high-frequency noise is usually amplified (Using TVSW OR SSB) and so a band-pass filter must be applied to the resulting data Since conventional seismic data is band limited, it provides limited subsurface geological information. Moreover, higher frequencies within the band are more attenuated. In order to improve the characterization of faults In this study, we used two different filters: the dip-steered median filter to remove random noise and increase the lateral continuity of reflections, and the fault-enhancement filter used to enhance the discontinuities of the reflections. After filtering, similarity and curvature attributes were applied in order to identify the distribution of faults along the data.

Transcript of Structural interpretation and faults mapping

Page 1: Structural interpretation and faults mapping

win-win partnership

STRUCTURAL INTERPRETATION AND FAULTS MAPPING

USING SEISMIC ATTRIBUTES AND NEURAL NETWORK OF

A 3D POST-STACK SEISMIC SURVEY IN KABOUDIA PERMIT

General setting Fault Enhancement Workflow

Applying The Filters & QC

The Kaboudia permit lies in eastern offshore Tunisia and covers an area of about 3104 km².

It is bounded to the West by the Monastir-Mahdia shoreline and to the North east by the

Halk el Menzel Oil Field, the water depth is generally less than 200 meter.

The Kaboudia block is located in the Pelagian platform in the NW-SE “Mahdia-Isis”

paleohigh which separate the Gabes basin from the Hammamet basin. It lies in a highly

complex structural settings since Late Triassic to Early Cretaceous period which is marked

by extensive cycle and is relayed by Early Albian to Early Quaternary compressive cycles

which are mainly related to the subsidence and collision between African and European

plates. The compressive cycle is recognized by four major tectonic unconformities related

to Early Albian-Austrian phase (top Serj unconformity), to Late Cretaceous-Eocene

Pyrenean phase (top Souar-base Fortuna unconformity), to the Middle-Late Miocene

Atlassic phase (top Fortuna-base Ain Grab unconformity) and to the Early Quaternary phase

(base Pliocene-base RafRaf unconformity). These structural activities were associated with

significant movements of pre-existing faults and folds leading to the generation of many

local unconformities, structural inversion, erosion and lateral variation of facies of different

stratigraphic series.

We set out to develop a seismic-fault detection method which recombines multiple

attributes into a new attribute that gives the optimal view of the targeted object .The seismic

volume has been subject to several stages of pre-conditionning to enhance discontinuities

and calculate similarity and curvature attributes These final attributes show detailed

geometry of the major fault system and numerous subtle lineaments in the study area.

METHODOLOGY:

Seismic attributes can provide essential details

from seismic data, but noise present in the data-

set often distort the outcome. To optimally

remove noise, data conditioning was carried out

in two steps, in the first step, a dip volume was

created using Fourier transform based algorithm

in inline and cross-line directions. This dip

volume was then used as a reference, and a

median filter was applied on the seismic data

guided by prepared dip volume. It reduces the

structurally oriented random and coherent noise

and increases the continuity and visibility of

faults and fractures by preserving and at the

same time sharpening the edges.

Definitions:

Similarity is a coherence attribute that expresses

how similar two or more traces are to each other

either in the crossline or inline directions Trace

discontinuities of the seismic data may be the

result of the faults or stratigraphic features.

In the other hand, curvature attributes are being

increasingly used to characterize faults and

fractures, based on the fact that certain areas with

curved surfaces are related to discontinuity

zones, which can be represented by faults and

fractures. The curvature attribute emphasizes the

positive (antiforms) and negative (sinforms)

curvatures, whereas on a flat surface or in

inflection zones the curvature is zero.

Inline 1249: Fault Enhanced Volume

Small Scaled Fault Interpretation

Inline 1249: Residual= Initial Seismic Volume- Fault Enhanced Volume

Noise spikes and bands of noise around low amplitude levels are removed

Noise is localized mainly around Major faults

The enhanced Seismic will help to better image weaker faults and enhance their Continuity

Inline 1249: Initial Seismic Volume

Fault Probability Cube

Seismicvolume

CalculateDip Volume

Fault EnhancedVolume

TWT SURFACES

Curvature

Mapping SeismicGeomorphology

TVSW

Spectral Bluing

Seismic Attribtes Grids

Detailed Small scale faults Interpretation

Similarity

Seismic Attribute Grids

METHODOLOGY:

In order to improve imaging of small scaled faults

two techniques that shape the seismic spectrum

to optimize the vertical resolution were used. The

first is TIME VARIANT SPECTRAL WHITENING

(TVSW). This method involves passing the input

data through a number of narrow band-pass

filters and determining the decay rates for each

frequency band. The inverse of these decay

functions for each frequency band is applied and

the results are summed. The second is SEISMIC

SPECTRAL BLUEING where an operator was

designed by matching the spectrum of seismic

data with well data and then convolved with input

data to create enhanced volume.

3D View of Similarity Cube, Full Stack Seismic and faults Interpretation

Serdj Horizon: Similarity attribute co-rendered with crossline

Direction Energy gradient to better Image fault Continuity

Serdj Horizon: Maximum curvature attribute co-rendred with Energy

it shows minor E-W Faults

E-W Minor Faults

Fault Probability Cube

computed Using Neural Network

Extracted On Ain Ghrab Horizon

Fault Probability Cube

computed Using Neural Network

Extracted On Serdj Horizon

Blue: Intial Volume

Red : Fault Enhanced

Green: TVSW

Cyan: SSB

Inline 1252: initial Seismic Volume With Fault interpretation (black) Inline 1252: TVSW Seismic Volume With Minor Fault interpretation (white)

the high-frequency noise is usually amplified (Using TVSW OR SSB)

and so a band-pass filter must be applied to the resulting data

Since conventional seismic data is band limited, it provides limited subsurface geological information. Moreover, higher frequencies within the

band are more attenuated. In order to improve the characterization of faults In this study, we used two different filters: the dip-steered median

filter to remove random noise and increase the lateral continuity of reflections, and the fault-enhancement filter used to enhance the

discontinuities of the reflections. After filtering, similarity and curvature attributes were applied in order to identify the distribution of faults along

the data.