Scribd
Upload
25 views5 pages

Inversion Vvi

The document discusses using machine learning and convolutional neural networks to perform automatic AVO analysis on seismic data from offshore South Korea. A CNN model was trained on data containing known gas reservoirs and then applied to a second dataset to identify AVO class III anomalies, which were confirmed by manual analysis.

Uploaded by

Bidyut Mandal
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
25 views5 pages

Inversion Vvi

The document discusses using machine learning and convolutional neural networks to perform automatic AVO analysis on seismic data from offshore South Korea. A CNN model was trained on data containing known gas reservoirs and then applied to a second dataset to identify AVO class III anomalies, which were confirmed by manual analysis.

Uploaded by

Bidyut Mandal
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 5

Modern seismic inversion methods transform seismic reflection data to pseudo acoustic impedance

logs. Inherent in this approach is an integration of low frequency model information with higher
frequencies from the seismic. This is necessary since the low frequencies present in logs and
therefore required by the inversion, are absent in the seismic data. We obtain the low frequency
impedances from a 3D solid model defined by interpreted horizons and impedance logs. It is
important to understand the effects of this a priori log information on the final inversion. We must
guard against using the logs to provide the answer. Our goal is to place the transformed high
frequency seismic information in a reasonable geologic setting defined by the low frequency model.
We investigate these ideas using a western Canadian example, a small 8600 bin 3D survey over a
Devonian reef within which, are located 19 wells, each with a set of sonic and density logs. First, we
compute an inversion assuming knowledge of only the most distant off-reef well. In this case, the low
frequency model is clearly erroneous. Next we re-do the inversion, adding a single on-reef well to
the model. Finally, we assume knowledge of all on-reef wells. We conclude that anomalous
impedance information resides mainly in the seismic domain. Lower frequencies from the model are
useful to provide a correct geological interpretation and sometimes necessary to calibrate the
anomalous behaviour. The modelling procedure is evolutionary, new well logs being used to update
both the model and the inversion. In this way, the inversion becomes a viable tool in both exploration
and production environments.

Low frequency models for seismic inversions: strategies for


success

Abstract
A critical issue in all seismic inversion procedures is the accounting for missing
information below the seismic band. We need that if we are to determine absolute
reservoir properties. Typically, inversionists create some sort of structural model from
workstation interpretations and then use well logs to populate it with the native
outcomes of elastic inversion – P Impedance, Vp/Vs and Density. However, the problem
is how to go about interpolating the logs between the wells. We are usually uncertain
about how to do this and often uncomfortable with the result.

We will demonstrate by example, a joint deterministic inversion – facies estimation


procedure which makes minimal prior assumptions and obviates the need for the
building of a low frequency model (LFM) which interpolates log curves between wells.
We then apply the same procedure to geostatistical inversion and show that facies can
successfully convey low frequency trend information between the two domains.
Seismic Impedance Inversion with Updated Low-Frequency
Models
December 2022 Satinder ChopraRitesh Kumar Sharma

Seismic impedance is widely used in our industry because it allows an integrated approach to
geological interpretation. The transformation of seismic amplitudes to impedance data can be
essentially seen as the change from an interface property to a layer property. This stratal interval
property (impedance) simplifies the lithologic and stratigraphic identification and can be directly
converted into lithologic or reservoir properties such as porosity, fluid fill and net pay. It also
allows for direct interpretation of three-dimensional geobodies.

As seismic data are bandlimited (typically exhibiting a bandwidth of 10-70 hertz), the direct
transformation of seismic amplitudes into impedance yields a relative impedance. The missing
low frequencies (1-10 hertz) are usually derived from well logs or stacking velocities and used as
a priori information during the inversion process. The quality of the low-frequency impedance
model used in the inversion has a pronounced effect on the final impedance result and thus needs
to be constructed carefully. More details on seismic impedance inversion can be picked up from
our earlier published articles of Geophysical Corner (May and June 2015 issues).

Seismic impedance inversion can be carried out on both poststack and prestack seismic data. In a
seismic gather, the near-offset amplitudes relate to changes in impedance of the subsurface
rocks, and thus depict the correct time of the reflection events. The far-offset amplitudes relate to
not only the changes in P-wave velocity and density, but the S-wave velocity as well. The
inversion of far-offset amplitudes in a gather yields the elastic impedance (as was described in
the October 2012 Geophysical Corner) and can be used for lithology and fluid discrimination.

Since the inversion process transforms seismic amplitudes directly into impedance values,
special attention needs to be paid to their preservation, which ensures that the observed
amplitude variations are related to geological effects. Besides this, preconditioning of seismic
data is usually carried out by adopting processes such as muting, bandpass filtering, random
noise removal and trim statics (see January and November 2016, and January, October and
November 2019 installments of Geophysical Corner). But sometimes these processes are not
enough. In such cases we can adopt some poststack processing steps for preconditioning noisy
prestack seismic data (see the July 2019 Geophysical Corner).
Machine learning derived AVO analysis on marine 3D seismic
data over gas reservoirs near South Korea
 November 2020

 Journal of Petroleum Science and Engineering 197(3):108105

Abstract
Amplitude variation with offset (AVO) analysis has become a crucial step in determining gas well locations. However,
large amounts of time and effort are required to confirm AVO anomalies, and interpretations can be inconsistent. To
avoid the need for the long processes involved in conventional manual analysis, we developed an automatic AVO
analysis method for common midpoint (CMP) gathers through using machine learning (ML) with a convolutional
neural network (CNN). To deal with complicated seismic data, the network was constructed based on VGG16
network architecture, which includes 16 layers. The resulting CNN-based algorithm was applied to two sets of three-
dimensional (3D) seismic data acquired off the east coast of South Korea. One dataset, which was obtained over gas
reservoirs and confirmed to show multiple AVO class III anomalies by comparison with geophysical well logging data,
was used for training and evaluation of the proposed CNN-based algorithm. The resulting trained model was tested
using the second dataset, which was obtained over an area near the gas reservoirs with a different depositional
environment. To demonstrate the applicability of the model to raw and final migrated CMP gathers, AVO class III
anomalies predicted by the ML-derived analysis were confirmed by manual AVO analysis of the test data.

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy