• Title/Summary/Keyword: synthetic seismic data

Search Result 96, Processing Time 0.028 seconds

Synthetic Training Data Generation for Fault Detection Based on Deep Learning (딥러닝 기반 탄성파 단층 해석을 위한 합성 학습 자료 생성)

  • Choi, Woochang;Pyun, Sukjoon
    • Geophysics and Geophysical Exploration
    • /
    • v.24 no.3
    • /
    • pp.89-97
    • /
    • 2021
  • Fault detection in seismic data is well suited to the application of machine learning algorithms. Accordingly, various machine learning techniques are being developed. In recent studies, machine learning models, which utilize synthetic data, are the particular focus when training with deep learning. The use of synthetic training data has many advantages; Securing massive data for training becomes easy and generating exact fault labels is possible with the help of synthetic training data. To interpret real data with the model trained by synthetic data, the synthetic data used for training should be geologically realistic. In this study, we introduce a method to generate realistic synthetic seismic data. Initially, reflectivity models are generated to include realistic fault structures, and then, a one-way wave equation is applied to efficiently generate seismic stack sections. Next, a migration algorithm is used to remove diffraction artifacts and random noise is added to mimic actual field data. A convolutional neural network model based on the U-Net structure is used to verify the generated synthetic data set. From the results of the experiment, we confirm that realistic synthetic data effectively creates a deep learning model that can be applied to field data.

Characteristics of Virtual Reflection Images in Seismic Interferometry Using Synthetic Seismic Data (합성탄성파자료를 이용한 지진파 간섭법의 가상반사파 영상 특성)

  • Kim, Ki Young;Park, Iseul;Byun, Joongmoo
    • Geophysics and Geophysical Exploration
    • /
    • v.21 no.2
    • /
    • pp.94-102
    • /
    • 2018
  • To characterize virtual reflection images of deep subsurface by the method of seismic interferometry, we analyzed effects of offset range, ambient noise, missing data, and statics on interferograms. For the analyses, seismic energy was simulated to be generated by a 5 Hz point source at the surface. Vertical components of particle velocity were computed at 201 sensor locations at 100 m depths of 1 km intervals by the finite difference method. Each pair of synthetic seismic traces was cross-correlated to generate stacked reflection section by the conventional processing method. Wide-angle reflection problems in reflection interferometry can be minimized by setting a maximum offset range. Ambient noise, missing data, and statics turn to yield processing noise that spreads out from virtual sources due to stretch mutes during normal moveout corrections. The level of processing noise is most sensitive to amplitude and duration time of ambient noise in stacked sections but also affected by number of missing data and the amount of statics.

The Use of Unsupervised Machine Learning for the Attenuation of Seismic Noise (탄성파 자료 잡음 제거를 위한 비지도 학습 연구)

  • Kim, Sujeong;Jun, Hyunggu
    • Geophysics and Geophysical Exploration
    • /
    • v.25 no.2
    • /
    • pp.71-84
    • /
    • 2022
  • When acquiring seismic data, various types of simultaneously recorded seismic noise hinder accurate interpretation. Therefore, it is essential to attenuate this noise during the processing of seismic data and research on seismic noise attenuation. For this purpose, machine learning is extensively used. This study attempts to attenuate noise in prestack seismic data using unsupervised machine learning. Three unsupervised machine learning models, N2NUNET, PATCHUNET, and DDUL, are trained and applied to synthetic and field prestack seismic data to attenuate the noise and leave clean seismic data. The results are qualitatively and quantitatively analyzed and demonstrated that all three unsupervised learning models succeeded in removing seismic noise from both synthetic and field data. Of the three, the N2NUNET model performed the worst, and the PATCHUNET and DDUL models produced almost identical results, although the DDUL model performed slightly better.

Seismic AVO Analysis, AVO Modeling, AVO Inversion for understanding the gas-hydrate structure (가스 하이드레이트 부존층의 구조파악을 위한 탄성파 AVO 분석 AVO모델링, AVO역산)

  • Kim Gun-Duk;Chung Bu-Heung
    • 한국신재생에너지학회:학술대회논문집
    • /
    • 2005.06a
    • /
    • pp.643-646
    • /
    • 2005
  • The gas hydrate exploration using seismic reflection data, the detection of BSR(Bottom Simulating Reflector) on the seismic section is the most important work flow because the BSR have been interpreted as being formed at the base of a gas hydrate zone. Usually, BSR has some dominant qualitative characteristics on seismic section i.e. Wavelet phase reversal compare to sea bottom signal, Parallel layer with sea bottom, Strong amplitude, Masking phenomenon above the BSR, Cross bedding with other geological layer. Even though a BSR can be selected on seismic section with these guidance, it is not enough to conform as being true BSR. Some other available methods for verifying the BSR with reliable analysis quantitatively i.e. Interval velocity analysis, AVO(Amplitude Variation with Offset)analysis etc. Usually, AVO analysis can be divided by three main parts. The first part is AVO analysis, the second is AVO modeling and the last is AVO inversion. AVO analysis is unique method for detecting the free gas zone on seismic section directly. Therefore it can be a kind of useful analysis method for discriminating true BSR, which might arise from an Possion ratio contrast between high velocity layer, partially hydrated sediment and low velocity layer, water saturated gas sediment. During the AVO interpretation, as the AVO response can be changed depend upon the water saturation ratio, it is confused to discriminate the AVO response of gas layer from dry layer. In that case, the AVO modeling is necessary to generate synthetic seismogram comparing with real data. It can be available to make conclusions from correspondence or lack of correspondence between the two seismograms. AVO inversion process is the method for driving a geological model by iterative operation that the result ing synthetic seismogram matches to real data seismogram wi thin some tolerance level. AVO inversion is a topic of current research and for now there is no general consensus on how the process should be done or even whether is valid for standard seismic data. Unfortunately, there are no well log data acquired from gas hydrate exploration area in Korea. Instead of that data, well log data and seismic data acquired from gas sand area located nearby the gas hydrate exploration area is used to AVO analysis, As the results of AVO modeling, type III AVO anomaly confirmed on the gas sand layer. The Castagna's equation constant value for estimating the S-wave velocity are evaluated as A=0.86190, B=-3845.14431 respectively and water saturation ratio is $50\%$. To calculate the reflection coefficient of synthetic seismogram, the Zoeppritz equation is used. For AVO inversion process, the dataset provided by Hampson-Rushell CO. is used.

  • PDF

Deep-Learning Seismic Inversion using Laplace-domain wavefields (라플라스 영역 파동장을 이용한 딥러닝 탄성파 역산)

  • Jun Hyeon Jo;Wansoo Ha
    • Geophysics and Geophysical Exploration
    • /
    • v.26 no.2
    • /
    • pp.84-93
    • /
    • 2023
  • The supervised learning-based deep-learning seismic inversion techniques have demonstrated successful performance in synthetic data examples targeting small-scale areas. The supervised learning-based deep-learning seismic inversion uses time-domain wavefields as input and subsurface velocity models as output. Because the time-domain wavefields contain various types of wave information, the data size is considerably large. Therefore, research applying supervised learning-based deep-learning seismic inversion trained with a significant amount of field-scale data has not yet been conducted. In this study, we predict subsurface velocity models using Laplace-domain wavefields as input instead of time-domain wavefields to apply a supervised learning-based deep-learning seismic inversion technique to field-scale data. Using Laplace-domain wavefields instead of time-domain wavefields significantly reduces the size of the input data, thereby accelerating the neural network training, although the resolution of the results is reduced. Additionally, a large grid interval can be used to efficiently predict the velocity model of the field data size, and the results obtained can be used as the initial model for subsequent inversions. The neural network is trained using only synthetic data by generating a massive synthetic velocity model and Laplace-domain wavefields of the same size as the field-scale data. In addition, we adopt a towed-streamer acquisition geometry to simulate a marine seismic survey. Testing the trained network on numerical examples using the test data and a benchmark model yielded appropriate background velocity models.

Integrated approach using well data and seismic attributes for reservoir characterization

  • Kim Ji- Yeong;Lim Jong-Se;Shin Sung-Ryul
    • 한국지구물리탐사학회:학술대회논문집
    • /
    • 2003.11a
    • /
    • pp.723-730
    • /
    • 2003
  • In general, well log and core data have been utilized for reservoir characterization. These well data can provide valuable information on reservoir properties with high vertical resolution at well locations. While the seismic surveys cover large areas of field but give only indirect features about reservoir properties. Therefore it is possible to estimate the reservoir properties guided by seismic data on entire area if a relationship of seismic data and well data can be defined. Seismic attributes calculated from seismic surveys contain the particular reservoir features, so that they should be extracted and used properly according to the purpose of study. The method to select the suitable seismic attributes among enormous ones is needed. The stepwise regression and fuzzy curve analysis based on fuzzy logics are used for selecting the best attributes. The relationship can be utilized to estimate reservoir properties derived from seismic attributes. This methodology is applied to a synthetic seismogram and a sonic log acquired from velocity model. Seismic attributes calculated from the seismic data are reflection strength, instantaneous phase, instantaneous frequency and pseudo sonic logging data as well as seismic trace. The fuzzy curve analysis is used for choosing the best seismic attributes compared to sonic log as well data, so that seismic trace, reflection strength, instantaneous frequency, and pseudo sonic logging data are selected. The relationship between the seismic attribute and well data is found out by the statistical regression method and estimates the reliable well data at a specific field location derived from only seismic attributes. For a future work in this study, the methodology should be checked an applicability of the real fields with more complex and various reservoir features.

  • PDF

Enhancement of Seismic Stacking Energy with Crossdip Correction for Crooked Survey Lines

  • Kim, Ji Soo;Lee, Sun Jung;Seo, Yong Seok;Ju, Hyeon Tae
    • The Journal of Engineering Geology
    • /
    • v.24 no.2
    • /
    • pp.171-178
    • /
    • 2014
  • In seismic reflection data processing, the crossdip correction effectively focuses the stacking energy near the sharp bends of a crooked survey line. Additionally, approximate 3-D information on the reflector (e.g., true crossdip angle and lateral continuity) are locally investigated as a by-product of the crossdip correction procedure. Improvement of the signal-to-noise ratio and estimation of reflector crossdip attitude are tested, in terms of both common midpoint bin direction and processing-line type, using synthetic seismic reflection data. To effectively image the reflection energy near bends in seismic survey lines, straight-line binning is preferred to slalom-line binning.

Comparison of synthetic seismograms referred to inhomogeneous medium (불균질 매질에 따른 인공 합성 탄성파 자료 비교)

  • Kim, Young-Wan;Jang, Seung-Hyung;Yoon, Wang-Joong;Suh, Sang-Yong
    • 한국지구물리탐사학회:학술대회논문집
    • /
    • 2007.06a
    • /
    • pp.197-202
    • /
    • 2007
  • Most of seismic reflection prospecting assumes subsurface formation to be homogeneous media. These models are not capable of estimating small scale heterogeneity which is verified by well log data or drilling core. And those synthetic seismograms by homogeneous media are limited to explain various changes at field data. So we developed a inhomogeneous velocity model which can estimate inhomogeneity of background medium to implement numerical modeling from homogeneous medium and inhomogeneous medium on the model. Background medium using three autocorrelation functions in order to generate inhomogeneous velocity media was according to dominant wavelength of background medium and correlation length of random medium. And then we compared shot gathers. The results show that numerical modeling implemented at inhomogeneous medium depicts complex wave propagation of field data.

  • PDF

3D Seismic Data Processing Methodology using Public Domain Software System (공유 소프트웨어 시스템을 이용한 3차원 탄성파 자료처리 방법론)

  • Ji, Jun;Choi, Yun-Gyeong
    • Geophysics and Geophysical Exploration
    • /
    • v.13 no.2
    • /
    • pp.159-168
    • /
    • 2010
  • Recent trend in petroleum/gas exploration is an application of 3D seismic exploration technique. Unlike the conventional 2D seismic data processing, 3D seismic data processing is considered as the one which requires expensive commercial software systems and high performance computer. This paper propose a practical 3D seismic processing methodology on a personal computer using public domain software such as SU, SEPlib, and SEPlib3D. The applicability of the proposed method has been demonstrated by successful application to a well known realistic 3D synthetic data, SEG/EAGE 3D salt model data.

Depositional Facies Analysis from Seismic Attributes: Implication of Reservoir Characterization

  • Park Yong-Joon
    • 한국석유지질학회:학술대회논문집
    • /
    • autumn
    • /
    • pp.2-16
    • /
    • 1999
  • This study includes structural analysis of the northern Pattani Basin, areal description of depositional facies, and their spatial relationships using 3-D seismic and well data. Well log data indicate that the representative depositional facies of the studied intervals are sandy, fluvial, channel-fill facies encased in shaly floodplain deposits. Seismic responses were predicted from a synthetic seismogram using a model of dominant depositional facies. Peak-to-trough amplitude and instantaneous frequency seismic attributes are used in depositional facies interpretation. Three Intervals A, B and C are interpreted on the successive stratal surfaces. The shallowest interval, A, is the Quaternary transgressive succession. Each stratal surface showed flow pattern variation of fluvial channel facies. Two transgressive cycles were identified in interval A. Interval B also indicated fluvial facies. Depositional facies architectures are described by interpreting seismic attributes on the successive stratal surfaces.

  • PDF