Seminars & Talks | Center for Energy & Geo Processing

12/09/2015 CeGP Seminar

Date/Time Wednesday, Dec. 9, 2015.
11:00 am to 12:00 pm (Atlanta), 7:00 pm to 8:00 pm (Dhahran)
Title Sparse-promoting Full Waveform Inversion based on Online Orthonormal Dictionary Learning
Speakers Mr. Lingchen Zhu
Abstract

Full waveform inversion (FWI) delivers high-resolution images of a subsurface medium model by minimizing iteratively the least-squares misfit between the observed and simulated seismic data. Due to the limited accuracy of the starting model and the inconsistency of the seismic waveform data, the FWI problem is inherently ill-posed, so that regularization techniques are typically applied to obtain better models. FWI is also a computationally expensive problem because modern seismic surveys cover very large areas of interest and collect massive volumes of data. The dimensionality of the problem and the heterogeneity of the medium both stress the need for faster algorithms and sparse regularization techniques to accelerate and improve imaging results.

This talk develops a compressive sensing approach for the FWI problem, where the sparsity of model perturbations is exploited within learned dictionaries. Based on stochastic approximations, the dictionaries are updated iteratively to adapt to dynamic model perturbations. Meanwhile, the dictionaries are kept orthonormal in order to maintain the corresponding transform in a fast and compact manner without introducing extra computational overhead to FWI. Such a sparsity regularization on model perturbations enables us to take randomly subsampled data for computation and thus significantly reduce the cost. Compared with other approaches that employ sparsity constraints in the fixed curvelet transform domain, our approach can achieve more robust inversion results with better model fit and visual quality.

Slides [PDF]

09/16/2015 CeGP Seminar

Date/Time Wednesday, Sept. 16, 2015.
11:00 am to 12:00 pm (Atlanta), 6:00 pm to 7:00 pm (Dhahran)
Title Microseismic Denoising and Detection using Autocorrelation
Speakers Dr. Entao Liu, Georgia Institute of Technology
Abstract The existing method of denoising based on cross-correlation suffers an issue of inaccurate alignment when the SNR is low. In order to overcome this drawback, we proposed a denoising scheme which is based on the autocorrelation. The alignment issue is taken care of automatically by this proposed simple method. The simulations based on synthetic and real waveform plus noise model show that this method is very robust to Gaussian noise. In addition, a detection algorithm is developed naturally without much extra computational burden.
Slides [PDF]

05/20/2015 CeGP Seminar

Date/Time Wednesday, May 20, 2015.
11:00 am to 12:00 pm (Atlanta), 6:00 pm to 7:00 pm (Dhahran)
Title SOC in Microearthquakes and Hydraulic Fracturing Microseisms
Speakers Dr. Tim Long, Professor Emeritus,
School of Earth and Atmospheric Sciences, Georgia Institute of Technology
Abstract The Gutenberg Richter recursion relation has long been recognized as evidence of chaotic behavior in the form of self-organized criticality (SOC) in earthquakes. The “b” value can be interpreted as a measure of fractal dimension. Earthquakes in the crystalline rocks of the Piedmont, like those associated with hydraulic fracturing have shallow hypocenters, typically less than a few kilometers. At these shallow depths, faults and/or joints can incorporate fluids such that as in Reservoir Induced seismicity changes in fluid pressure trigger earthquakes. The character of earthquake swarms and the spectral properties of earthquakes among other factors implicate fluids as variable component in shallow earthquake generation. Increased fluid pressure is known from RIS and Hydraulic Fracturing to weaken a fault under stress and trigger an earthquake. SOC comes into play when the earthquake rupture breaks down the asperities holding a fault plane apart and collapsing the fault. Consequently, fluids in the fault plane are compressed under lithostatic pressure resulting for some faults an increase in fluid pressure. Under increased fluid pressure the fluids will migrate away increasing the fluid pressure in surrounding rocks. Hence, for shallow crystalline rocks with appropriate stress and fracture density, the occurrence of one earthquake may through SOC lead to larger events. Where SOC is a large component in the seismicity the foreshocks can increase exponentially. The timing and location of shallow events will be unpredictable in time or location. Their hypocenters could migrate beyond the point of first rupture or the initiation or fluid injection.
Slides [PDF]

03/04/2015 CeGP Seminar

Date/Time Wednesday, March 04, 2015
11:00 am to 12:00 pm (Atlanta), 7:00 pm to 8:00 pm (Dhahran)
Title Microseismic Parameters Inversion without Arrival-Time Picking
Speakers Dr. Entao Liu, Georgia Institute of Technology
Abstract Conventional microseismic location schemes depend on manual pick of the arrival time of the waveform. However, the manual pick approach becomes inaccurate or even incapable when the noise level is high. There is a clear demand of effective and efficient microseismic location schemes without manual picking. In this talk, two microseismic events location methods based on stacking and full waveform matching are presented. Instead of solving an inverse problem, we propose to use a differential evolution algorithm to fulfill the grid points scanning. As a necessary tool, the moment tensor inversion are reviewed as well.
Slides [PDF]

02/18/2015 CeGP Seminar

Date/Time Wednesday, February 18, 2015
11:00 am to 12:00 pm (Atlanta), 7:00 pm to 8:00 pm (Dhahran)
Title Salt Dome Detection and Tracking Using Texture-Based Gradient and Tensor-Based Subspace Learning
Speakers Dr. Tamir Hegazy and Zhen Wang, Georgia Institute of Technology
Abstract The identification of salt dome boundaries in migrated seismic data volume is important for locating petroleum reservoirs. In this paper, we develop new algorithms that can label boundaries of salt domes both effectively and efficiently. Our research is twofold. First, we utilize a texture-based gradient to accomplish salt dome detection. We show that by employing a dissimilarity measure based on 2D discrete Fourier transform (DFT), the algorithm is capable of detecting salt dome boundaries with both high accuracy and excellent efficiency. Once the detection is performed for an initial 2D seismic section, we propose to track the initial boundaries through the data volume to accomplish an efficient labeling process by avoiding parameters tuning that would have been necessary if detection had been performed for every seismic section. The tracking involves a tensor-based subspace learning process, in which we build texture tensors using patches from different seismic sections. We validate our detection and tracking algorithms using seismic datasets acquired from Netherland offshore F3 block in the North Sea with very encouraging results.
Slides [PDF]

02/04/2015 CeGP Seminar

Date/Time Wednesday, February 4, 2015
11:00 am to 12:00 pm (Atlanta), 7:00 pm to 8:00 pm (Dhahran)
Title Seismic Denoising through Dictionary Learning
Speakers Lingchen Zhu, Georgia Institute of Technology
Abstract Seismic data comprises many traces that provide a spatio-temporal sampling of the reflected wave field. However, such information may suffer from different source of noise during acquisition which could possibly reduce the contribution of the seismic data in reservoir locating. Traditionally, fixed transforms are used to segregate the noise from the data by exploiting their different characteristics in transform domains. However, their performance are not satisfactory enough due to their lack of adaptability to some specific data morphology. In this paper, we propose a novel seismic data denoising scheme using a dictionary learning method based on a sparsity-promoting model. Both dictionary learning and denoising are processed from all overlapping patches of the given noisy seismic data for low complexity. Different from the previous dictionary learning methods that have to learn all atom elements, our method exploits underlying sparse structure of the learned atoms using previous methods over another generic dictionary and significantly reduces the dictionary elements need to be learned. By combining the advantages of multi-scale representations with the power of dictionary learning, more degrees of freedom can be provided to the sparse representation, and therefore the characteristics of seismic data can be efficiently captured in sparse coefficients for denoising. Numerical experiments on synthetic seismic data indicate that our scheme achieves the best performance in terms of peak signal-to-noise ratio (PSNR) and produces the least visual distortion.
Slides [PDF]

12/08/2014 CeGP Seminar

Date/Time Monday, December 8, 2014
10:00 am to 11:00 am (Atlanta), 6:00 pm to 7:00 pm (Dhahran)
Title Enhanced Reservoir Characterization and Uncertainty Quantification: A Multi-Data History Matching Approach
Speakers Dr. Ibrahim Hoteit, King Abdullah University of Science and Technology (KAUST)
Abstract Reservoir characterization and prediction assume an essential role in reservoir management, assisting in determining production strategies for efficient recovery and future planning. History matching is at the heart of calibrating the reservoir simulators to best emulate and forecast the true reservoir formation. This process traditionally involved the fine-tuning of critical reservoir parameters based almost entirely on production data that are only observed at some wells. While adequate matches could be realized, these data are spatially rather sparse to accurately characterize the reservoir. Consequently, various 4D monitoring techniques have been developed for better exploration of the reservoir formation in space and time. These include time-lapse seismic and electromagnetic surveys, and gravity and satellite INSAR data, which provide different, but complimentary in information about the reservoir parameters. Up-to-date, very little has been undertaken to exploit the synergy from combining these different datasets, often separately interpreted and analyzed before they are provided as inputs to the reservoir simulator. In this talk I will present and discuss the development of an efficient multi-data reservoir characterization framework for simultaneously and directly incorporating all these datasets into the history matching process for best possible reconstruction, prediction, and uncertainty quantification of the reservoir formation. I will also outline how the resulting information could be used for planning and decision making.
Slides [PDF]

11/19/2014 CeGP Seminar

Date/Time Wednesday, November 19, 2014
11:00 am to 12:00 pm (Atlanta), 7:00 pm to 8:00 pm (Dhahran)
Title Texture attributes for detecting salt bodies in seismic data
Automatic fault surface detection by using 3D Hough transform
Speakers Dr. Tamir Hegazy and Zhen Wang, Georgia Institute of Technology
Abstract

Texture attributes for detecting salt bodies in seismic data

Texture-based methods have proven to be useful in the detection of salt bodies in seismic data. In this abstract, we present three computationally inexpensive texture attributes that strongly differentiate salt bodies from other geological formations. The proposed method combines the three texture attributes along with region boundary smoothing for delineating salt boundaries. Our first proposed attribute is directionality, which differentiates between regions where texture lacks any specific direction (potentially, salt) and areas with directional texture. The second attribute is the smoothness of texture, while the third is based on edge content. Our results show that the directionality attribute effectively detects salt bodies in all the seismic images used in testing. The other two attributes correct the false positives detected by the directionality. The overall results show that the proposed method can fairly detect salt regions when compared to manual interpretation.

Automatic fault surface detection by using 3D Hough transform

Detection of faults plays an important role in the characterization of reservoir regions. In this paper, we propose an automatic fault surface detection method using 3D Hough transform to improve the interpretation efficiency. We first highlight the likely fault points in seismic data by thresholding the corresponding discontinuity volumes. Then, we apply 3D Hough transform to detect the likely fault planes in seismic volumes. After filtering out the noisy planes, we apply the weighted plane fitting method to extract the smooth fault surfaces from the remaining fault planes. Experimental results show that the proposed method has the capability of detecting fault surfaces in real seismic data with high accuracy and fewer human interventions.

Slides [PDF]

11/10/2014 CeGP Seminar

Date/Time Monday, November 10, 2014
09:30 am to 10:30 am (Atlanta), 5:30 pm to 6:30 pm (Dhahran)
Title Introduction to Marine Seismic Acquisition
Speakers Mr. Azizur Rahman Khan, Saudi Aramco
Abstract

About the tutorial

We have come a long way since the first successful seismic survey that discovered oil field in Texas, USA in the year 1924. Seismic surveys can save oil companies hundreds of millions of dollars by giving precise information about the subsurface geological features. Marine Seismic surveys have been used for decades to assess the location and size of oil and gas deposits which are buried several kilometers deep beneath the ocean floor.Various types of techniques and equipment are used for seismic exploration in offshore. It depends upon the marine environment, geological objectives and financial consideration.

Tutorial Outline & Objective

The tutorial will cover the fundamentals of seismic wave generation and propagation, reflection of seismic waves, various techniques used in marine seismic data acquisition, design & operation of marine seismic source, design & functioning of seismic sensors, seismic data recording system, positioning & navigation and Geophysical & Operational challenges in offshore seismic surveys. At the end of the tutorial the participants should be able to understand different marine seismic survey techniques, their merits and demerits and their application in oil and gas exploration.

Who should attend ?

The tutorial is designed for non-geophysicists who wish to gain an overview of different kinds of equipment & techniques used in marine seismic survey for hydrocarbon exploration.

Slides [PDF]

10/15/2014 CeGP Seminar

Date/Time Wednesday, October 15, 2014.
10:30 am to 1:30 pm (Atlanta), 5:30 pm to 8:30 pm (Dhahran)
Title Introduction to Field Acquisition
Speakers Dr. Abdullatif A. Al-Shuhail, King Fahd University of Petroleum and Minerals (KFUPM)
Abstract This tutorial will introduce the seismic exploration method and focus on seismic data acquisition. The objective is to prepare the participants for the forthcoming field trip to a seismic crew. Topics will include the following:

  • Introduction to geophysics
  • Seismic waves
  • Propagation and interface effects
  • Seismic signal and noise
  • Seismic wavelets
  • Seismic data acquisition
    • Equipment
    • 2D field procedures
    • 3D seismic exploration
    • Time-lapse (4D) seismic exploration
Slides [PDF]

09/17/2014 CeGP Seminar

Date/Time Wednesday,September 17, 2014.
11:30 am to 1:00 pm (Atlanta), 6:30 pm to 8:00 pm (Dhahran)
Title Parallel Computing Basics with a Case Study on the CeGP Cluster
Speakers Dr. Tamir Hegazy, Dr. Entao Liu, and Dr. Zhiling Long, Georgia Institute of Technology
Abstract The computational demand of big data, including geophysical applications, entails the use of parallel computing. This seminar, which is composed of three parts, touches upon important related topics. In the first part, we will introduce the basic architectural and programming paradigms of parallel computing. We will also provide the theoretical considerations for parallelizing sequential programs and analyzing the expected speedups. In the second part of the seminar, we will introduce the CeGP cluster and exercise how to classify it using the knowledge introduced in the first part. Finally, in the third part, we will demonstrate how to write and run parallel programs on the CeGP cluster using the Matlab Parallel Processing Toolbox, MPI, and MatlabMPI. After this introductory seminar, CeGP researchers and students will have the basic knowledge to start using the cluster for their CeGP projects.
Slides [PDF]

09/10/2014 CeGP Seminar

Date/Time Wednesday, September 10, 2014.
11:30 am to 1:00 pm (Atlanta), 6:30 pm to 8:00 pm (Dhahran)
Title Numerical Simulation of Exploration Geophysics: Introduction to S3I (Seismic Simulation, Survey, and Imaging)
Speakers Dr. Entao Liu and Lingchen Zhu, Georgia Institute of Technology
Abstract The 2D/3D reflection seismology is a fundamental method exploration geophysics, which based on the wave propagation in the subsurface complex media. Numerical simulation of wave equation and inverse problem solving techniques are highly expected. In this talk, we will introduce the acoustic/elastic wave equation and its numerical schemes and techniques in simulation. A tutorial of the newly developed MATLAB based package, named S3I (Seismic Simulation, Survey, and Imaging), will be given as well.
Slides [PDF]