GeoTomo LLC

Houston, TX, United States

GeoTomo LLC

Houston, TX, United States
Time filter
Source Type

Geotomo Llc | Date: 2016-04-01

A computer-implemented method of joint inversion of seismic data comprising receiving seismic data, selecting, by a computer system, a travel time data and waveform data, the selecting using the seismic data and generating a velocity model of a near-surface. The method comprises preparing parameters with which to update the velocity model by: calculating synthetic travel time data and synthetic waveform data; determining a travel time misfit gradient using the travel time data and synthetic travel time data; determining a wave-property misfit gradient based on the waveform data and the synthetic waveform data; calculating a join inversion gradient comprising the travel time misfit gradient and the wave-property misfit gradient; and estimating a step length. The method further comprises updating the velocity model using the step length and join inversion gradient; and repeating the preparing and updating steps until a travel time misfit and wave-property misfit reach a predetermined value.

Zhang J.,GeoTomo LLC | Zhang X.,GeoTomo LLC | Zhang W.,GeoTomo LLC
SEG Technical Program Expanded Abstracts | Year: 2013

Similar to a web search engine, we develop a microseismic search engine that can help to estimate event location, magnitude, and focal mechanism all together in less than a second. The method includes the calculation of all possible microseismic events over a 3D grid with a known velocity model. We then index and rank all of the seismic waveforms following the characteristics of phase and amplitude information and create a database by applying multiple randomized K-dimensional tree method. When a microseismic event occurs, approximate best matches to the entry waveform shall be found immediately by comparing the input data with the characteristic features in the database. The method does not just return one solution, but a series of solutions just like the web search engine. Thus, the solution space delineates the resolution and uncertainty of the results. Also similar to a web search engine, it does not require any input parameter or processing experience, thus the solutions are the same for any user. We demonstrate the method with both synthetic and real data. It shows a great potential for routinely monitoring microseismic events in real time during hydraulic fracturing operation. © 2013 SEG.

Zhang C.,University of Science and Technology of China | Zhang J.,University of Science and Technology of China | Sun Z.,GeoTomo LLC
SEG Technical Program Expanded Abstracts | Year: 2014

Refraction traveltimes have long been applied for deriving long-wavelength statics solutions. They are also applied for deriving residual statics, but it requires providing with substantially accurate traveltime picks for the calculation. In this study, we present a residual statics method that applies interferometric theory to produce four stacked super-virtual refraction gathers with significantly improved signal-to-noise ratio. They include forward and backward super-virtual refraction gathers for receivers and shots. Picking first arrivals on these four gathers followed by applying a set of equations is able to derive reliable residual statics solutions. This approach can help dealing with noisy data and also avoid using traveltime picks from shot gathers. We demonstrate the approach by applying to synthetic data as well as real data. © 2014 SEG.

Zhang Z.,Hefei University of Technology | Zhang W.,Hefei University of Technology | Zhang W.,GeoTomo LLC | Li H.,Hefei University of Technology | Chen X.,Hefei University of Technology
Geophysical Journal International | Year: 2013

Simulating seismic waves with uniform grid in heterogeneous high-velocity contrast media requires small-grid spacing determined by the global minimal velocity, which leads to huge number of grid points and small time step. To reduce the computational cost, discontinuous grids that use a finer grid at the shallow low-velocity region and a coarser grid at highvelocity regions are needed. In this paper, we present a discontinuous grid implementation for the collocated-grid finite-difference (FD) methods to increase the efficiency of seismic wave modelling. The grid spacing ratio n could be an arbitrary integer n = 2. To downsample the wavefield from the finer grid to the coarser grid, our implementation can simply take the values on the finer grid without employing a downsampling filter for grid spacing ratio n = 2 to achieve stable results for long-time simulation. For grid spacing ratio n = 3, the Gaussian filter should be used as the downsampling filter to get a stable simulation. To interpolate the wavefield from the coarse grid to the finer grid, the trilinear interpolation is used. Combining the efficiency of discontinuous grid with the flexibility of collocated-grid FD method on curvilinear grids, our method can simulate large-scale high-frequency strong ground motion of real earthquake with consideration of surface topography. © The Authors 2012. Published by Oxford University Press on behalf of The Royal Astronomical Society.

Yilmaz O.,GeoTomo LLC
SEG Technical Program Expanded Abstracts | Year: 2011

Near-surface modeling for statics corrections is an integral part of a land seismic data processing workflow. The past, present, and the future methods for near-surface modeling can be categorized into five groups: 1 uphole surveys, 2 shallow seismic surveys, 3 traveltime tomography, 4 waveform inversion, and 5 joint inversion of seismic and non-seismic data. I present the i-stats - an image-based workflow for modeling near-surface anomalies, which 1 does not require first-break picking as for traveltime tomography, 2 does not require source wavelet estimation as for waveform inversion, 3 does not fail velocity inversions as in traveltime tomography, 4 does not suffer from velocity-depth ambiguity, and 5 does not exhaust computational resources as in waveform and joint inversions. The i-stats method is based on prestack depth migration of shot records from topography using a range of near-surface velocities. The resulting depth images form an image volume which can then be interpreted to pick the reflector associated with the base of the near-surface and to pick the velocities for the near surface from the corresponding horizon-consistent semblance spectrum. The resulting 'effective-medium' model for the near-surface comprises laterally varying velocities, only, but yields essentially the same statics that one calculates from a more complicated model for the near-surface that may be estimated from tomography or inversion methods. The effective-medium model of the near-surface actually conforms to the assumption of vertical raypath within the near-surface that underlies the statics corrections. I demonstrate the i-stats method to correct for the deleterious effect of the near-surface anomalies associated with sand dunes on subsurface reflections. © 2011 Society of Exploration Geophysicists.

Zhang W.,GeoTomo LLC. | Zhang J.,GeoTomo LLC.
SEG Technical Program Expanded Abstracts | Year: 2011

For inverting land seismic data, it sometimes requires to handle large topography variations in the full waveform tomographic imaging. This may not be a trivial issue, since the forward modeling using a finite difference approach may produce inaccurate results if the topography variations are too large and the surface numerical conditions are no longer valid. To solve the problem, we design a variable grid mesh system for acoustic finite-difference modeling that addresses two concerns at the same time, that is, topography variations and finer grids required for lower velocity medium. The grid system remains constant laterally, but the grid size increases downward. In such a system, topography shall be sufficiently sampled with sophisticated boundary conditions, and the high velocity area in the deeper part is not oversampled, ensuring efficient computation. Along topography, we apply boundary conditions in multiple directions so that ensures wavefield continuity across lateral grids. In inversion, we convert the variable grids to a regular and uniform grid system so that sensitivities are uniformly weighted. We tested with foothill synthetic model for forward modeling and also for waveform inversion. © 2011 Society of Exploration Geophysicists.

He L.,University of Science and Technology of China | Zhang J.,University of Science and Technology of China | Zhang W.,GeoTomo LLC
SEG Technical Program Expanded Abstracts | Year: 2011

The complexity in the near-surface area may significantly affect seismic wave propagation, causing reflection statics issues. Therefore, it is important to accurately resolve the near-surface velocity structures and address the problems of statics corrections. We design several near-surface numerical models, and apply four imaging methods to explore the performance of the methods with each model. These methods include refraction delay-time method, GLI solution, the first-arrival traveltime tomography, and time-domain early-arrival waveform tomography. In the case of layer models with or without tomography variations, we found that nearly all methods produce reasonable statics solutions but with different computation runtime cost and sometime also different velocity structure solutions. Delay-time method and GLI may produce incorrect models, but correct statics times. This is because of the velocity-depth ambiguity in the two solutions. Therefore, their solutions may be applied for statics corrections but not for dynamic corrections. If the true models include vertical velocity gradients or any non-layer feature, traveltime tomography and waveform tomography prove to be better approaches. For the complex foothill numerical model that includes low velocity zones in the near-surface area, only early-arrival waveform tomography is capable of imaging the structures accurately. © 2011 Society of Exploration Geophysicists.

Loading GeoTomo LLC collaborators
Loading GeoTomo LLC collaborators