Applications of Time-Domain Multiscale Waveform Tomography to Marine and Land Data C. Boonyasiriwat 1, J. Sheng 3, P. Valasek 2, P. Routh 2, B. Macy 2, W. Cao 1, and G.T. Schuster 1 1 Department of Geology and Geophysics, University of Utah 2 Seismic Technology Development, ConocoPhillips 3 Formerly University of Utah, Currently at Nexus Geoscience
Outline 1 IntroductionIntroduction Time-domain multiscale waveform tomographyTime-domain multiscale waveform tomography Processing workflowProcessing workflow Field data results:Field data results: Gulf of MexicoGulf of Mexico Saudi ArabiaSaudi Arabia SummarySummary
Waveform Tomography 2 Wave-equation based model building technique.Wave-equation based model building technique. 0 4 Depth (km) 016 Horizontal Position (km) Reconstructed V p Velocity Model True V p Velocity of Marmousi II Model Depth (km) Boonyasiriwat et al., 2008
Constant Density Acoustic Inversion Result 2.1 True Velocity (Good) Initial Velocity Model Reconstructed Velocity Model
Marmousi II: Key Learnings 2.2 Acoustic inversion of elastic data can provide an accurate result at shallow depths.Acoustic inversion of elastic data can provide an accurate result at shallow depths. Kinematics are correctly predicted but amplitudes are not.Kinematics are correctly predicted but amplitudes are not. The elastic effect is accumulated and degrades the inversion result (accuracy and resolution) with increasing depth.The elastic effect is accumulated and degrades the inversion result (accuracy and resolution) with increasing depth.
Problem and Solution 3 Problem: Find a velocity model from seismic data that minimizes the data residual Proposed Solution: - Use a gradient-based method - Use a multiscale method in X-T domain
Observed Wavefield Waveform Tomography True Velocity Calculated Wavefield Initial Velocity Velocity Update Wavefield Residual Iterate until wavefield residual is small 4
Outline 5 IntroductionIntroduction Time-domain multiscale waveform tomographyTime-domain multiscale waveform tomography Processing workflowProcessing workflow Field data results:Field data results: Gulf of MexicoGulf of Mexico Saudi ArabiaSaudi Arabia SummarySummary
Why Use Multiscale? Low Frequency High Frequency Coarse Scale Fine Scale Image from Bunks et al. (1995) Model parameter (m) Misfit function ( f ) 6
Multiscale Waveform Tomography 1. Collect data d(x,t) 2. Generate synthetic data d(x,t) by FD method syn. 3. Adjust v(x,z) until ||d(x,t)-d(x,t) || minimized by CG. syn To prevent getting stuck in local minima: a). Invert early arrivals initially a). Invert early arrivals initially mute 7 b). Use multiscale: low freq. high freq. b). Use multiscale: low freq. high freq.
Outline 8 IntroductionIntroduction Time-domain multiscale waveform tomographyTime-domain multiscale waveform tomography Processing workflowProcessing workflow Field data results:Field data results: Gulf of MexicoGulf of Mexico Saudi ArabiaSaudi Arabia SummarySummary
9 Processing Workflow Pre-Processing of Data Estimating Source Wavelet Generating Initial Model Multiscale Waveform Tomography Validating Velocity Tomograms
9 Processing Workflow 3D-to-2D conversion Attenuation compensation Random noise removal Pre-Processing of Data Estimating Source Wavelet Generating Initial Model Multiscale Waveform Tomography Validating Velocity Tomograms
9 Processing Workflow Pre-Processing of Data Estimating Source Wavelet Generating Initial Model Multiscale Waveform Tomography Validating Velocity Tomograms Pick the water-bottom Stack along the water-bottom Generate a stacked section
9 Processing Workflow Pre-Processing of Data Estimating Source Wavelet Generating Initial Model Multiscale Waveform Tomography Validating Velocity Tomograms Traveltime picking Initial model: RMS velocity Refraction traveltime inversion
9 Processing Workflow Pre-Processing of Data Estimating Source Wavelet Generating Initial Model Multiscale Waveform Tomography Validating Velocity Tomograms Low-pass filtering Inversion from low- to high-frequency bands high-frequency bands
9 Processing Workflow Pre-Processing of Data Estimating Source Wavelet Generating Initial Model Multiscale Waveform Tomography Validating Velocity Tomograms Migration images Common image gathers
Outline 10 IntroductionIntroduction Time-domain multiscale waveform tomographyTime-domain multiscale waveform tomography Processing workflowProcessing workflow Field data results:Field data results: Gulf of MexicoGulf of Mexico Saudi ArabiaSaudi Arabia SummarySummary
515 Shots 480 Hydrophones 12.5 m dt = 2 ms T max = 10 s 11 Gulf of Mexico Data
Low-pass Filtering 12
Reconstructed Velocity 13 Velocity (m/s)
Kirchhoff Migration Images 14
Kirchhoff Migration Images 14
Comparing CIGs 15
Comparing CIGs 16 CIG from Traveltime Tomogram CIG from Waveform Tomogram
Comparing CIGs 17
Comparing CIGs 18 CIG from Traveltime Tomogram CIG from Waveform Tomogram
Comparing CIGs 19
Comparing CIGs 20 CIG from Traveltime Tomogram CIG from Waveform Tomogram
Outline 21 IntroductionIntroduction Time-domain multiscale waveform tomographyTime-domain multiscale waveform tomography Processing workflowProcessing workflow Field data results:Field data results: Gulf of MexicoGulf of Mexico Saudi ArabiaSaudi Arabia SummarySummary
Saudi Arabia Land Survey 0 km 1.6 km X-Coord. (km) 0 50 Y-Coord. (km) Offset (km) 0 2 Time (s) 100 m CSGs, 240 traces/gather 4. Pick 246,000 traveltimes m station interval, max. offset = 3.6km max. offset = 3.6km 3. Line Length = 46 km 10 Depth (km) X-Coord. (km) (m/s) 5. Traveltime tomography -> V(x,y,z) 22
Brute Stack Section Time (s) CDP 23
Traveltime Tomostatics + Stacking Time (s) CDP 24
Waveform Tomostatics + Stacking Time (s) CDP 25
Outline 26 IntroductionIntroduction Time-domain multiscale waveform tomographyTime-domain multiscale waveform tomography Processing workflowProcessing workflow Field data results:Field data results: Gulf of MexicoGulf of Mexico Saudi ArabiaSaudi Arabia SummarySummary
Summary Acoustic waveform inversion was successfully applied to both marine and land datasets, and can provide accurate velocity subsurface structures. Issues: Cost > 100 iterations: How to reduce cost?Cost > 100 iterations: How to reduce cost? Acoustic vs. Elastic: How far can we go with acoustic?Acoustic vs. Elastic: How far can we go with acoustic? Anisotropy needed?Anisotropy needed? Source wavelet important: Source-independent inversion.Source wavelet important: Source-independent inversion. Missing low frequencies: Better initial model via reflection tomography.Missing low frequencies: Better initial model via reflection tomography. 27
Acknowledgment 28 We would like to thank UTAM sponsors for financial support.UTAM sponsors for financial support. Amarada Hess and Saudi Aramco for providing us the datasets.Amarada Hess and Saudi Aramco for providing us the datasets.