Hao Hu, Ph.D. in Geophysics
Assistant Professor, University of Oklahoma
Senior Research Geophysicist, TGS
I am a geophysicist using seismic signals/methods to find answers for :
How to balance climate change and energy consumption via exploration of geothermal/fossil resources and CO2 geological storage;
How to solve cutting-edge scientific and technical problems in understanding the subsurface structures, from shallow to deep;
How to conduct fundamental studies of seismic theory and algorithm, especially in seismic wave propagation, imaging, and inversion.
How to develop industrial programs that can perform efficiently, accurately, and stably, especially for elastic full waveform inversion and imaging.
My research interests mainly focus on:
• Exploration of unconventional/conventional energy resources using seismic methods
fracture characterization for geothermal/fossil resource reservoirs,
CO2 geological storage for carbon neutralization,
Geological structures of target reservoirs from seismic images;
• Understanding the subsurface structures from shallow to deep using seismic signals
near-surface imaging of marine and land structures,
upper mantle/crust and subducting slab structures,
mantle heterogeneities and core-mantle boundary (CMB) topography;
• Fundamental theory studies
high-resolution seismic modeling, imaging methods, and inversion theory,
Planetary seismology especially for Mars,
Machine learning in Geophysics.
Expertise areas
Algorithm developments/applications Fundamental Research
* Geothermal/fossil/CO2-storage reservoir * 3D seismic wave modeling and high-resolution imaging
* 3D Seismic imaging: marine and land * Nonlinear signal processing
* Seismic signal processing and inversion * Stochastic inversion of heterogeneity
* Proprietary surface wave separation/removal * Planetary seismology (Mars)
* Machine learning in seismic studies * HPC using large-scale GPU/CPU
Programming skills
C/C++, Fortran: independently developed various types of forward modeling algorithms (e.g., acoustic /elastic /decoupled /anisotropic), elastic reverse-time migration, least-squares Gaussian beam migration; modified many open-source codes, e.g., specfem3D, Geomechanics modeling.
Matlab: independently developed many algorithms (e.g., surface waves, forward modeling and imaging algorithms, signal processing, fracture characterization), GUI design and optimization (CougarSeismic); acceleration using CPU/GPU.
Parallel computing: high-performance computing (CPU/GPU) from hardware architecture to algorithm; experienced in MPICH, OpenMP, and CUDA (GPU).
Python: Numerical calculation, Devito, Numba, TensorFlow.
Others: Shell, Linux, Git.
Selected Research Topics
Seismic fracture characterization for geothermal/fossil resource reservoirs
Subsurface fractures are important for understanding regional stress states and for controlling the fluid flows in the potential natural resource reservoirs. The knowledge of fractures is critical for drilling and hydraulic fracturing, as well as evaluating CO2 geological storage. We have developed novel fracture characterization methods using recorded multiple-component seismic data and applied them on real field datasets. Particularly, we developed a creative method to invert the discrete fracture spatial distribution using ML from the seismic data. Recently, I also developed a new stochastic inversion method using the recorded reflected wave fluctuations to invert for the subsurface heterogeneity that is closely related to the reservoir properties.
Left: Reservoir model with multiple coexisting fracture sets. Right: a) Map view of true fracture network with two sets of fractures (one set along the horizontal axis, the second set is circular); b) P-to-P double beam result; c) P-to-Santi (anti-plane S) result; d) P-to-Sin (in-plane) result. (Figures from Hu. et a., 2021)
Soda Lake geothermal field is one of the first geothermal fields developed in Nevada. We apply our Double-Beam Neural Network (DBNN) method to the 3D Soda Lake seismic data to identify additional blind geothermal resources, particularly the shallow steam-charged fracture zones with large fracture compliance values. We identify four possible drilling targets showing high fracture compliance values, with one of them (Well 41-33) previously verified as a hot steam zone via drilling. Left: Well locations (yellow) superimposed on the seismic acquisition geometry. The circles, P0, P1, P2, and P3 are proposed prospects. Right: Inverted fracture maps at depth 250. High compliance prospect regions are labeled as P0, P1, P2, and P3.
Selected pulications related to this topic
Zheng Y., H. Hu, M. N. Bugti, J. Parsons, L. Huang, K. Gao, T. Cladouhos. Characterizing Steam-Filled Fracture Zones at the Soda Lake Geothermal Field Using Seismic Double-Beam Neural Network (DBNN), PROCEEDINGS, 48th Workshop on Geothermal Reservoir Engineering, pp.9, Stanford University, Stanford, California, February 6-8, 2023. Link. PDF.
Hu H., Alali M. A., Almomin A., and Y. Zheng, 2021, Novel 3D Seismic Characterization of Fractures Using Elastic P-to-S Double-Beams, Geophysics, 86(6): 1-51.
Zheng, Y., J. Li, H. Hu, K. Gao, L. Huang, and T. Cladouhos (2021). Seismic Double-beam Neural Network Approach to Characterizing Small-Scale Fractures in Geothermal Fields, Geothermal Rising Conference 2021 proceedings.
Hu H., and Y. Zheng, 2020, Stochastic inversion of Gaussian random media using transverse coherence functions for reflected waves, Journal of Geophysical Research: Solid Earth, 125(12), https://doi.org/10.1029/2020JB020385.
Hu H., Y. Zheng, X. Fang, and M. C. Fehler, 2018, 3D Seismic characterization of fractures with Random spacing Using the Double-Beam Method, Geophysics, 83(5): M63-M74.
Hu H. and Y. Zheng, 2018, 3D Seismic Characterization of Fractures in a Dipping Layer Using the Double-beam Method, Geophysics, .83(2): V123-V134.
Seismic imaging of subsurface using active sources
Seismic imaging can provide valuable and reliable information about subsurface and near-surface structures. Such information is critical for understanding the tectonic/sedimentary processes that guide natural resource exploration. I have been working on developing various seismic techniques to improve image quality for marine and land seismic data. These unique methods include reverse-time migration using primaries and multiples, least-squares migration, and Gaussian beam migrations.
Examples: Left: Traditional seismic imaging for marine structures. Right: Our high-resolution image using least-squares Gaussian Beam migration. Notice the improvements in the black (better resolution) and white (better amplitude) boxes. (Figures from Hu. et a., 2016)
3D DAS-VSP surveys can complement surface seismic data to provide a more holistic subsalt image with improved illumination and reduced costs. However, the DAS-VSP geometry intrinsically has limited angular coverage which leads to strong swing migration artifacts that contaminate and degrade the seismic image. To enhance the image quality of DAS-VSP data, we propose a structurally adaptive aperture for Reverse Time Migration (RTM) to confine the image to a predefined angular range. Application of the algorithm to a field DAS-VSP dataset collected in a complex salt environment shows a significant uplift of the subsalt image (Hu et al., 2023, EAGE).
Selected pulications related to this topic
Hao Hu, Ge Zhan, Faqi Liu, Carlos Cardanons, Bin Wang, 2023, Enhancing subsalt imaging of DAS-VSP data by structurally adaptive aperture RTM, EAGE extended abstract. Link. PDF
Hu H., K. Xia, F. Hilterman, Y. Zhang, 2020, Amplitude-versus-angle analysis of local angle-domain common image gathers with prestack Gaussian beam migration of Seismic data, IEEE, Transactions on Geoscience and Remote Sensing, 58(8): 5969-5975.
Liu X., Y. Liu, H. Lu and H. Hu, 2017, Prestack correlative least-squares reverse time migration, Geophysics, 82(2): S159-S172.
Hu H., Y. Liu, Y. Zheng, X. Liu and H. Lu, 2016, Least-squares Gaussian beam migration. Geophysics, 81(3), S87-S100.
Hu H., Y. Liu, A. Osen and Y. Zheng, 2015, Compression of local slant stacks by the estimation of multiple local slopes and the matching pursuit decomposition. Geophysics, 80, WD175-187.
Hu H., Y. Wang and X. Chang, 2015, Migration of free-surface-related multiples: removing artefacts using a water-layer model, Journal of Applied Geophysics, 112, 147-156.
Y. Liu, H. Hu, X. Xie and Y. Zheng, 2015, Reverse time migration of internal multiples for subsalt imaging. Geophysics, 80, S175-S185.
Imaging Earth’s interior structures using earthquake signals
I use 3D seismic imaging methods developed for industrial applications to image Earth structures using passive earthquake data, such as the sedimentary layers and subducting slabs. I also worked with Prof. Zheng to use deep-focus earthquakes to image discontinuities in the mantle above the slab to understand the interaction between the Tonga slab and the Samoa Plume. I also investigate the subducting slab using our free-constrain moment tensor inversion and tilted-slab-related shear wave splitting.
Examples: Structural discontinuiey image of Tonga subduction zone using earthquake data. (Figures from Li. et a., 2019)
Animation: How reverse-time migration (RTM) can image stratigraphic layers and slab?
Selected pulications related to this topic
Li L., Y. Chen, Y. Zheng, H. Hu and J. Wu, 2019, Seismic Evidence for Plume-Slab Interaction by High-resolution Imaging of the 410-km Discontinuity Under Tonga, Geophysical Research Letters, 46(23): 13687-13694.
Zhang Y., A. Li and H. Hu, 2019, Crustal structure in Alaska from receiver function analysis, Geophysical Research Letters, 46(3): 1284-1292.
Zhou. H. W., H. Hu, Z. Zou, Y. Wo and O. Youn, 2018, Reverse time migration: A prospect of seismic imaging, Earth-Science Reviews, 179: 207-227.
The RTM image of stratigraphic layers and slab using synthetic seismograms from 100 earthquakes
Unique non-linear surface-wave analysis and applications
Surface waves are energetic waves propagating along the Earth’s surface and seafloor. They can be used to understand the physical properties of near-surface structures. I developed a new and unique method to remove/isolate/separate the surface waves from the entire dataset using a novel high-resolution nonlinear signal comparison method. This work has resulted in two patents and one commercial license in exploring near-surface resources. This method could also be used to detect a low-velocity zone (LVZ) on Mars using the InSight observations. This method can provide critical information for detecting the thermal evolution of Mars. We also applied the ML to automatically identify the surface waves in the recorded seismic data.
High-resoluion dispersion map using our unique nonlinear signal comparison (NLSC)
Feild seismic gather from vibroseis source to demonstate the surface wave seperation
Selected pulications related to this topic
Li, D., X. Tian, H. Hu, X. Tang, X. Fang and Y. Zheng, 2020, Gaussian beam imaging of fractures near the wellbore using sonic logging tools after removing dispersive borehole waves, Geophysics, 85(4): 1-47.
Hu H. and Y. Zheng, 2019, Data-driven dispersive surface-wave prediction and mode separation using high-resolution dispersion estimation, Journal of Applied Geophysics, 171: 1-10.
Hu H., M. Senkaya and Y. Zheng, 2019, A novel measurement of the surface wave dispersion with high and adjustable resolution: Multi-channel nonlinear signal comparison, Journal of Applied Geophysics, 160: 236-241.
Zheng Y. and H. Hu, 2017, Nonlinear signal comparison and high-resolution measurement of surface wave dispersion, Bulletin of the Seismological Society of America, 107(3):1551-1556.
Understanding earth heterogeneities using stochastic inversion of seismic waveform fluctuations
Heterogeneities in Earth's interior can be used to detect the composition mixing, thermal and chemical processes of minerals. The seismic waves carry rich information of heterogeneities in terms of their scales and strengths. We can measure the amplitude and phase fluctuations of recorded seismic waves to detect the heterogeneities using stochastic inversion. We developed the stochastic inversion to use reflected waves and relative-phase waves (e.g., PKPbc-PKPdf) to expand its applicable regions, such as reservoir porosities and the mineral mixing nearby the core-mantle-boundary.
3D synthetic model with random heterogeneities.
Modeled waveforms and their phase and amplitude fluctuations.
Detecting random heterogeneities of lowerest mantle using waveforms recorded by Japan HiNet network. (a) The wavepaths of PKPbc and PKPdf. (b) The map of stations (blue triangles) and detected region on CMB (black crosses). (c) One example of observations. (d) The measured coherence function and the best-fit one. (e) The inverted spectrum of heterogeneities.
Selected pulications related to this topic
Hu H., J. Li, J. Zhang, Y. Zheng, V. Cormier, Amount of Subducted Basaltic Crust at the Core-Mantle Boundary Beneath Japan Revealed by PKPdf and PKPbc Waves, in preparation.
Hu H., and Y. Zheng, 2020, Stochastic inversion of Gaussian random media using transverse coherence functions for reflected waves, Journal of Geophysical Research: Solid Earth, 125(12), https://doi.org/10.1029/2020JB020385.
Fundamental theory studies of seismic wave propagation, imaging, and inversion
Studies of seismic wave propagation, imaging, and inversion are the cornerstones of using seismic signals to understand the planets. For the question-driven research, we always meet unexpected obstacles that need us to create new methods or develop existing methods to overcome. I am interested in fundamental studies, especially high-frequency high-efficient full-waveform modeling, seismic imaging with high resolution and fidelity, convergency of inversion solutions, and non-constrain moment tensor inversion. Especially, the fast high-frequency 3D global seismic wave modeling will be an ice-breaker for imaging the earth’s interior with high resolution.
Elastic full waveform modeling for high-frequency body waves (> 2hz) through the Earth's interior. The wavepath is PKP in this demonstration.
Speeding up the inversion convergence. (Figure from Roya et al., 2018)
Improving seismic imaging in terms of resolution, fidelity, and efficiency. (a) the traditional image. (b) the new image. (Hu. et a., 2016)
Non-constrain moment tensor inversion using high-frequency body waves. Left: waveforms of P, pP, S and sS; Right: True and inverted beach balls.
Selected pulications related to this topic
Hu H., Y. Zheng, L. Huang, and K. Gao, Imaging of vertical faults using multi-component seismic data, in preparation.
Ding Y., H. Hu*, A. Malallah, M. C. Fehler, L. Huang, and Y. Zheng, Mapping subsurface karsts and voids using directional elastic wave packets, Geophysics, Geophysics, 86(6): 1-67.
Thongsang P., H. Hu*, H.W. Zhou and A. Lau, 2020, Imaging enhancement in angle-domain common-image-gathers using the connected-component labeling method, Pure and Applied Geophysics, 177: 4897–4912.
Eftekhar R., H. Hu* and Y. Zheng, 2018, Convergence acceleration in scattering series and seismic waveform inversion using nonlinear Shanks transformation, Geophysical Journal International, 214(3): 1732–1743.
Hu H., Y. Liu, Y. Zheng, X. Liu and H. Lu, 2016, Least-squares Gaussian beam migration. Geophysics, 81(3), S87-S100.
Y. Liu, H. Hu, X. Xie and Y. Zheng, 2015, Reverse time migration of internal multiples for subsalt imaging. Geophysics, 80, S175-S185.
Hu H., Y. Wang and X. Chang, 2015, Migration of free-surface-related multiples: removing artifacts using a water-layer model, Journal of Applied Geophysics, 112, 147-156.
Hu H., Y. Liu, X. Chang, Y. Wang, X. Du and R. Yang, 2013, Analysis and application on boundary treatment for the computation of reverse-time migration: Chinese Journal of Geophysics, 2033-2042.
Optimizing the machine learning (ML) applications in seismic studies
I am also interested in applying supervised/unsupervised machine learning (ML) to reduce human involvement and improve the efficiency and accuracy in geophysical studies, such as automatically identifying the surface waves from the seismic data, characterizing the rock facies from well-logging data, and inverting for the subsurface discrete fractures from the surface recorded seismic data, and improving image quality from ML and least-squares migration.
Recognizing surface waves from the common-shot gathers using unsupervised ML. (From Xia et al., 2018)
Characterizing rock facies using ML. (From Wei et al., 2019)
Reconstruction of small-scale fracture map from seismic data. (From Zheng et al., 2021)
Selected pulications related to this topic
Zheng, Y., J. Li, H. Hu, K. Gao, L. Huang and T. Cladouhos (2021). Seismic Double-beam Neural Network Approach to Characterizing Small-Scale Fractures in Geothermal Fields, Geothermal Rising Conference 2021 Proceedings.
Li J., H. Hu and Y. Zheng, 2019, Physics-guided machine learning identification of discrete fractures from double beam images, SEG, Expanded Abstracts.
Wei Z., H. Hu*, A. Lau and H. W. Zhou, 2019, Characterizing the rock facies using convolutional neural network with feature engineering and a data padding strategy, Pure and Applied Geophysics, 176(8): 3593-3605.
Xia K., F. Hilterman and H. Hu, 2018, Unsupervised Machine Learning Algorithm for Detecting and Outlining Surface Waves, Journal of Applied Geophysics, 157, 73-86.