Dmitry Borisov
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Dmitry Borisov has a total of 43 co-authors. Here are the top co-authors by number of shared publications:
- Steve Sloan
US Army Corps of Engineers, Vicksburg, MS, USA
5 shared publications - Jeroen Tromp
Department of Geosciences, Princeton University , Princeton, NJ 08544
4 shared publications - Yanwei Zhang
Kansas Geology Survey , 1930 Constant Ave, Lawrence, KS 66047
4 shared publications - Fuchun Gao
Total E&P Research and Technology, Houston, Texas, USA.
3 shared publications - Ryan Modrak
University of Alaska Fairbanks 4 , Fairbanks, Alaska, .
3 shared publications - Shelby L. Peterie
University of Kansas 1 , Kansas Geological Survey, Lawrence, Kansas 66047, USA.;;;
3 shared publications
Latest Publications
- 23 June 2026
- Published by Society of Exploration Geophysicists in Geophysics
- Vol. 91 (4) , B139-B149
Seismic imaging of a salt-dissolution void in central Kansas
A subsurface anomaly in McPherson County, Kansas, was caused by salt dissolution following a casing breach in a brine wastewater disposal well during the 1980s. The breach led to the formation of a void and possible subsidence within and above the 150-m-deep Permian salt layer. Although borehole camera images confirmed the presence of the air-filled void, limited camera illumination restricted conclusive measurements of its horizontal extent. High-frequency seismic reflection data were acquired to delineate the extent of the void and assess potential surface subsidence. To achieve a more accurate mapping of the void, a 2D elastic multiscale full-waveform inversion (FWI) technique was used, which involved analyzing seismic body waves with frequencies ranging from 20 to 50 Hz and a maximum source offset of 1050 m. First-arrival travel-time tomography refined the initial velocity model above the salt layer, whereas automated first-arrival picking from seismic shot gathers was achieved through a custom-trained convolutional neural network. The FWI velocity updates accurately located the salt cavity, estimating it to be approximately 100 m wide and over 50 m tall, consistent with historical invasive measurements. Cavities of this nature can pose significant environmental or public safety risks because potential subsidence or collapse could lead to ground instability and negatively impact groundwater or infrastructure. Consequently, FWI and high-resolution reflection imaging serve as complementary tools for delineating and monitoring these salt-related anomalies. This specific salt interval is heavily used in this area for seasonal storage of liquid petroleum, solution and room and pillar mined for salt consumption, and highway applications, but it is unsuitable (uneconomical) for energy or CO2 storage due to its shallow depth and lack of potential for high-pressure fluid containment. However, this study helps establish a dedicated workflow for effective monitoring of deeper reservoirs more suitable for high-pressure storage.
- 23 June 2026
- Published by Society of Exploration Geophysicists in Geophysics
- Vol. 91 (4) , B139-B149
Seismic imaging of a salt-dissolution void in central Kansas
A subsurface anomaly in McPherson County, Kansas, was caused by salt dissolution following a casing breach in a brine wastewater disposal well during the 1980s. The breach led to the formation of a void and possible subsidence within and above the 150-m-deep Permian salt layer. Although borehole camera images confirmed the presence of the air-filled void, limited camera illumination restricted conclusive measurements of its horizontal extent. High-frequency seismic reflection data were acquired to delineate the extent of the void and assess potential surface subsidence. To achieve a more accurate mapping of the void, a 2D elastic multiscale full-waveform inversion (FWI) technique was used, which involved analyzing seismic body waves with frequencies ranging from 20 to 50 Hz and a maximum source offset of 1050 m. First-arrival travel-time tomography refined the initial velocity model above the salt layer, whereas automated first-arrival picking from seismic shot gathers was achieved through a custom-trained convolutional neural network. The FWI velocity updates accurately located the salt cavity, estimating it to be approximately 100 m wide and over 50 m tall, consistent with historical invasive measurements. Cavities of this nature can pose significant environmental or public safety risks because potential subsidence or collapse could lead to ground instability and negatively impact groundwater or infrastructure. Consequently, FWI and high-resolution reflection imaging serve as complementary tools for delineating and monitoring these salt-related anomalies. This specific salt interval is heavily used in this area for seasonal storage of liquid petroleum, solution and room and pillar mined for salt consumption, and highway applications, but it is unsuitable (uneconomical) for energy or CO2 storage due to its shallow depth and lack of potential for high-pressure fluid containment. However, this study helps establish a dedicated workflow for effective monitoring of deeper reservoirs more suitable for high-pressure storage.
- Open Access
- 27 October 2025
- Published by Oxford University Press (OUP) in Geophysical Journal International
- Vol. 244 (1)
Enhancing machine learning based seismic inversion with noise-augmented training datasets
SUMMARY: An 18-layer, U-shaped convolutional neural network was trained to predict Vs models and identify near-surface void locations. To enhance seismic inversion accuracy for real world applications, the model is trained on synthetic data sets augmented with field noise. While models trained on noise-augmented data showed poorer performance on synthetic testing data sets, they achieved lower root-mean-square error values and the best results on field data. The velocity model resulting from full-waveform inversion based on noise-augmented model accurately detected a void-like low-velocity zone near the known void location. This approach shows that training with field-noise-augmented data allows machine learning models to generalize better to real-world conditions, increasing their reliability for velocity inversion in noisy environments. The results highlight the strong potential of this strategy, particularly if a diverse range of real noise samples is incorporated during training.
- 27 October 2025
- Published by Oxford University Press (OUP) in Geophysical Journal International
- Vol. 244 (1)
Open AccessEnhancing machine learning based seismic inversion with noise-augmented training datasets
SUMMARY: An 18-layer, U-shaped convolutional neural network was trained to predict Vs models and identify near-surface void locations. To enhance seismic inversion accuracy for real world applications, the model is trained on synthetic data sets augmented with field noise. While models trained on noise-augmented data showed poorer performance on synthetic testing data sets, they achieved lower root-mean-square error values and the best results on field data. The velocity model resulting from full-waveform inversion based on noise-augmented model accurately detected a void-like low-velocity zone near the known void location. This approach shows that training with field-noise-augmented data allows machine learning models to generalize better to real-world conditions, increasing their reliability for velocity inversion in noisy environments. The results highlight the strong potential of this strategy, particularly if a diverse range of real noise samples is incorporated during training.
- 2 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 22 (1545598X) , 1-5
Using Deep Learning for Data-Driven Seismic Velocity and Source Wavelet Inversion
Full-waveform inversion requires an accurate representation of the source wavelet to enhance the reliability of model reconstruction. Traditional wavelet estimation methods depend on the accuracy of the existing velocity model. We propose an approach for the joint estimation of velocity and source wavelet in a data-driven manner. To achieve this, a dataset with randomly generated pairs of velocity models and source wavelets is created. Each wavelet is injected into its respective velocity model to simulate seismic records. A Convolutional Neural Network (CNN) is then trained to predict both velocity and wavelet features from the resulting shot gathers. The trained CNN effectively mitigates the bias in velocity estimation caused by an inaccurate source wavelet. By jointly estimating velocity and wavelet, the method exhibits improved seismic velocity reconstruction compared to a network that predicts velocity alone. The approach demonstrates potential for producing the input necessary to enable high-resolution FWI-based imaging on both synthetic and shallow seismic field datasets.
- 2 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 22 (1545598X) , 1-5
Using Deep Learning for Data-Driven Seismic Velocity and Source Wavelet Inversion
Full-waveform inversion requires an accurate representation of the source wavelet to enhance the reliability of model reconstruction. Traditional wavelet estimation methods depend on the accuracy of the existing velocity model. We propose an approach for the joint estimation of velocity and source wavelet in a data-driven manner. To achieve this, a dataset with randomly generated pairs of velocity models and source wavelets is created. Each wavelet is injected into its respective velocity model to simulate seismic records. A Convolutional Neural Network (CNN) is then trained to predict both velocity and wavelet features from the resulting shot gathers. The trained CNN effectively mitigates the bias in velocity estimation caused by an inaccurate source wavelet. By jointly estimating velocity and wavelet, the method exhibits improved seismic velocity reconstruction compared to a network that predicts velocity alone. The approach demonstrates potential for producing the input necessary to enable high-resolution FWI-based imaging on both synthetic and shallow seismic field datasets.
- 5 January 2025
- Published by Society of Exploration Geophysicists in Geophysics
- Vol. 90 (3) , R101-R112
Optimal transport-based full-waveform inversion for shallow seismic data
Full-waveform inversion (FWI) is widely used to reconstruct subsurface properties at different geologic scales. For shallow land applications using surface waves, a lack of information on the source wavelet, dispersion, and presence of higher modes increases the nonlinearity of the inverse problem. Moreover, the inversion can become more challenging with the presence of near-surface (NS) complexities associated with scattering, attenuation, and high-contrast variations in the elastic parameters. To address those challenges, we exploit the potential of graph-space optimal transport (GSOT)-based waveform inversion. Compared with the least-squares (LS) formulation, GSOT provides a more convex misfit function and reduces dependence on the accuracy of the initial model. Although a few field-data applications have shown the potential and benefits of using GSOT-based FWI with body waves, there are limited real applications of the inversion with a GSOT misfit function for NS characterization. Moreover, despite considerable effort with blind benchmark tests in exploration seismology, typically synthetic FWI examples for NS applications are demonstrated through an “inverse crime” approach, wherein the same forward-modeling engine is used to generate observed and simulated data. Synthetic FWI examples performed here compare the performance of LS- and GSOT-based FWI with more realistic scenarios. Those scenarios include using different wavefield propagators, inaccurate source wavelets, the elastic inversion of anelastic data, and the use of a coarse grid for the inversion of shallow seismic modeled and field data collected over a man-made elongated cavity located approximately 10 m below ground surface. We demonstrate that the GSOT misfit function improves the initial 1D velocity models and guides the updates toward the actual subsurface properties. This enables the recovery of higher-mode Rayleigh waves and reconstruction of the cavity with better precision.
- 5 January 2025
- Published by Society of Exploration Geophysicists in Geophysics
- Vol. 90 (3) , R101-R112
Optimal transport-based full-waveform inversion for shallow seismic data
Full-waveform inversion (FWI) is widely used to reconstruct subsurface properties at different geologic scales. For shallow land applications using surface waves, a lack of information on the source wavelet, dispersion, and presence of higher modes increases the nonlinearity of the inverse problem. Moreover, the inversion can become more challenging with the presence of near-surface (NS) complexities associated with scattering, attenuation, and high-contrast variations in the elastic parameters. To address those challenges, we exploit the potential of graph-space optimal transport (GSOT)-based waveform inversion. Compared with the least-squares (LS) formulation, GSOT provides a more convex misfit function and reduces dependence on the accuracy of the initial model. Although a few field-data applications have shown the potential and benefits of using GSOT-based FWI with body waves, there are limited real applications of the inversion with a GSOT misfit function for NS characterization. Moreover, despite considerable effort with blind benchmark tests in exploration seismology, typically synthetic FWI examples for NS applications are demonstrated through an “inverse crime” approach, wherein the same forward-modeling engine is used to generate observed and simulated data. Synthetic FWI examples performed here compare the performance of LS- and GSOT-based FWI with more realistic scenarios. Those scenarios include using different wavefield propagators, inaccurate source wavelets, the elastic inversion of anelastic data, and the use of a coarse grid for the inversion of shallow seismic modeled and field data collected over a man-made elongated cavity located approximately 10 m below ground surface. We demonstrate that the GSOT misfit function improves the initial 1D velocity models and guides the updates toward the actual subsurface properties. This enables the recovery of higher-mode Rayleigh waves and reconstruction of the cavity with better precision.
- 25 March 2024
- Published by Elsevier in Mechanical Systems and Signal Processing
- Vol. 213
Experimental full waveform inversion for elastic material characterization with accurate transducer modeling
- 25 March 2024
- Published by Elsevier in Mechanical Systems and Signal Processing
- Vol. 213
Experimental full waveform inversion for elastic material characterization with accurate transducer modeling
- 1 August 2023
- Published by Pleiades Publishing Ltd in Journal of Communications Technology and Electronics
- Vol. 68 (8) , 828-835
Some Features of Scattering of a Plane Electromagnetic Wave on a Plane Magnetodielectric Plate
The two-dimensional problem of scattering of a plane electromagnetic wave on a flat homogeneous magnetodielectric plate of finite dimensions is considered. The problem is solved by the method of integral equations for auxiliary surface currents. Algebraization of integral equations was carried out using the collocation method and overlapping three-step basis functions. The results characterizing the efficiency of the proposed algorithm are presented. The influence of the geometric and material parameters of the plate on its effective scattering surface (ESS) in the direction of wave incidence, which characterizes the effects of “invisibility” of the plate, is studied. It is shown that the minimum of the specified ESS is achieved at a plate thickness close to the thickness that ensures the in-phase geometry of the geometric-optical rays that have passed through the plate and the rays that have passed by the plate. The ESS of a plate made of a material with positive and negative refractive indices is compared.
- 1 August 2023
- Published by Pleiades Publishing Ltd in Journal of Communications Technology and Electronics
- Vol. 68 (8) , 828-835
Some Features of Scattering of a Plane Electromagnetic Wave on a Plane Magnetodielectric Plate
The two-dimensional problem of scattering of a plane electromagnetic wave on a flat homogeneous magnetodielectric plate of finite dimensions is considered. The problem is solved by the method of integral equations for auxiliary surface currents. Algebraization of integral equations was carried out using the collocation method and overlapping three-step basis functions. The results characterizing the efficiency of the proposed algorithm are presented. The influence of the geometric and material parameters of the plate on its effective scattering surface (ESS) in the direction of wave incidence, which characterizes the effects of “invisibility” of the plate, is studied. It is shown that the minimum of the specified ESS is achieved at a plate thickness close to the thickness that ensures the in-phase geometry of the geometric-optical rays that have passed through the plate and the rays that have passed by the plate. The ESS of a plate made of a material with positive and negative refractive indices is compared.
- 22 June 2022
- Published by Elsevier in Materialia
- Vol. 24
Subsurface polycrystalline reconstruction based on full waveform inversion - A 2D numerical study
- 22 June 2022
- Published by Elsevier in Materialia
- Vol. 24
Subsurface polycrystalline reconstruction based on full waveform inversion - A 2D numerical study
- 13 April 2022
- Published by Oxford University Press (OUP) in Geophysical Journal International
- Vol. 230 (3) , 1725-1750
Full-waveform centroid moment tensor inversion of passive seismic data acquired at the reservoir scale
Passive seismic inversion at the reservoir scale offers the advantages of low cost, negligible environmental impact, and the ability to probe a target area with low-frequency energy not afforded by even the most modern active-source seismic technology. In order to build starting models suitable for full-waveform wavespeed tomography, characterization of earthquake sources is an indispensable first step. We present a workflow for the centroid moment tensor (CMT) inversion of seismic events identified in a passive seismic data set acquired by a large and dense array of three-component broadband seismic sensors in a mountainous setting in the Himalayan foothills. The dataset comprised 256 instruments operating for 2×4 months over an area of 8,000 km2. An initial three-dimensional (3-D) wavespeed model was determined for the region via the analysis of first-arriving traveltime picks. Of the 2,607 identified seismic events that were well recorded at frequencies between 0.2–50 Hz, 86 with magnitudes 1.3 ≤ M ≤ 3.0 initially had their CMT focal mechanisms determined by a waveform fitting procedure built on a Green’s function approach in a one-dimensional (1-D) layered average wavespeed model, for stations within an offset of 10 km, in the frequency range 0.2–1.4 Hz. Here, we obtain updated CMT mechanisms for the 86 events in that catalog via multicomponent full-waveform inversion in the 3-D wavespeed model. Our workflow includes automated data- and model-driven data selection using a combination of different metrics derived from signal-to-noise considerations and waveform-fitting criteria, and relies upon spectral-element simulations of elastic wave propagation in the 3-D wavespeed model, honoring topography. Starting from the initial CMT solutions, we seek improvement to the data fit within the frequency band 0.5–2.5 Hz by minimizing the waveform difference between observed and synthetic data, while accommodating wavespeed-model errors by allowing for small time-shifts. We balance uneven data coverage and tune their contributions via data-space weighting functions. We quantify the improvements to the data fit in terms of different metrics. We summarize the changes to the CMT solutions, and present and analyze the resulting catalog for the region, including their breakdown into double-couple and non-double couple components, and their relation to mapped faults in the region.
- 13 April 2022
- Published by Oxford University Press (OUP) in Geophysical Journal International
- Vol. 230 (3) , 1725-1750
Full-waveform centroid moment tensor inversion of passive seismic data acquired at the reservoir scale
Passive seismic inversion at the reservoir scale offers the advantages of low cost, negligible environmental impact, and the ability to probe a target area with low-frequency energy not afforded by even the most modern active-source seismic technology. In order to build starting models suitable for full-waveform wavespeed tomography, characterization of earthquake sources is an indispensable first step. We present a workflow for the centroid moment tensor (CMT) inversion of seismic events identified in a passive seismic data set acquired by a large and dense array of three-component broadband seismic sensors in a mountainous setting in the Himalayan foothills. The dataset comprised 256 instruments operating for 2×4 months over an area of 8,000 km2. An initial three-dimensional (3-D) wavespeed model was determined for the region via the analysis of first-arriving traveltime picks. Of the 2,607 identified seismic events that were well recorded at frequencies between 0.2–50 Hz, 86 with magnitudes 1.3 ≤ M ≤ 3.0 initially had their CMT focal mechanisms determined by a waveform fitting procedure built on a Green’s function approach in a one-dimensional (1-D) layered average wavespeed model, for stations within an offset of 10 km, in the frequency range 0.2–1.4 Hz. Here, we obtain updated CMT mechanisms for the 86 events in that catalog via multicomponent full-waveform inversion in the 3-D wavespeed model. Our workflow includes automated data- and model-driven data selection using a combination of different metrics derived from signal-to-noise considerations and waveform-fitting criteria, and relies upon spectral-element simulations of elastic wave propagation in the 3-D wavespeed model, honoring topography. Starting from the initial CMT solutions, we seek improvement to the data fit within the frequency band 0.5–2.5 Hz by minimizing the waveform difference between observed and synthetic data, while accommodating wavespeed-model errors by allowing for small time-shifts. We balance uneven data coverage and tune their contributions via data-space weighting functions. We quantify the improvements to the data fit in terms of different metrics. We summarize the changes to the CMT solutions, and present and analyze the resulting catalog for the region, including their breakdown into double-couple and non-double couple components, and their relation to mapped faults in the region.
- 1 January 2021
- Published by Elsevier in Materials Today: Proceedings
- Vol. 38, 1467-1469
Determination of the time of formation of particles during arc centrifugal dispersion
- 1 January 2021
- Published by Elsevier in Materials Today: Proceedings
- Vol. 38, 1467-1469
Determination of the time of formation of particles during arc centrifugal dispersion