RLRongzhi Lin

Rongzhi Lin

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Rongzhi Lin has a total of 35 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Contrasting Crustal Signatures Across the 2021 M S 6.0 Luxian (China) Induced Earthquake Revealed by Dense Seismic and Magnetotelluric Data

    by Tianyang Li, Chenguang Wang, Nian Yu, Rongzhi Lin, Yu Jeffrey Gu, Tao Yu, Sheng Zhang

    The 2021 Luxian earthquake has raised concerns about the seismic potential in the southern Sichuan Basin and highlighted the need to gain a further understanding of the area's seismic risks. To elucidate the underlying mechanisms driving the largest recorded hydraulic fracturing‐induced earthquake in China, this study investigates the local seismogenic environment of this notable event through an integrated seismic and magnetotelluric (MT) approach. Utilizing a linear Radon transform‐based mode‐separation method, we derived a 3‐D fine‐resolution velocity structure from dense seismic data. Corroborated by resistivity measurements using a co‐located MT array, our integrated data set provides compelling evidence for a complex upper‐crustal geological structure across the Huayingshan fault (HYS‐F). Apart from the low velocity and resistivity zones at 3–4 km depth within the HYS‐F, consistent with the location of fluid injection sites, we newly discovered a large‐scale low‐resistivity anomaly beneath the mainshock at depths exceeding 15 km. A comparable structural pattern emerges in the southeast of the HYS‐F, where the distribution and focal mechanisms of recently detected earthquakes, including and earthquakes, highlight the combined influence of shallow fluid injection and the ascent of deep metamorphic saline fluids in creating a breeding condition for seismicity. A prominent high‐velocity and high‐resistivity anomaly in the northwest, extending from the surface to depths exceeding 20 km, is potentially linked to intruded ancient materials originating from the lower crust. Our findings reveal that complex geological structures and dynamic processes strongly impact the occurrence and distribution of earthquakes in the Luxian earthquake region.

  • Exploring the Effectiveness of Improved Reverse Time Migration on Multi-GPUs in Unmanned Aerial Vehicle-Based Ground-Penetrating Radar

    by Wuji Wang, Tianyang Li, Wenxin Kong, Cheng Huang, Tao Yu, Rongzhi Lin, Nian Yu

    Unmanned Aerial Vehicle (UAV)-based ground-penetrating radar (GPR) migration imaging encounters substantial challenges in practical applications due to various noise interferences. This study presents a GPU-accelerated imaging algorithm that integrates reverse time migration and total variation denoising (RTM-TVD) to enhance the image quality of UAV-based GPR migration profiles. By employing two acceleration strategies, this algorithm achieves a tenfold increase in speed compared to the traditional RTM method. Additionally, a quantitative evaluation is conducted to assess the effects of terrain roughness, flight altitude, and noise levels on imaging quality. This evaluation utilizes two key metrics: image entropy and the Structural Similarity Index (SSIM). Our findings reveal that the relationship between terrain roughness and noise levels with entropy values follows a positive linear trend, whereas a negative linear trend is observed in SSIM values. The advantages of the RTM-TVD approach are validated through field GPR data collected from the Lemon Creek Glacier in the US.


  • High-dimensional compressive irregular-grid data reconstruction with a fast multidimensional singular spectrum analysis

    by Rongzhi Lin, Siwei Yu, Jianwei Ma, Yi Guo, Mauricio Sacchi

    Spatially irregularly sampled seismic data is unavoidable due to natural obstacles or acquisition designed for compressive sensing. Seismic reconstruction aims to regularize field data and map them from an irregular acquisition grid to regular-grid coordinates. We develop reconstructing high-dimensional arbitrary irregular-grid data with a fast multidimensional singular spectrum analysis (FMSSA) algorithm. The FMSSA filtering algorithm, replacing the traditional multidimensional singular spectrum analysis (MSSA) algorithm, acts as a projection operator to avoid explicitly constructing block Hankel matrices, accelerate the rank-reduction procedure, and reduce the memory load. Our method, the interpolated-FMSSA, can reconstruct data deployed on an irregular grid by introducing an interpolation operator adapted to connect irregular-grid observations and desired regular-grid data without losing accurate spatial coordinates information. In addition, two commonly used Fourier-based methods for irregular-grid data reconstruction, a modified projection onto convex sets algorithm and the fast iterative shrinkage-thresholding algorithm, are used for comparison. Synthetic and real data examples show significant improvement in computational efficiency compared to the traditional I-MSSA method and improvement in reconstruction accuracy compared with the Fourier-based methods for 3D and 5D irregular-grid data reconstruction.


  • Self-Supervised Transfer Learning POCS-Net for Seismic Data Interpolation

    by Yao Chen, Siwei Yu, Rongzhi Lin

    Deep learning has been widely applied to seismic data interpolation. However, most existing methods are based on supervised learning, suffering from limitations such as low generalization ability and the necessity for a labeled training dataset. To address these issues, we propose a novel self-supervised transfer learning framework. The backbone network used is our previously developed projection-onto-convex-sets network (POCS-Net). To our knowledge, this represents the first integration of a data- and model-driven dual approach with a self-supervised learning method. The proposed approach consists of two steps. In the first step, the network undergoes pre-training with synthetic training samples using a supervised learning framework. The parameters obtained from this pre-training are then used to initialize the transfer training in the second step. In the second step, the training dataset is constructed by further down-sampling the already corrupt data. The proposed framework was evaluated through numerical experiments on 2D synthetic and 3D field prestack data, demonstrating its superiority over existing methods. Compared to supervised learning using synthetic dataset, the signal-to-noise ratio of 2D synthetic and 3D field data improved by about 9 dB.


  • Open Access

    Mesozoic intraoceanic subduction shaped the lower mantle beneath the East Pacific Rise

    by Jingchuan Wang, Vedran Lekić, Nicholas C. Schmerr, Yu J. Gu, Yi Guo, Rongzhi Lin

    The Pacific large low-shear-velocity province (LLSVP), as revealed by cluster analysis of global tomographic models, hosts multiple internal anomalies, including a notable gap (~20° wide) between the central and eastern Pacific. The cause of the structural gap remains unconstrained. Directly above this structural gap, we identify an anomalously thick mantle transition zone east of the East Pacific Rise, the fastest-spreading ocean ridge in the world, using a dense set of SS precursors. The area of the thickened transition zone exhibits faster-than-average velocities according to recent tomographic images, suggesting perturbed postolivine phase boundaries shifting in response to lowered temperatures. We attribute this observation to episodes of Mesozoic-aged (250 to 120 million years ago) intraoceanic subduction beneath the present-day Nazca Plate. The eastern portion of the Pacific LLSVP was separated by downwelling because of this ancient oceanic slab. Our discovery provides a unique perspective on linking deep Earth structures with surface subduction.


  • Optimal Seismic Sensor Placement Based on Reinforcement Learning Approach: An Example of OBN Acquisition Design

    by Yi Guo, Rongzhi Lin, Mauricio D. Sacchi

    Seismic acquisition costs are directly associated with the number of sensors used in the survey. Limiting the number of sensors in a seismic survey can be beneficial, especially when sensors are expensive to purchase, deploy, and maintain. This work explores an optimal design method for ocean bottom node (OBN) detector deployment. The proposed method is based on a reinforcement learning approach. We assume access to an initial dataset over the area of study. These data are used to extract an over-complete pre-learned basis library via the proper orthogonal decomposition method, which leads to a fast least-squares seismic data reconstruction algorithm. Then, the sensor selection procedure entails using Q-learning to find the sensor configuration that maximizes the reconstruction quality.


  • Simultaneous deblending and source reconstruction for compressive 3D simultaneous-source acquisition data via interpolated multichannel singular spectrum analysis

    by Rongzhi Lin, Yi Guo, Fernanda Carozzi, Mauricio D. Sacchi

    Iterative rank reduction implemented via multichannel singular spectrum analysis (MSSA) filtering has been proposed for data deblending. The original algorithm is based on the projected gradient-descent method with a projection given by the MSSA filter. Unfortunately, MSSA filters operate on data deployed on a regular grid. We have developed a way to adopt a recently proposed modification to MSSA, interpolated-MSSA, to deblend and reconstruct sources in situations in which the acquired blended data correspond to sources with arbitrary irregular-grid coordinates. In essence, we develop an iterative rank-reduction deblending method that can honor true source coordinates. In addition, we indicate how the technique also can be used for source regularization and interpolation. We compare our algorithm with traditional iterative rank reduction that adopts a regular source grid and ignores errors associated with allocating off-the-grid source coordinates to the desired output grid. Synthetic and field data examples indicate how our method can deblend and reconstruct sources simultaneously.


  • Regularization by denoising for simultaneous source separation

    by Breno Bahia, Rongzhi Lin, Mauricio Sacchi

    Denoisers can help solve inverse problems via a recently proposed framework known as regularization by denoising (RED). The RED approach defines the regularization term of the inverse problem via explicit denoising engines. Simultaneous source separation techniques, themselves being a combination of inversion and denoising methods, provide a formidable field to explore RED. We investigate the applicability of RED to simultaneous-source data processing and introduce a deblending algorithm called REDeblending (RDB). The formulation permits developing deblending algorithms in which the user can select any denoising engine that satisfies RED conditions. Two popular denoisers are tested, but the method is not limited to them: frequency-wavenumber thresholding and singular spectrum analysis. We offer synthetically blended data examples to showcase the performance of RDB.


  • Iterative Deblending of Simultaneous-Source Seismic Data via a Robust Singular Spectrum Analysis Filter

    by Rongzhi Lin, Breno Bahia, Mauricio D. Sacchi

    We solve the simultaneous source separation problem by adopting the projected gradient descent (PGD) method to iteratively estimate the data one would acquire via a conventional seismic acquisition. The projection operator is a windowed robust singular spectrum analysis (SSA) filter that suppresses source interferences in the $f-x$ (frequency-space) domain. We reformulate the SSA filter as a robust optimization problem solved via a bifactored gradient descent (BFGD) algorithm. Robustness becomes achievable by adopting Tukey’s biweight loss function for the design of the robust SSA filter. The SSA filter requires breaking down common-receiver gathers or common offset gathers into small overlapping windows. The traditional SSA method needs the filter rank as an input parameter, which can vary from window to window. The latter has been a shortcoming for the application of classical SSA filtering to complex seismic data processing. The proposed robust SSA filter is less sensitive to rank-selection, making it appealing for deblending applications that require windowing. Additionally, the robust SSA projection provides an effective attenuation of random source interferences during the initial iterations of the PGD method. Comparing classical and robust SSA filters, we also report an acceleration of the PGD method convergence when we adopt the robust SSA filter. Finally, we provide synthetic and real data examples, and discuss heuristic strategies for parameter selection.