Jingnan Li

Jingnan Li

Auto-generated profile, based on publicly available information

Jingnan Li

SINOPEC Geophysical Research Institute, Nanjing, CHINA

Research Focus & Metrics

Academic Career Length

12Years (2015-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
22 January 2020 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
XCXintao Chai
Xintao Chai
GTGenyang Tang
Genyang Tang
SWShangxu Wang
Shangxu Wang
RPRonghua Peng
Ronghua Peng
WCWei Chen
Wei Chen
JLJingnan Li
Jingnan Li
Inspired by image-to-image translation, we applied deep learning (DL) to regularly missing data reconstruction, aimed at translating incomplete data into their corresponding complete data. With this purpose in mind, we first construct a network architecture based on an end-to-end U-Net convolutional network, which is a generic DL solution for various tasks. We then meticulously prepare the training data with both synthetic and field seismic data. This article is implemented in Python based on Keras (a high-level DL library). We described the network architecture, the training data, and the training settings in detail. For training the network, we employed a mean-squared-error loss function and an Adam optimization algorithm. Next, we tested the trained network with several typical data sets, achieving good performances (even in the presence of big gaps) and validating the feasibility, effectiveness, and generalization capability of the assessed framework. The feature maps for a sample going through the well-trained network are uncovered. Compared with the f-x prediction interpolation method, DL performs better and is capable of avoiding several assumptions (e.g., linearity, sparsity, etc.) associated with conventional interpolation methods. We demonstrated the influences of the network depth, the kernel size of the convolution window, and the pooling function on the DL results. We applied the trained network to dense data reconstruction successfully. The proposed method can overcome noise to some extent. We finally discussed some practical aspects and extensions of the evaluated framework.

26 August 2015 in Geophysics
Verified
Journal Article
Closed access
SWShoudong Wang
Shoudong Wang
DYDengfeng Yang
Dengfeng Yang
JLJingnan Li
Jingnan Li
HSHuijuan Song
Huijuan Song
We have developed a novel method for Q estimation. As a seismic wave propagates through subsurface viscoelastic media, seismic attenuation decreases the amplitude and reduces the bandwidth, which leads to variation of the logarithmic spectral area of seismic wave. According to this property, we have derived an analytical relation between Q factor and the logarithmic spectral area difference of seismic wave. To improve the accuracy of Q estimation, we took account of the amplitude loss due to energy partitioning from transmission. Tests of the method on synthetic models determined the validity of our method, and it was insensitive to different types of noises and evaluated density. A further comparison between the presented method and two widely used Q estimation methods, namely, the logarithmic spectral ratio and centroid frequency shifting methods, with a synthetic model containing random noise demonstrated that our method is the more robust, in these situations, among the three approaches. Then, we also compared the effect of bandwidth and time-window selection on the Q estimation for these three methods, and the result revealed that our method is less sensitive to the bandwidth and time window than the other two methods. Application of this method to real zero-offset vertical seismic profile data shows a reasonable correlation between estimated interval-Q and interval velocities, which further indicates the validity of the method.

18 November 2015 in Geophysics
Verified
Journal Article
Closed access
XCXintao Chai
Xintao Chai
SWShangxu Wang
Shangxu Wang
JWJianxin Wei
Jianxin Wei
JLJingnan Li
Jingnan Li
HYHanjun Yin
Hanjun Yin
We have developed a method of nonstationary sparse reflectivity inversion (NSRI) that directly retrieves the reflectivity series from nonstationary seismic data without the intrinsic instability associated with inverse Q filtering methods. We have investigated the NSRI performance in the presence of input error (e.g., the phase shift and the peak frequency of the wavelet), which determined that NSRI results are reasonable in the case of moderate error. NSRI was then applied to data collected from a laboratory physical model made of highly attenuating media, for which true reflectivities were known, but the wave propagation and the Q filtering mechanism were not. Analysis of data from the physical model, therefore, represented a blind test for evaluating the effectiveness and accuracy of NSRI. The physical model had a specified spatial scale of 1:5000 (relative to the field scale) and an approximate 1:1 velocity scale. Because the Q input is required by NSRI, attenuation of the P-wave was measured in the laboratory at ultrasonic frequencies with a pulse transmission technique and the spectral ratio method. Although multiples, side reflections from the model boundaries, diffractions, and other event types were clearly observed from the raw data, no preprocessing was done to avoid affecting the Q filtering effects. Results from the physical modeling data demonstrated that the derived formula for an attenuated seismogram was correct and estimated Q values were reliable. The functions and advantages of NSRI were confirmed. The Q compensated seismogram generated by NSRI was superior to that generated using gain-limited inverse Q filtering. We have also investigated NSRI’s capabilities in analyzing a marine data set from the Gulf of Suez. These examples provided a basis for discussing assumptions and limitations of NSRI.

23 May 2016 in Journal of Geophysics and Engineering
Verified
Journal Article
Closed access
PDF file
JLJingnan Li
Jingnan Li
SWShangxu Wang
Shangxu Wang
DYDengfeng Yang
Dengfeng Yang
CDChunhui Dong
Chunhui Dong
YTYonghui Tao
Yonghui Tao
YZYatao Zhou
Yatao Zhou
Seismic wave propagation in subsurface media suffers from absorption, which can be quantified by the quality factor Q. Accurate estimation of the Q factor is of great importance for the resolution enhancement of seismic data, precise imaging and interpretation, and reservoir prediction and characterization. The centroid frequency shift method (CFS) is currently one of the most commonly used Q estimation methods. However, for seismic data that contain noise, the accuracy and stability of Q extracted using CFS depend on the choice of frequency band. In order to reduce the influence of frequency band choices and obtain Q with greater precision and robustness, we present an improved CFS Q measurement approach—the weighted CFS method (WCFS), which incorporates a Gaussian weighting coefficient into the calculation procedure of the conventional CFS. The basic idea is to enhance the proportion of advantageous frequencies in the amplitude spectrum and reduce the weight of disadvantageous frequencies. In this novel method, we first construct a Gauss function using the centroid frequency and variance of the reference wavelet. Then we employ it as the weighting coefficient for the amplitude spectrum of the original signal. Finally, the conventional CFS is adopted for the weighted amplitude spectrum to extract the Q factor. Numerical tests of noise-free synthetic data demonstrate that the WCFS is feasible and efficient, and produces more accurate results than the conventional CFS. Tests for noisy synthetic data indicate that the new method has better anti-noise capability than the CFS. The application to field vertical seismic profile (VSP) data further demonstrates its validity5.

16 February 2017 in Pure and Applied Geophysics
Verified
Journal Article
Closed access
PDF file
JLJingnan Li
Jingnan Li
SWShangxu Wang
Shangxu Wang
JWJingbo Wang
Jingbo Wang
CDChunhui Dong
Chunhui Dong
Sanyi YuanSanyi Yuan
Sanyi Yuan

20 January 2020 in Geophysics
Verified
Journal Article
Closed access
DYDengfeng Yang
Dengfeng Yang
JLJun Liu
Jun Liu
JLJingnan Li
Jingnan Li
DLDaoli Liu
Daoli Liu
The centroid frequency shift (CFS) method is a widely used Q estimation approach. However, the CFS approach assumes that the amplitude spectrum of a source wavelet has a particular shape, which can cause systematic error in Q estimation. Moreover, the amplitude spectrum at high and low frequencies is susceptible to random noise, which can reduce the robustness of Q estimation using the CFS method. To improve the accuracy and robustness of Q estimation, we have developed a Q extraction method using the bisection algorithm based on the centroid frequency shift of power spectrum (BPCFS). In the BPCFS approach, we first obtain the source and the received wavelet. Then, we calculate the centroid frequency of the attenuated wavelet and that of the received wavelet from the power spectrum. Based on the obtained centroid frequencies, we establish an equation containing only one variable — the Q factor. Introducing the Jeffrey divergence to measure the attenuation of the power spectrum, we prove that this equation has only one root when Q is greater than zero. The root of this equation, which is the desired Q factor, is obtained through the bisection algorithm; we do not make any assumption about the shape of the amplitude spectrum. The noise-free numerical tests indicate that the BPCFS gives more accurate results than the CFS, which demonstrates that the shape of the wavelet spectrum has little effect on the Q estimation accuracy for BPCFS. Gaussian random and blue noise tests also show that the stability of BPCFS is better than that of CFS. The frequency band selection for BPCFS is also more flexible when the amplitude spectrum of the seismic wavelet is band limited. The application to real vertical seismic profile data further demonstrate the effectiveness and feasibility of the new method.