Exploration Geophysics - Volume 52, Issue 2, 2021

Random noise attenuation is an important step in seismic signal processing. This paper develops a seismic denoising method which combines the improved complementary ensemble empirical mode decomposition (ICEEMD) and adaptive interval threshold. The seismic data are decomposed into intrinsic mode functions (IMFs) by ICEEMD, which can overcome the problem of uncertain number of modes when adding different random noise as well as the problems of spurious modes and the residual noise from using the ensemble empirical mode decomposition (EEMD) and the complementary ensemble empirical mode decomposition (CEEMD). After the decomposition, the noise in IMFs is filtered out by the adaptive interval threshold. The de-noised data are reconstructed by stacking the filtered IMFs. The proposed approach is validated via the synthetic and field data. The results demonstrate that the approach can effectively improve the de-noising performance.

The deformation and failure development of the overlying strata in a stope is the key to improving the upper limit of coal mining under the aquifers. Taking a mine in Huainan as an example, a comprehensive evaluation method for multi-physics field (strain (stress) field and geoelectric field) dynamic monitoring in an underground borehole was proposed. Based on the analysis of a numerical simulation combined with geological data, a design scheme of the borehole parameters and sensor selection was optimised. In addition, the deformation of the rock strata in different mining periods was monitored by optical cables and resistivity units, and the multi-physics field was utilised to comprehensively analyse the deformation characteristics and development of the overlying strata and to perform correlation analysis on the strain and resistivity. In this way, the relationships between the strain-resistivity correlation coefficient and the rock mass deformation and fracturing were presented. As indicated by the results, the overlying strata are principally subjected to compressive stress in the vertical direction (at approximately 45–90°). Horizontally (at approximately 0–45°), tensile stress plays a major role. In the case of minor rock strata deformation, the strain-resistivity correlation coefficient (R²) ranges from 0.87247 to 0.95682; comparatively, its value abruptly declines to a range between 0.67968 and 0.84675 if the deformation of the rock strata is rather large. Once the fracturing and caving of the rock strata take place, the R² approaches 0. On this basis, the mechanism of the deformation and failure of the overlying strata is further revealed, and a relationship between the rock strata and strain distribution is obtained. Under the action of advance abutment pressure during mining, transverse fractures within and between strata are initially developed. As mining proceeds, vertical fractures are generated due to the formation of a goaf in later periods. According to the strain and resistivity distribution features, the development heights of the 11-2 coal caving zone and the water-flowing fractured zone are determined to be 12.4 and 40–42.35 m, respectively. The results can provide technical references for improving the upper limit of mining in mines with similar geological conditions.

Spectrum induced polarisation (SIP) is a frequency-domain method commonly used in electrical geophysical exploration. However, SIP is very sensitive to the signal-to-noise ratio of the signal, so noise reduction is very important. Independent component analysis can be used to reduce the noise of geophysical exploration data, but it cannot be used when the observed signal has only one dimension. In this paper, an empirical wavelet transform-independent component analysis method is proposed, which can be used for noise reduction in spectrum induced polarisation data. Firstly, the original measurement data are adaptively decomposed into a finite number of intrinsic modal functions by empirical wavelet transform, and then the intrinsic modal functions are selected to construct a virtual noise channel according to their correlation with the induced polarisation signal. Finally, the induced polarisation signal in the multi-dimensional mixed data is extracted by independent component analysis. Our experiment shows results show that this method can effectively remove noise signals and improve the signal-to-noise ratio of induced polarisation data.

When implementing reverse-time migration (RTM), reflected waves that are generated during migration may produce false reflectors. One effective method to remove these artefacts is to apply a de-primary RTM algorithm. This de-primary RTM algorithm uses the Hilbert transform to remove RTM artefacts without explicitly separating wavefields. However, multiply reflected down-going waves, such as surface-related multiples and internal multiples that are generated when simulating source and receiver wavefields, are not excluded during cross-correlation even if we adopt the de-primary RTM technique. Therefore, these down-going multiples may produce troublesome RTM artefacts. In this paper, we attempt to reduce these RTM artefacts by combining the de-primary RTM and impedance-matching techniques. Impedance-matching is only effective for normal incidence, so we also apply an absorbing boundary condition to the free surface to remove surface-related multiples. We evaluate the proposed method through the SEG/EAGE salt model. These numerical examples show that the proposed method effectively reduces RTM artefacts and provides clear migration images.

Full waveform inversion (FWI) is an efficient tool to build the subsurface velocity models. However conventional FWI suffers from the cycle skipping problem, which causes FWI to fail in converging to the global minimum. A good initial model can mitigate this problem, but it is hard to be provided. Low frequencies in the observed data are helpful to recover the low-wavenumber components of the subsurface velocity models, which can provide good initial models. However, field data usually lack low frequencies because of the physical limitation of the instruments or the environmental noise. In addition, multiscale approach may not work well to tackle the cycle skipping problem when there isn’t sufficient low-frequency information in the observed data. Therefore, we proposed an amplitude increment coding-based data selection method to find which parts of the data are mismatched, and set these parts of data to 0 to mitigate the cycle skipping problem. In this case, we use the global-correlation misfit function which behaves better in mitigating the interference of the incorrect amplitude information and highlighting the phase information with weaker nonlinearity. In addition, the amplitude increment coding-based data selection method can be combined with the encoded blended-source scheme to improve computational efficiency. Numerical tests on Marmousi model demonstrate that FWI with an amplitude increment-based data selection method can generate convergent results when the observed data lack low frequencies.

Fresnel zones are helpful for obtaining a high signal-to-noise ratio (S/N)-migrated result. A migrated dip-angle gather provides a simple domain for estimating 2D Fresnel zones for 3D migration. We develop a deep-learning-based technology to automatically estimate Fresnel zones from migrated dip-angle gathers, thus avoiding the cumbersome task of manually checking and modifying the boundaries of the Fresnel zones. A pair of 1D Fresnel zones are incorporated to represent a 2D Fresnel zone in terms of the inline and crossline dip angles because it is difficult to directly extract 2D Fresnel zones from a 2D dip-angle gather. The proposed convolutional neural network (CNN) is established by modifying VGGNet. As picking boundaries of the Fresnel zones is a regression problem, we remove the last soft-max layer from the VGGNet. The last three convolution layers and a pooling layer are also removed because the feature maps are small enough. To improve the contrast and definition, we enhance the features of the reflected events in the dip-angle gather. Data normalisation is carried out to accelerate the training process using a simple-rescaling method before training the modified VGGNet. Field data examples demonstrate the effectiveness and efficiency of the proposed method.

The finite difference method is widely used in seismic wave numerical simulation, reverse time migration and full waveform inversion. However, the numerical dispersion problem seriously affects the results of seismic imaging and inversion. Based on this, we introduced cosine-combined window function (CCWF) used in harmonic analysis of the power system and compared the amplitude response and error properties with other window functions. Then, an optimised CCWF and a new weighting method are proposed, which results in finite-difference operators with not only large spectral coverage but also small precision error fluctuation. In this paper, the analysis shows the good stability of using finite difference operators. Finally, we perform numerical forward modelling, which denotes that our method is superior than other optimum methods. From an economic point of view, this method will effectively reduce the computation cost and improve efficiency.

Due to high imaging precision and adaptability for steep-dip interfaces, reverse time migration (RTM) has become the preferred choice in seismic imaging. However, low-frequency noise is a common problem in RTM and effects the clarity of the final image. The noise is formed by the cross-correlation of source and receiver wavefields that propagate along the same paths. A wavefield decomposition imaging condition (WDIC) can eliminate this low-frequency noise by separating the wavefield into up–down or left–right directions and then cross-correlating source and receiver wavefields along the different propagation paths. Nevertheless, many backscattered waves propagating along horizontal, nearly horizontal, vertical and nearly vertical directions exist in vertical seismic profile (VSP) data. These waves are difficult to separate and generate the residual low-frequency noise when applying WDIC to VSP RTM. To overcome this, Poynting vector filtering is combined with WDIC to attenuate the noise. Several numerical case studies testify to the effectiveness of the analysis and method.