Co-station coherent stack enhancement method for low signal-to-noise ratio active source seismic surface wave

Abstract

The present application relates to the technical field of geophysical exploration, and discloses a kind of active source seismic surface wave's common station pair coherent stack enhancement methods for low signal-to-noise ratio.It includes: generating Green function according to the relative relationship of seismic source position and station pair: when station pair is located on the same side of seismic source, the cross-correlation of the seismic record of two stations in station pair is calculated, and cross-correlation Green function is generated;when station pair is located on the opposite side of seismic source, the cross-correlation of the seismic record of two stations in station pair and the seismic record of reference station at seismic source is calculated respectively, and then convolution calculation is carried out, and convolution Green function is generated;Based on the cross-correlation Green function and convolution Green function after stacking, frequency dispersion energy diagram is calculated respectively;Stack two kinds of frequency dispersion energy diagram to generate final frequency dispersion diagram, which is used to extract frequency dispersion curve.The active source seismic surface wave signal common station pair coherent stack enhancement method proposed in the present application effectively improves the surface wave signal-to-noise ratio and frequency dispersion diagram quality without increasing the cost of data acquisition.

Description

Technical Field

[0001] This invention relates to the field of geophysical exploration technology, specifically to a method for coherent stacking enhancement of co-station pairs for low signal-to-noise ratio active source seismic surface waves. Background Technology

[0002] In the field of near-surface geophysical exploration, active source surface wave (MASW) exploration technology has been widely used in engineering geological surveys, environmental geological investigations, and urban underground space exploration due to its advantages such as low cost, non-destructive testing, and high resolution. Park et al. first proposed the Multichannel Analysis of Surface Waves (MASW) method, which standardized the selection criteria for field data acquisition parameters and performed dispersion analysis on single-shot active source surface wave records. Based on the dispersion map, the dispersion curve was extracted and inverted to obtain the average one-dimensional layered structure beneath the array. A key assumption of this method is that the velocity structure beneath the array has no lateral variation. Therefore, the array length is required to be as short as possible while meeting the detection depth. A two-dimensional model is constructed by using a series of one-dimensional models through a rolling arrangement. Hayashi & Suzuki introduced common-center cross-correlation stacking to improve the accuracy and resolution of subsurface shear wave velocity structure inversion. Li et al. proposed a multichannel surface wave analysis method based on short array stacked cross-correlation gathers, aiming to improve horizontal resolution using short arrangements without reducing the detection depth. Ikeda et al. used continuous wavelet transform to superimpose surface wave information excited by different sources to improve the signal-to-noise ratio of the data, which helps to extract more reliable dispersion curves.

[0003] However, traditional multichannel surface wave (MASW) analysis methods rely on rolling arrangement observations, resulting in relatively low data acquisition efficiency. With the large-scale application of nodal seismographs, more efficient full-arrangement observation methods have become increasingly common. Existing interferometric stacking methods cannot fully utilize observation data, especially when processing low signal-to-noise ratio data, where the enhancement effect is limited, leading to missing or non-convergent dispersion energy, and accurately extracting dispersion information remains difficult.

[0004] Therefore, this invention combines a passive source surface wave three-station interferometry method with active source surface waves, using seismic interferometry and co-station stacking techniques to synergistically enhance the signal-to-noise ratio of surface wave data by stacking different excitation records multiple times, which can significantly improve the dispersion quality and ultimately achieve a deeper and more reliable understanding of near-surface geological structures. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a coherent superposition enhancement method for co-station pairs of surface waves from active source seismic sources with low signal-to-noise ratio (SNR). This method involves interferometric processing of cross-correlation or convolution between observation records from different stations in the active source excitation records to extract the surface wave signal between the two stations. Furthermore, the results of multiple active source excitations are superimposed to further improve the SNR, significantly enhancing the surface wave signal quality. The resulting dispersion map exhibits better continuity and convergence, as well as a wider bandwidth, effectively improving the reliability of dispersion extraction and facilitating more detailed research on underground structures.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0007] A method for coherent stacking enhancement of surface waves from active-source seismic sources with low signal-to-noise ratio using shared-station pairs includes:

[0008] Seismic records triggered multiple times by active sources were obtained from multiple stations in a linear array.

[0009] For each earthquake record, stations are paired up to form station pairs, and a Green's function is generated based on the relative relationship between the epicenter location and the station pair:

[0010] When a pair of stations is located on the same side of the epicenter, the cross-correlation of the seismic records of the two stations in the pair is calculated, and the cross-correlation Green's function is generated.

[0011] When the stations are located on opposite sides of the source, the cross-correlation between the seismic records of the two stations in the station pair and the seismic record of the reference station at the source is calculated separately, and then convolution calculation is performed to generate the convolution Green's function.

[0012] For all seismic records of each station pair, weighted superposition of cross-correlation Green's function and convolution Green's function are performed respectively;

[0013] Based on the superimposed cross-correlation Green's function and the convolutional Green's function, the dispersion energy map is calculated respectively;

[0014] The final dispersion map is generated by superimposing the two types of dispersion energy maps, which is used to extract the dispersion curve.

[0015] In one embodiment, when the station pair is located on the same side of the epicenter, the cross-correlation of the seismic records of the two stations in the station pair is calculated to generate a cross-correlation Green's function, specifically including:

[0016] The spatial location of the epicenter is The two stations in the station pair are denoted as platform i and platform j, and their spatial positions are respectively: ;

[0017] when At that time, the cross-correlation Green's function for:

[0018] ;

[0019] when At that time, the cross-correlation Green's function for:

[0020] ;

[0021] These represent the seismic records in the frequency domain at platforms i and j, respectively. They represent The complex conjugate, It represents angular frequency.

[0022] In one embodiment, when the station pair is located on opposite sides of the epicenter, the cross-correlation between the seismic records of the two stations in the station pair and the seismic record of the reference station at the epicenter is calculated, and then convolution calculation is performed to generate a convolution Green's function, specifically including:

[0023] The spatial location of the epicenter is The two stations in the station pair are denoted as platform i and platform j, and their spatial positions are respectively: ;

[0024] Convolution Green's function for:

[0025] ;

[0026] These represent the seismic records in the frequency domain at platforms i and j, respectively. This represents the complex conjugate of the frequency domain seismic records from the reference station at the epicenter. It represents angular frequency.

[0027] In one embodiment, the weighted superposition of cross-correlation Green's functions and convolution Green's functions for all seismic records of each station pair specifically includes:

[0028] ;

[0029] ;

[0030] The Green's function represents the superposition of cross-correlation. The Green's function represents the convolution after stacking. This represents the weight of the cross-correlation Green's function corresponding to the k-th excitation of the active source. This represents the weight of the convolutional Green's function corresponding to the k-th excitation of the active source. and These represent the cross-correlation Green's function and the convolution Green's function, respectively. It represents angular frequency.

[0031] In one embodiment, The calculation method is as follows:

[0032] ;

[0033] The calculation method is as follows:

[0034] ;

[0035] These represent the seismic records in the frequency domain at platforms i and j, respectively. They represent The complex conjugate, These represent the frequency domain seismic records of the reference station at the epicenter and their complex conjugates, respectively. It represents angular frequency.

[0036] In one embodiment, the calculation of the dispersion energy map based on the superimposed cross-correlation Green's function and the convolutional Green's function specifically includes:

[0037] ;

[0038] ;

[0039] in, These represent the dispersion energy maps calculated using the cross-correlation Green's function and the convolution Green's function, respectively. Generally refers to specific transformation operators corresponding to different calculation methods. Represents angular frequency. Indicates the scanning speed. The Green's function represents the superposition of cross-correlation. This represents the convolution Green's function after the convolution is performed.

[0040] Compared with the prior art, the beneficial technical effects of the present invention are:

[0041] The proposed coherent superposition enhancement method for active-source seismic surface wave signals from multiple stations effectively improves the signal-to-noise ratio (SNR) and dispersion map quality without increasing data acquisition costs. This method involves the generation and processing of cross-correlation components and convolutional integrals: After each source excitation, all stations are paired to form station pairs, and their empirical Green's function is calculated. If the station pairs are on the same side of the source, the cross-correlation function of the waveforms from the two receiving stations is directly calculated to generate the cross-correlation components of the station pairs. If the station pairs are on opposite sides of the source, the cross-correlation between the waveforms from the two receiving stations and the record from the reference station at the source point is calculated separately, and then these two results are convolved to generate the convolutional integral of the station pairs. Based on this, for each station pair, all cross-correlation components and convolutional integrals obtained from all source excitations are weighted and superimposed; then, the superimposed dispersion map is calculated; finally, the two results are superimposed again to generate a high-quality dispersion map for extracting dispersion curves. This method, through the synergistic use of interferometry and multiple superposition techniques, significantly suppresses incoherent noise and improves the SNR. Furthermore, the strategy of calculating and then superimposing the dispersion maps of the cross-correlation components and convolutional components separately helps to extract weak but effective surface wave signals more effectively, fully integrating the advantages of each component. In the implementation case, the enhanced dispersion map, such as... Figure 6 d) Figure 7 In the d), compared to the original dispersion result, such as Figure 6 a) Figure 7 (a) In this invention, the continuity is significantly improved, and the fundamental and higher-order surface waves transform from modal aliasing to clearly identifiable patterns, exhibiting better convergence and higher reliability of dispersion energy distribution over a wider frequency band. Therefore, this invention significantly improves the quality and reliability of dispersion maps, providing strong support for accurately extracting surface wave dispersion information and further refining the velocity structure of underground media. Attached Figure Description

[0042] Figure 1 This is a schematic diagram illustrating the principle of the coherent superposition enhancement method for shared-station pairs in an embodiment of the present invention. Wherein, a) represents three typical source-station pair relationships; b) represents the interference modes of different station pairs along the active source survey line; These represent the spatial locations of the earthquake source and the two stations, respectively. ⊖ indicates cross-correlation calculation, and ⊕ indicates convolution calculation.

[0043] Figure 2 This is a schematic diagram of the active source seismic survey line observation system and the time-frequency characteristics of the seismic source in an embodiment of the present invention. Wherein, a) represents the spatial distribution of seismic stations and seismic sources; b) represents the frequency-time curve of the swept source.

[0044] Figure 3This is an active source excitation record diagram in an embodiment of the present invention. Wherein, a), c), and e) represent the original seismic records excited by the source at stations 1, 25, and 50, respectively, and b), d), and f) represent the corresponding wavelet compression waveforms.

[0045] Figure 4 The diagram shows the coherence results based on a single active source excitation. In this diagram, a) shows the results from station 1 (endpoint of the survey line), all of which are cross-correlation components; b) shows the results from station 25 (middle of the survey line), with cross-correlation components within 25 m and convolutional components outside; c) shows the results from station 50 (endpoint of the survey line), all of which are cross-correlation components. Figure 4 The interference results of station 1 with all other stations are shown. The red dashed lines correspond to reference apparent velocities of 100 m / s and 500 m / s.

[0046] Figure 5 The image shows the coherent superposition results of all active source excitation records. Where a) represents the cross-correlation component; b) represents the convolution component.

[0047] Figure 6 This is a comparison chart of the dispersion energy calculated from stations 1 to 15 based on different data sources. Among them, a) is the dispersion plot calculated from a single source record; b) is the result of superimposing the cross-correlation components of all sources; c) is the result of superimposing the convolution components of all sources; d) is the result of superimposing the dispersion plots of b) and c).

[0048] Figure 7 This is a comparison chart of the dispersion energy calculated from stations 12 to 26 based on different data sources. Among them, a) is the dispersion plot calculated from a single source record; b) is the result of superimposing the cross-correlation components of all sources; c) is the result of superimposing the convolution components of all sources; d) is the result of superimposing the dispersion plots of b) and c). Detailed Implementation

[0049] A preferred embodiment of the present invention will now be described in detail with reference to the accompanying drawings.

[0050] This invention proposes an enhancement method for active-source excited seismic surface wave signals, used to recover high signal-to-noise ratio (SNR) surface wave signals from low SNR data, thereby accurately extracting surface wave dispersion information. Active-source surface wave exploration technology is an indispensable tool in near-surface engineering geological exploration. It typically utilizes artificially controlled sources such as hammer, tamping, and sweep-frequency sources to generate seismic waves, and extracts surface wave dispersion information by analyzing geophone array records at the surface, thereby inverting the shear wave velocity structure of the subsurface medium. Therefore, high SNR surface wave data is fundamental for accurate dispersion information extraction. However, due to factors such as limited source excitation energy, energy loss caused by poor coupling between the source and the surface, and environmental background noise interference, the raw surface wave signals acquired in actual exploration operations often face insufficient SNR. The active-source seismic surface wave enhancement method proposed in this invention reconstructs surface wave signals between different station pairs using seismic interferometry and co-station pair stacking techniques, significantly improving the SNR. This method significantly improves the quality of surface wave data, directly enhancing the accuracy and robustness of dispersion curve extraction, and laying a solid data foundation for subsequent inversion of high-precision near-surface shear wave velocity structures.

[0051] The proposed method for coherent superposition enhancement of active-source seismic surface wave signals using co-station pairs is based on seismic interferometry, and its principle is as follows: According to convolution model theory, the array observation records can be represented as:

[0052] (1)

[0053] in, These represent the time-varying observation data (seismic record), source wavelet, Green's function, instrument response, and noise, respectively. Transforming this to the frequency domain, it can be written as:

[0054] (2)

[0055] From formulas (1) and (2), it can be seen that when the earthquake source... Poor coupling, insufficient effective output power, or environmental noise When it is strong, it will cause the observation data The signal-to-noise ratio decreases, making it difficult to extract the Green's function from the underground medium. This poses challenges to the study of underground structures. Conventional methods for accurately extracting the Green's function include improving source coupling or increasing source energy, and conducting data acquisition when environmental noise levels are low (e.g., at night). However, implementing these measures places higher demands on the construction environment, significantly increases data acquisition costs, and reduces acquisition efficiency, thus presenting numerous limitations.

[0056] This invention improves the signal-to-noise ratio (SNR) through data processing based on signal correlation and noise randomness without increasing data acquisition costs. The basic principle is as follows: Figure 1 As shown. From formula (2), the Green's function between the source and the station can be written as:

[0057] (3)

[0058] Figure 1 It shows the distribution of the observation system, that is, multiple seismic stations are linearly distributed, forming a dense survey line, and the seismic source is excited sequentially at different station locations. Figure 1 a) in the diagram illustrates three possible spatial relationships between the seismic source and the stations. The spatial locations of the seismic source and the two stations are as follows: , Figure 1 b) shows different station pair combinations, and the Green's function between different station pairs can be extracted using seismic interferometry.

[0059] 1. Cross-correlation components

[0060] When the stations are located on the same side of the seismic source, only the observation data from the two stations need to be used to calculate the cross-correlation of the waveform and extract the Green's function of the surface wave. For example... From the convolution model, we can see that:

[0061] (4)

[0062] The same type of instrument is generally used for observation. and The same applies to surface wave imaging, but since surface wave imaging utilizes its dispersion information, it is necessary to normalize the surface wave energy at different frequencies separately, so only phase information is considered. (Denominator) Being a real number does not affect the phase; the molecule Plays a major role. In actual data, it is impossible to obtain data from observational data. Separation noise Therefore, the noisy Green's function, ignoring energy distributions at different frequencies, can be written as:

[0063] (5)

[0064] Similarly, when At that time, there were:

[0065] (6)

[0066] This is the cross-correlation Green's function, also known as the cross-correlation component.

[0067] 2. Convolutional integral

[0068] When stations are located on either side of the seismic source, the station at the seismic source is used as a bridge, and its records are treated as source wavelets to calculate the convolution and extract the surface wave Green's function. At that time, as can be seen from the convolution model:

[0069] (7)

[0070] That is, the Green's function from station i to station j is equal to the convolution of the Green's functions from the source to the two stations.

[0071] Substituting formula (4) into formula (7) gives us...

[0072] (8)

[0073] Similar to the cross-correlation components, the response functions of the same instrument cancel each other out. Ignoring the energy term and simplifying equation (8), the noisy Green's function can be expressed as:

[0074] (9)

[0075] This is the Green's function for convolution, also known as the convolution integral.

[0076] 3. Overlay of shared stations

[0077] For any two stations in each active source excitation record, a noisy Green's function can be obtained based on the cross-correlation component or convolutional integral. Therefore, the number of noisy Green's functions obtainable for any station is the same as the number of active source excitations. Assuming the noise follows a random distribution, multiple data superpositions can effectively suppress environmental noise and enhance coherent signals, thereby improving the signal-to-noise ratio. This invention uses a weighted superposition of Green's functions:

[0078] (10)

[0079] in, This represents the weight of the Green's function corresponding to the k-th excitation of the active source. For the cross-correlation component, different weights are applied to different frequencies using the energy recorded by observations at stations i and j. For the convolution component, since the waveform observed at station s is used as the wavelet in the calculation, data from three stations are required when calculating the weights. The specific formula is as follows:

[0080] (11)

[0081] The number of cross-correlation Green's functions depends on the number of effective source excitations to the outer region from the station, while the number of convolution Green's functions depends on the number of effective source excitations to the inner region from the station. Therefore, the proportions of these two Green's functions in different station pairs may vary significantly: at smaller station spacing, the cross-correlation component dominates, far exceeding the convolution component; as the station spacing increases, the convolution component gradually becomes dominant and significantly exceeds the cross-correlation component. Furthermore, the enhancement effects of the cross-correlation component and the convolution component also differ under different observation scales and data acquisition conditions. If the effective surface wave signal mainly exists in one of the weaker or lower-proportion components, directly superimposing all the Green's functions of that station pair may be masked by the strong signal of the dominant component or other noise, leading to the loss or distortion of key dispersion information. To avoid these problems, this invention superimposes the two components separately to enhance the effective surface wave signal.

[0082] 4. Dispersion calculation and superposition

[0083] Dispersion map calculation is a crucial step in accurately extracting surface wave dispersion information. To meet diverse research needs, existing research has developed various algorithms for dispersion map calculation. Although the specific implementation details differ, these algorithms share a common core process: first, a sub-array is defined with a certain aperture centered on a specific station; then, at a predetermined target angular frequency... The above iterates through all possible velocity values ​​within the range of phase velocities at fixed intervals. Based on different A dispersion map is obtained by performing specific transformations on the observed data. The surface wave dispersion curve can be picked out based on the energy peak of the dispersion map. The dispersion map is calculated based on the cross-correlation components and convolutional quantities obtained by superposition, as shown in formula (12):

[0084] (12)

[0085] in These represent the dispersion plots calculated using the cross-correlation components and the convolutional integral, respectively. This refers generally to specific transformation operators corresponding to different calculation methods, such as the phase-shifting method (Park et al. 1998), whose corresponding calculation formula F is defined as... Subsequently, the two dispersion energy maps from the same central station were superimposed to obtain the final high-quality dispersion map, which was used to extract the dispersion curve.

[0086] The proposed method of surface wave coherent superposition enhancement using a shared station was applied to the detection of underground air-raid shelters to demonstrate its effectiveness in real-world data. Details of the experimental observation system can be found here. Figure 2A total of 50 short-period nodal seismographs were deployed along a 50-meter survey line, with an average spacing of approximately 1 meter between stations. Active source excitation was performed at 45 stations within the survey line area, using time-frequency adjustable swept-frequency signals. The swept-frequency parameters are detailed below. Figure 2 (b) The output wavelet frequency increases linearly from 0 Hz to 160 Hz between 0 s and 20 s, maintains the peak frequency for 5 s (from the 20th s to the 25th s), and then decreases linearly to 0 Hz between the 25th s and the 45th s.

[0087] Figure 3 The image shows the original seismic records and their wavelet compression waveforms from three sets of active source-excited seismic records located at stations 1, 25, and 30. The original swept-frequency records presented complex waveforms that made it difficult to directly identify different signals. After wavelet compression, the signal energy was clearly focused, with surface waves exhibiting the strongest energy. However, these waves were still subject to strong environmental noise interference, and the signal-to-noise ratio dropped sharply after the offset exceeded 30 m, with the effective signal being obscured by noise. Subsequently, the stations were paired, and cross-correlation components or convolutional quantities were calculated using interferometry to extract the surface wave signals. The coherence results for station 1 with all other stations are shown in [reference needed]. Figure 4 . Figure 4 a) and Figure 4 In c), the source excitation location is located at the end of the survey line, all coherent components are cross-correlation components, and the signal-to-noise ratio of the near-station spacing data is significantly higher than that of the far-station spacing data. Figure 4 In b), the source is located in the center of the survey line. Taking station 1 as a reference, stations within 25 m are located on the same side of the source, and the cross-correlation components are calculated. When the station spacing exceeds 25 m, the station pairs are located on both sides of the source. The station at the source needs to be used as a bridge to calculate the convolution component. The signal-to-noise ratio does not change significantly with the station spacing.

[0088] In fact, after the above interferometric processing, the excitation records from different sources all reflect the Green's function between the station pairs. Therefore, the coherence results of the same station pair can be superimposed. The cross-correlation components and convolutions after superposition are as follows: Figure 5 As shown in the figure, the results indicate that the signal stability of both the cross-correlation component and the convolution component is significantly enhanced after multiple superpositions, and their signal-to-noise ratio (SNR) is significantly improved compared to the single-excitation coherent result. However, when the inter-station spacing exceeds 30 m, the SNR of the cross-correlation component still decreases rapidly, mainly due to two reasons: first, the source energy is relatively weak and therefore attenuates sharply after propagating to 30 m, resulting in a decrease in effective signal coherence; second, as the inter-station spacing increases, the number of sources outside the station decreases, and the number of superpositions also decreases, limiting the noise suppression effect of the superposition process. Conversely, the convolution component shows that when the inter-station spacing is very close, the number of sources inside the station is small, and the number of superpositions is very low. The number of superpositions increases with the increase of the inter-station spacing. Therefore, at large inter-station spacings, the SNR of the convolution component is significantly better than that of the cross-correlation component.

[0089] Using stations 1 to 15 as a subarray, surface wave dispersion maps were calculated and compared using active source records and co-station records for the superimposed enhanced results. Figure 6 ). Figure 6 In section a), the dispersion map calculated directly from the original active source record is interrupted between 20 and 40 Hz, and the resolution below 20 Hz is very low, easily introducing large errors during dispersion curve extraction. The continuity is better in the 40 to 70 Hz range, but it is easily identified as a fundamental surface wave. The dispersion map calculated from the enhanced data, whether for cross-correlation components or convolutional components, such as... Figure 6 Both b) and c) in the above examples exhibit better continuity and higher resolution, demonstrating good consistency, and the fundamental and higher-order surface wave dispersions are clearly distinguishable, providing more accurate dispersion information. Note the subtle differences in the performance of the cross-correlation component and the convolution component across the fundamental and higher-order frequency bands, as well as at certain frequencies; superimposing the two again yields... Figure 6 (d) further integrates their respective advantageous frequency bands to obtain high-quality dispersion over a wide bandwidth.

[0090] Similarly, the dispersion diagrams of the subarrays of stations 12 to 26 are calculated as follows: Figure 7 The dispersion map of this array is compared to Figure 6 Higher noise levels Figure 7 While the active source dispersion map (a) in the figure shows good continuity, it actually suffers from severe aliasing of dispersion energy clusters between the fundamental and higher-order modes, making it difficult to distinguish between different modes. Furthermore, the uneven resolution of energy clusters at different frequencies indicates poor convergence. See also... Figure 7 (b) The superimposed enhanced cross-correlation component performs better in the fundamental mode, but the higher-order modes weaken to the point of being indistinguishable above 40 Hz. In contrast, the higher-order modes of the convolution component exhibit good continuity and convergence over a wider frequency range. (See also...) Figure 7 (c) However, the stability of the fundamental surface wave is insufficient at low frequencies below 15 Hz. The result of superimposing the two components shows the best performance in both the fundamental and higher-order modes, indicating that it effectively extracts and fuses the effective information in the two components, making full use of the potential in the original data, thereby obtaining a high-quality dispersion curve with higher accuracy and reliability.

[0091] The above examples and comparative results demonstrate that the coherent superposition method for co-station pairs proposed in this invention can effectively mine and utilize low signal-to-noise ratio surface wave signals in active source data to obtain high-quality dispersion maps, effectively improving the accuracy and reliability of dispersion curve extraction, and providing important support for the study of underground fine structures.

[0092] This invention proposes an enhancement method for active-source excited seismic surface wave signals, capable of recovering high signal-to-noise ratio (SNR) surface wave signals from low SNR data, thereby accurately extracting surface wave dispersion information. Active-source surface wave exploration technology is an indispensable tool in near-surface engineering geological exploration. It typically utilizes artificially controlled sources such as hammer, tamping, and frequency-sweeping sources to generate seismic waves, and extracts surface wave dispersion information by analyzing geophone array records at the surface, subsequently inverting to obtain the shear wave velocity structure of the subsurface medium. Therefore, high SNR surface wave data is fundamental for accurate dispersion information extraction. However, due to factors such as limited source excitation energy, energy loss caused by poor coupling between the source and the surface, and environmental background noise interference, the raw surface wave signals acquired in actual exploration operations often face insufficient SNR. The active-source seismic surface wave enhancement method proposed in this invention reconstructs surface wave signals between different station pairs using seismic interferometry and co-station pair stacking techniques, significantly improving the SNR. This method significantly improves the quality of surface wave data, directly enhancing the accuracy and robustness of dispersion curve extraction, and laying a solid data foundation for subsequent inversion of high-precision near-surface shear wave velocity structures.

[0093] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the invention. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0094] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0095] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention, and no reference numerals in the claims should be construed as limiting the scope of the claims.

[0096] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for coherent stacking enhancement of surface waves from active-source seismic sources with low signal-to-noise ratio using co-station pairs, characterized in that, include: Seismic records triggered multiple times by active sources were obtained from multiple stations in a linear array. For each earthquake record, stations are paired up to form station pairs, and a Green's function is generated based on the relative relationship between the epicenter location and the station pair: When a pair of stations is located on the same side of the epicenter, the cross-correlation of the seismic records of the two stations in the pair is calculated, and the cross-correlation Green's function is generated. When the stations are located on opposite sides of the source, the cross-correlation between the seismic records of the two stations in the station pair and the seismic record of the reference station at the source is calculated separately, and then convolution calculation is performed to generate the convolution Green's function. For all seismic records of each station pair, weighted superposition of cross-correlation Green's function and convolution Green's function are performed respectively; Based on the superimposed cross-correlation Green's function and the convolutional Green's function, the dispersion energy map is calculated respectively; The final dispersion map is generated by superimposing the two types of dispersion energy maps, which is used to extract the dispersion curve.

2. The method for coherent superposition enhancement of low signal-to-noise ratio active source seismic surface waves using co-station pairs according to claim 1, characterized in that, When the station pair is located on the same side of the epicenter, the cross-correlation of the seismic records of the two stations in the station pair is calculated to generate the cross-correlation Green's function, specifically including: The spatial location of the epicenter is The two stations in the station pair are denoted as platform i and platform j, and their spatial positions are respectively: ; when At that time, the cross-correlation Green's function for: ； when At that time, the cross-correlation Green's function for: ； These represent the seismic records in the frequency domain at platforms i and j, respectively. They represent The complex conjugate, It represents angular frequency.

3. The method for coherent superposition enhancement of low signal-to-noise ratio active source seismic surface waves using co-station pairs according to claim 1, characterized in that, When the station pair is located on opposite sides of the epicenter, the cross-correlation between the seismic records of the two stations in the pair and the seismic record of the reference station at the epicenter is calculated separately, and then convolution is performed to generate the convolution Green's function, specifically including: The spatial location of the epicenter is The two stations in the station pair are denoted as platform i and platform j, and their spatial positions are respectively: ; Convolution Green's function for: ； These represent the seismic records in the frequency domain at platforms i and j, respectively. This represents the complex conjugate of the frequency domain seismic records from the reference station at the epicenter. It represents angular frequency.

4. The method for coherent superposition enhancement of low signal-to-noise ratio active source seismic surface waves using co-station pairs according to claim 1, characterized in that, For each pair of stations, all seismic records are weighted and superimposed using cross-correlation Green's functions and convolution Green's functions, specifically including: ； ； The Green's function represents the superposition of cross-correlation. The Green's function represents the convolution after stacking. This represents the weight of the cross-correlation Green's function corresponding to the k-th excitation of the active source. This represents the weight of the convolutional Green's function corresponding to the k-th excitation of the active source. and These represent the cross-correlation Green's function and the convolution Green's function, respectively. It represents angular frequency.

5. The method for coherent superposition enhancement of low signal-to-noise ratio active source seismic surface waves using co-station pairs according to claim 4, characterized in that, The calculation method is as follows: ； The calculation method is as follows: ； These represent the seismic records in the frequency domain at platforms i and j, respectively. They represent The complex conjugate, These represent the frequency domain seismic records of the reference station at the epicenter and their complex conjugates, respectively. It represents angular frequency.

6. The method for coherent superposition enhancement of low signal-to-noise ratio active source seismic surface waves using co-station pairs according to claim 1, characterized in that, The calculation of the dispersion energy map based on the superimposed cross-correlation Green's function and the convolutional Green's function specifically includes: ； ； in, These represent the dispersion energy maps calculated using the cross-correlation Green's function and the convolution Green's function, respectively. Generally refers to the transformation operators corresponding to different calculation methods. Represents angular frequency. Indicates the scanning speed. The Green's function represents the superposition of cross-correlation. This represents the convolution Green's function after the convolution is performed.

Images