A method and device for picking up earthquake phases and weak signals based on multi-band cross-correlation

Abstract

The present invention relates to a method and device for picking up earthquake phases and weak signals based on multi-band cross-correlation. The method comprises: obtaining theoretical first arrival times of reference phases and target phases, using the theoretical first arrival times to intercept target waveform data of the reference phase and each target phase or weak signal before and after the weak signal in waveform data; calculating the cross-correlation function of the reference phase and the target phase based on the target waveform data, obtaining the multi-band cross-correlation probability density through the cross-correlation function, and determining the final cross-correlation first arrival time; and using the theoretical first arrival time and the final cross-correlation first arrival time to pick up the observed first arrival time of the target phase. The present invention can efficiently achieve high-precision detection of earthquake phases and weak signals, providing reliable technical support for earthquake monitoring and early warning and research on the internal structure of the Earth.

Description

Technical Field

[0001] The present invention relates to the technical field of earthquake monitoring and signal processing, and in particular to a method and device for picking up earthquake phases and weak signals based on multi-band cross-correlation. Background Art

[0002] Earthquake phase information is essential data for geophysical research. Its accurate acquisition is not only crucial for earthquake monitoring and early warning, but is also widely used in earthquake location, focal mechanism analysis, and travel-time tomography of the Earth's interior. Earthquake phase acquisition refers to the process of accurately detecting and extracting the arrival times of seismic phases, such as P and S waves, from data recorded by seismic stations.

[0003] Currently, seismic phase picking methods are mainly divided into manual picking, traditional methods, and artificial intelligence methods. Although manual picking has high accuracy, it is limited by a shortage of professionals, a large workload, low efficiency, and a high reliance on experience. Traditional methods such as the short-time window method (STA / LTA) and the autoregressive Akaike Information Criterion (AIC), although computationally simple, are sensitive to noise, highly parameter-dependent, and difficult to adapt to complex signal environments. Artificial intelligence methods excel in automatically processing massive amounts of data and adapting to complex signal and noise environments, significantly improving seismic phase picking accuracy. However, they still suffer from problems such as reliance on large amounts of labeled data, high consumption of computing resources, poor model interpretability, and limited generalization capabilities.

[0004] Existing traditional methods rely on single-band signal processing, resulting in high computational complexity, poor real-time performance, and insufficient weak signal detection capabilities. Manual picking offers high precision but low efficiency, making it difficult to process massive amounts of data. While combining the two approaches can improve complex seismic data processing capabilities, these approaches face technical challenges such as collaborative optimization of manual and automated algorithms, designing real-time interaction mechanisms, and balancing real-time performance and detection accuracy in complex noisy environments. Summary of the Invention

[0005] In order to solve the problems existing in the above-mentioned prior art, the purpose of the present invention is to provide a method and equipment for picking up earthquake phases and weak signals based on multi-band cross-correlation, improve the ability of picking up earthquake phases in time and detecting weak signals, combine the collaborative mechanism of manual and automatic processing, take into account both efficiency and accuracy, and solve the problem of accurate picking up of earthquake phases in complex noise environments.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] A method for picking up earthquake phases and weak signals based on multi-band cross-correlation, comprising:

[0008] Obtaining theoretical first arrival times of the reference phase and the target phase, and using the theoretical first arrival times to intercept the reference phase in the waveform data and target waveform data of each target phase or the time period before and after the weak signal;

[0009] Based on the target waveform data, calculating the cross-correlation function between the reference seismic phase and the target seismic phase, obtaining the multi-band cross-correlation probability density through the cross-correlation function, and determining the final cross-correlation first arrival time;

[0010] The observed first arrival time of the target seismic phase is picked up using the theoretical first arrival time and the final cross-correlation first arrival time.

[0011] Optionally, obtaining the theoretical first arrival times of the reference seismic phase and the target seismic phase includes:

[0012] A one-dimensional earth velocity reference model is set, and the theoretical first arrival times of the reference seismic phase and the target seismic phase are calculated using the one-dimensional earth velocity reference model.

[0013] Optionally, acquiring the waveform data includes:

[0014] Acquiring original seismic data, wherein the original seismic data includes: the longitude and latitude of the earthquake event, the focal depth, and the longitude and latitude of the station;

[0015] The raw seismic data is preprocessed, wherein the preprocessing includes removing instrument response, removing mean, removing trend and upsampling to obtain the waveform data.

[0016] Optionally, intercepting the target waveform data includes:

[0017] According to the theoretical first arrival time, a signal analysis time window including the target seismic phase is set, and the signal analysis time window is used to intercept the target waveform data of the reference seismic phase and each target seismic phase or the time period before and after the weak signal.

[0018] Optionally, intercepting the target waveform data includes:

[0019] Calculate the mean of the phases within the signal analysis time window:

[0020]

[0021] Among them, x i is the waveform data point within the signal time window, and N is the total number of data points;

[0022] According to the theoretical first arrival time, a noise time window is set to obtain a background noise time period before the theoretical first arrival time;

[0023] Calculate the signal-to-noise ratio of the reference phase and the target phase or the weak signal based on the noise time window and the signal analysis time window:

[0024]

[0025] Among them, E signal and E noise are the energies of the signal time window and the noise time window respectively.

[0026] Optionally, the energy of the signal time window and the noise time window includes:

[0027]

[0028] Among them, x i and y i are the waveform data points in the signal time window and the noise time window respectively, N s and N n are the total number of data points in the signal time window and the noise time window respectively.

[0029] Optionally, calculating the cross-correlation function between the reference seismic phase and the target seismic phase includes:

[0030] Using multiple signal frequency bands covering the main frequency components of the target seismic phase as a multi-band filter bank to decompose the target waveform data into multiple frequency band signals;

[0031] Using multiple frequency band signals, the cross-correlation function between the reference phase and the target phase is calculated:

[0032]

[0033] Where τ is the time delay parameter, which represents the offset of the target phase relative to the reference phase, l is the length of the cross-correlation time window, is the cross-correlation function of the ith frequency band.

[0034] Optionally, determining the final cross-correlation first arrival time includes:

[0035] Determine the cross-correlation first arrival time:

[0036] T cc =τ·δ

[0037] Where, δ is the signal sampling interval;

[0038] By weighted fusion of the cross-correlation first arrival times of different frequency bands, the multi-band cross-correlation probability density is constructed:

[0039]

[0040] Where M is the total number of frequency bands, w i is the weight of the ith frequency band, weight w i According to T cc Dynamic adjustment of frequency characteristics;

[0041] The time corresponding to the probability density peak in the multi-band cross-correlation probability density is extracted as the final cross-correlation first arrival time, and the results are screened for reliability based on comprehensive signal quality assessment parameters. Finally, the accuracy of the first arrival time is optimized through manual verification.

[0042] To achieve the above-mentioned object, the present invention further provides a seismic phase and weak signal picking device based on multi-band cross-correlation, comprising:

[0043] A theoretical calculation module is used to obtain the theoretical first arrival time of the reference phase and the target phase, and to intercept the reference phase in the waveform data and the target waveform data of each target phase or the time period before and after the weak signal using the theoretical first arrival time;

[0044] a cross-correlation analysis module, configured to calculate a cross-correlation function between the reference seismic phase and the target seismic phase based on the target waveform data, obtain a multi-band cross-correlation probability density through the cross-correlation function, and determine a final cross-correlation first arrival time;

[0045] The manual picking module is used to pick the observed first arrival time of the target seismic phase by using the theoretical first arrival time and the final cross-correlation first arrival time.

[0046] The beneficial effects of the present invention are:

[0047] 1. High-precision seismic phase picking: This invention uses multi-band cross-correlation technology to fully utilize the multi-band characteristics of seismic signals, significantly improving the accuracy of seismic phase picking, especially in weak signal and complex noise environments.

[0048] 2. Efficiency and User-Friendliness: Through optimized algorithm design and modular processing, this system improves data processing efficiency while maintaining high accuracy, meeting the real-time requirements of earthquake monitoring and early warning. Furthermore, it provides an intuitive, user-friendly interface that supports real-time monitoring, parameter adjustment, and manual intervention, significantly enhancing user experience and operational convenience.

[0049] 3. Provide reliable technical support for related fields: The present invention provides high-precision, high-efficiency and reliable technical support for fields such as earthquake monitoring, early warning, source mechanism analysis and research on the internal structure of the earth. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0051] Figure 1 This is a flow chart of a method for picking up earthquake phases and weak signals based on multi-band cross-correlation according to an embodiment of the present invention;

[0052] Figure 2 Schematic diagram of three earthquake phase ray paths according to an embodiment of the present invention;

[0053] Figure 3 The theoretical first arrival time T under the high signal-to-noise ratio condition of the embodiment of the present invention is pre Schematic diagram of the aligned first arrival picking of P, PcP and PKiKP earthquake phases;

[0054] Figure 4 The first arrival time T is actually picked up according to the embodiment of the present invention. obs Schematic diagram of picking the first arrival of earthquake phases after alignment;

[0055] Figure 5 Schematic diagram of picking up the first arrival of earthquake phases in a low signal-to-noise ratio and high noise environment according to an embodiment of the present invention;

[0056] Figure 6 The actual first arrival time T is obtained under high noise environment according to the embodiment of the present invention. obs Schematic diagram of the aligned earthquake phase picking effect. DETAILED DESCRIPTION

[0057] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0058] This embodiment discloses a method for picking up earthquake phases and weak signals based on multi-band cross-correlation, including: obtaining the theoretical first arrival times of the reference phase and the target phase, and using the theoretical first arrival time to intercept the reference phase in the waveform data and the target waveform data of each target phase or the time period before and after the weak signal; based on the target waveform data, calculating the cross-correlation function of the reference phase and the target phase, obtaining the multi-band cross-correlation probability density through the cross-correlation function, and determining the final cross-correlation first arrival time; using the theoretical first arrival time and the final cross-correlation first arrival time, picking up the observed first arrival time of the target phase.

[0059] Specifically, this embodiment discloses a method for picking up earthquake phases and weak signals based on multi-band cross-correlation, including:

[0060] Step S1: Acquire seismic data and pre-process it to obtain waveform data;

[0061] Step S2: Initialize model parameters, including: reference model, reference phase, target phase, noise time window, signal time window, cross-correlation time window and multi-band filter bank;

[0062] Step S3: Calculate the theoretical first arrival time T of each earthquake phase based on the reference model and time window selection pre , signal-to-noise ratio and mean;

[0063] Step S4: Calculate the multi-band cross-correlation probability density of the target phase based on the reference phase and the multi-band filter bank, and construct the first arrival time prediction model T cc ;

[0064] Step S5: Reference theoretical first arrival time T pre and cross-correlation first arrival time T cc , manually accurately pick the observed first arrival time T of the target seismic phase obs and store all newly arrived data;

[0065] Step S6: After completing the seismic phase picking of the current seismic data, iteratively process the next seismic data, repeat steps S3 to S5, and perform seismic phase picking of the next seismic data until all seismic data processing is completed;

[0066] Step S7: storing a target phase first arrival time data table of multi-channel seismic data.

[0067] Furthermore, obtaining waveform data includes: obtaining original seismic data, the original seismic data including: the longitude and latitude of the earthquake event, the focal depth and the longitude and latitude of the station; preprocessing the original seismic data, the preprocessing including: removing instrument response, removing mean, removing trend and upsampling to obtain waveform data.

[0068] Specifically, step S1 includes the following steps: obtaining original seismic data, which includes basic information such as the longitude and latitude of the earthquake event, focal depth, and longitude and latitude of the station; and preprocessing the original seismic data, including removing instrument response, removing mean, removing trend, and increasing sampling.

[0069] Furthermore, obtaining the theoretical first arrival times of the reference phase and the target phase includes: setting a one-dimensional earth velocity reference model, and using the one-dimensional earth velocity reference model to calculate the theoretical first arrival times of the reference phase and the target phase.

[0070] Intercepting target waveform data includes: setting a signal analysis time window containing the target seismic phase according to the theoretical first arrival time, and using the signal analysis time window to intercept the reference seismic phase and the target waveform data of the time period before and after each target seismic phase or weak signal.

[0071] After intercepting the target waveform data, it includes: calculating the mean value of the seismic phase within the signal analysis time window, setting the noise time window according to the theoretical first arrival time, and obtaining the background noise time period before the theoretical first arrival time; based on the noise time window and the signal analysis time window, calculating the signal-to-noise ratio of the reference seismic phase and the target seismic phase or weak signal.

[0072] Specifically, further, step S2 specifically includes the following steps:

[0073] Reference model: Set a one-dimensional earth velocity reference model to calculate the theoretical first arrival time T of the reference phase and the target phase pre ;

[0074] Reference phase: Select P wave or S wave as the reference phase, which is determined according to the type of earthquake event;

[0075] Target phase: Set the earthquake phase or weak signal type to be detected (such as PCP wave, ScS wave or other specific phase);

[0076] Noise time window: select the background noise time period before the theoretical first arrival time of the earthquake phase for calculating the signal-to-noise ratio;

[0077] Signal time window: According to the theoretical first arrival time of the earthquake phase, set the signal analysis time window including the target phase;

[0078] Cross-correlation time window: Set the time window for multi-band cross-correlation calculation, usually a part of the signal time window;

[0079] Multi-band filter bank: It is preferred to use multiple signal frequency bands covering the main frequency components of the target phase to capture the signal characteristics of different frequency bands and avoid inaccurate first arrival picking caused by single filter spectrum truncation.

[0080] Step S3 specifically includes the following steps:

[0081] According to the one-dimensional earth velocity reference model, the theoretical first arrival time T of the reference phase and each target phase or weak signal is calculated. pre ;

[0082] Based on the first arrival time of the seismic phase and different time window parameters, the waveform data of the reference seismic phase and each target seismic phase or weak signal before and after the time period are intercepted respectively;

[0083] Based on the signal time window, calculate the mean of the seismic phase within the signal time window:

[0084]

[0085] Among them, x i is the waveform data point within the signal time window, and N is the total number of data points;

[0086] Based on the noise time window and signal time window, the signal-to-noise ratio (SNR) of the reference phase and the target phase is calculated to evaluate the signal quality and provide a reference for subsequent multi-band cross-correlation calculations:

[0087]

[0088] Among them, E signal and E noise are the energy of the signal time window and the noise time window respectively, and the calculation formula is:

[0089]

[0090] Among them, x i and y i are the waveform data points in the signal time window and the noise time window respectively, N s and N n are the total number of data points in the signal time window and the noise time window respectively.

[0091] Furthermore, calculating the cross-correlation function between the reference seismic phase and the target seismic phase includes: using multiple signal frequency bands covering the main frequency components of the target seismic phase as a multi-band filter group to decompose the target waveform data into multiple frequency band signals; using multiple frequency band signals to calculate the cross-correlation function between the reference seismic phase and the target seismic phase.

[0092] Specifically, step S4 includes the following steps:

[0093] Multi-band signal decomposition: A multi-band filter bank is used to decompose the waveform data of the reference phase and the target phase into multiple band signals: each band signal is represented by x i [n] and y i [n], where i represents the frequency band index and n represents the time point;

[0094] Calculate the multi-band cross-correlation function between the reference phase and the target phase: Calculate the reference phase x according to each frequency band signal i [n] In phase with target i The cross-correlation function of [n] is:

[0095]

[0096] Where τ is the time delay parameter, which represents the offset of the target phase relative to the reference phase, l is the length of the cross-correlation time window, is the cross-correlation function of the ith frequency band.

[0097] Furthermore, determining the final cross-correlation first arrival time includes: determining the cross-correlation first arrival time, constructing a multi-band cross-correlation probability density by weighted fusion of the cross-correlation first arrival times of different frequency bands, extracting the time corresponding to the probability density peak in the multi-band cross-correlation probability density as the final cross-correlation first arrival time, and screening the results for reliability based on comprehensive signal quality assessment parameters, and finally optimizing the accuracy of the first arrival time through manual verification.

[0098] Specifically, determine the cross-correlation first arrival time T cc :

[0099] T cc =τ·δ (6)

[0100] Where, δ is the signal sampling interval;

[0101] Construct the cross-correlation first arrival time Tcc probability density model: By weighted fusion of the cross-correlation results of each frequency band, the multi-band cross-correlation probability density P[k] is obtained:

[0102]

[0103] Where M is the total number of frequency bands, w i is the weight of the ith frequency band, weight w i Dynamically adjusted according to Tcc frequency characteristics.

[0104] This embodiment also discloses an earthquake phase and weak signal picking device based on multi-band cross-correlation, including: a theoretical calculation module, used to obtain the theoretical first arrival time of the reference phase and the target phase, and use the theoretical first arrival time to intercept the reference phase in the waveform data and the target waveform data of each target phase or weak signal before and after the time period; a cross-correlation analysis module, used to calculate the cross-correlation function of the reference phase and the target phase based on the target waveform data, obtain the multi-band cross-correlation probability density through the cross-correlation function, and determine the final cross-correlation first arrival time; a manual picking module, used to use the theoretical first arrival time and the final cross-correlation first arrival time to pick up the observed first arrival time of the target phase.

[0105] This embodiment also discloses a seismic phase and weak signal picking device based on multi-band cross-correlation, including: a data acquisition module for executing step S1; a parameter initialization module for executing step S2; a theoretical calculation module for executing step S3; a cross-correlation analysis module for executing step S4; a manual picking module for executing step S5; an iterative processing module for executing step S6; and a data storage module for executing step S7.

[0106] In a second aspect, this embodiment further provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor.

[0107] In a third aspect, this embodiment further provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-mentioned method for picking up earthquake phases and weak signals based on multi-band cross-correlation.

[0108] Fourthly, this embodiment also provides a computer program product, including a computer program, that, when executed, implements a method for accurately picking seismic phases and weak signals. The program features a modular design, supports data in multiple formats, provides an interactive graphical user interface, optimizes algorithms to improve efficiency, and integrates manual intervention to ensure high-precision picking. This system is suitable for earthquake monitoring, early warning, and research, reducing labor costs and meeting real-time processing requirements.

[0109] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0110] like Figure 1 As shown, this embodiment provides a method for picking up earthquake phases and weak signals based on multi-band cross-correlation, which specifically includes the following steps:

[0111] Step S1: Acquire seismic data and pre-process it to obtain waveform data;

[0112] Step S2: Initialize model parameters, including: reference model, reference phase, target phase, noise time window, signal time window, cross-correlation time window and multi-band filter bank;

[0113] Step S3: Calculate the theoretical first arrival time T of each earthquake phase based on the reference model and time window selection pre , signal-to-noise ratio and mean;

[0114] Step S4: Calculate the multi-band cross-correlation probability density of the target phase based on the reference phase and the multi-band filter bank, and construct the first arrival time prediction model T cc ;

[0115] Step S5: Reference theoretical first arrival time T pre and cross-correlation first arrival time T cc , manually accurately pick the observed first arrival time T of the target seismic phase obs and store all newly arrived data;

[0116] Step S6: After completing the seismic phase picking of the current seismic data, iteratively process the next seismic data, repeat steps S3 to S5, and perform seismic phase picking of the next seismic data until all seismic data processing is completed;

[0117] Step S7: storing a target phase first arrival time data table of multi-channel seismic data.

[0118] Furthermore, step S1 specifically includes the following steps:

[0119] Obtaining original earthquake data, which includes basic information such as the longitude and latitude of the earthquake event, focal depth, and longitude and latitude of the station;

[0120] The raw seismic data are preprocessed, including removing instrument response, removing mean, removing trend and upsampling.

[0121] Furthermore, step S2 specifically includes the following steps:

[0122] Reference model: Set a one-dimensional earth velocity reference model to calculate the theoretical first arrival time T of the reference phase and the target phase pre In this embodiment, the ak135 one-dimensional velocity model is selected as the reference model. Furthermore, other standard models such as the PREM model and the iasp91 model can also be selected according to the specific application scenario. These reference models all provide standardized Earth's internal velocity structure parameters to ensure the consistency and comparability of the theoretical first arrival time calculation. In practical applications, the most appropriate reference model can be flexibly selected according to the target phase type and the characteristics of the study area.

[0123] Reference phase: Select P wave or S wave as the reference phase. In this embodiment, the reference phase is P wave. The P wave related target phase preferably uses P wave as the reference phase (e.g. P diff , PKIKP shock equal);

[0124] Target Phase: Set the earthquake phase or weak signal type to be detected. Set the earthquake phase or weak signal type to be detected. In this embodiment, PcP and PKiKP are specifically selected as target phases (up to 3 target phases can be processed simultaneously). Figure 2 The results show that the reference phase P wave and the target phases PcP and PKiKP have different ray path characteristics. The PKiKP phase has a long propagation path, large energy attenuation, and is affected by complex deep structures. Its signal-to-noise ratio is usually significantly lower than that of other phases, making it a typical weak signal that is difficult to detect.

[0125] Noise time window: select the background noise time period before the theoretical first arrival time of the earthquake phase for calculating the signal-to-noise ratio. In this embodiment, the theoretical first arrival time T pre The [-30s, -10s] waveform data is used as the noise time window;

[0126] Signal time window: According to the theoretical first arrival time of the earthquake phase, the signal analysis time window including the target phase is set. In this embodiment, the theoretical first arrival time T pre The [-8s, 8s] waveform data is used as the signal time window;

[0127] Cross-correlation time window: Set the time window for multi-band cross-correlation calculation, which is usually a part of the signal time window. In this embodiment, the theoretical first arrival time T pre The [-8s, 8s] waveform data is used as the cross-correlation time window;

[0128] Multi-band filter banks: It is preferred to use multiple signal frequency bands covering the main frequency component of the target seismic phase to capture signal characteristics in different frequency bands and avoid inaccurate first-break picking caused by truncation of the spectrum by a single filter. In this embodiment, a multi-band filter bank in the range of 0.5-5 Hz is used, focusing on the main frequency band of 1-3 Hz. The filter parameters are set in intervals of 0.1 Hz to extract phase characteristics in different frequency bands and improve the accuracy of first-break picking.

[0129] Furthermore, step S3 specifically includes the following steps:

[0130] According to the one-dimensional earth velocity reference model, the theoretical first arrival time T of the reference phase and each target phase or weak signal is calculated. pre In this embodiment, the ray tracing method is used to calculate the theoretical first arrival time of the earthquake phase.

[0131] Based on the first arrival time of the seismic phase and different time window parameters, the waveform data of the time window segments before and after the reference seismic phase and each target seismic phase or weak signal are intercepted respectively. In this embodiment, the waveform data interception operation for each seismic phase is as follows: taking the theoretically calculated first arrival time Tpre as the reference point, the complete waveform data within the corresponding time window range is intercepted before and after it (for example, the signal time window [-8s, 8s]). It should be noted that there are two main reasons: 1) there are systematic differences between different one-dimensional earth velocity reference models themselves; 2) seismic waves are affected by the heterogeneity and attenuation effects of the earth's internal medium during actual propagation, which may result in significant deviations between the theoretically calculated first arrival time Tpre and the actually observed first arrival time Tobs. Figure 3 The present invention shows that the P, PcP and PKiKP phases are based on T under high signal-to-noise ratio conditions. pre The aligned earthquake phase waveforms clearly show the initial arrival characteristics, amplitude attenuation and waveform changes of each phase. Figure 3 and Figure 5 The comparison shows the P, PcP and PKiKP phases based on T under different signal-to-noise ratio conditions in the embodiment of the present invention. pre Aligned waveform features. In a high signal-to-noise ratio environment ( Figure 3 ), each phase shows clear waveform characteristics: the P wave has a sharp initial arrival, the PcP wave has a moderate amplitude, and the PKiKP wave is weak but still recognizable. Figure 5), background noise is significantly enhanced, and while the P-wave first arrival signature remains visible, the first arrival times of the PcP and PKiKP phases are difficult to detect. This comparison intuitively illustrates the energy attenuation of seismic waves along different propagation paths, particularly the vulnerability of deep phases like the PKiKP to noise, providing an important feature analysis foundation for subsequent signal processing.

[0132] Based on the signal time window, the mean of the seismic phase in the signal time window is calculated; based on the noise time window and the signal time window, the signal-to-noise ratio (SNR) of the reference seismic phase and the target seismic phase is calculated. In this embodiment, the energy characteristics of the seismic phase are characterized by calculating the waveform mean in the signal time window, and the signal-to-noise ratio (SNR) is calculated based on the energy ratio of the noise time window to the signal time window. Among them, the mean reflects the overall energy level of the seismic phase, and the SNR is used to evaluate the signal quality. These two parameters provide an important reference basis for the subsequent multi-band cross-correlation analysis: the mean can be used to compare the energy of different seismic phases, and the SNR is used as a criterion for judging signal quality, guiding the parameter selection and result evaluation of the subsequent analysis process, thereby improving the overall reliability of seismic phase picking. Figure 3 The mean value of the medium reference phase P wave is 0.1, and the signal-to-noise ratio (SNR) is 34.05. The P wave under high SNR has a sharp first arrival phase.

[0133] Furthermore, step S4 specifically includes the following steps:

[0134] Multi-band signal decomposition: A multi-band filter group is used to decompose the waveform data of the reference phase and the target phase into multiple frequency band signals. In this embodiment, a pre-set multi-band filter group (0.5-5Hz, interval 0.1Hz, main frequency band of 1-3Hz) is used to perform frequency band decomposition on the waveform data of the reference phase P wave and the target phase (PcP, PKiKP, etc.). In the specific implementation, each filter adopts a zero-phase 4-pole Butterworth filter design to ensure that each frequency band signal does not produce phase distortion. This step can effectively separate the phase characteristics of different frequency bands, especially for weak phases such as PKiKP, whose high-frequency components often contain important features, avoiding the influence of filter cutoff.

[0135] Calculate the multi-band cross-correlation function between the reference phase and the target phase, and determine the cross-correlation first arrival time T cc , construct the cross-correlation first arrival time T cc Probability density model. In the embodiment of the present invention, the multi-band cross-correlation technology is used to accurately provide the reference time of the first arrival of the seismic phase. The specific implementation process is as follows: First, a multi-band (0.5-5Hz) cross-correlation analysis is performed on the reference seismic phase and the target seismic phase, and the cross-correlation function in each frequency band is calculated (for example Figure 3 The cross-correlation waveform on the right side of the figure is obtained, and the probability density distribution of the multi-band cross-correlation results is obtained, as shown in Figure 3As shown in the figure, the shaded area above the target phase waveform intuitively shows the probability density distribution of the multi-band cross-correlation results, where the color depth represents the credibility of the first arrival time prediction in different frequency bands. The dotted line in the figure marks the most accurate T in the main frequency band (1-3Hz). cc When the prediction is reached. T constructed by kernel density estimation method cc The probability density model integrates the results of each frequency band and selects the time corresponding to the probability density peak as the final cross-correlation first arrival time. This multi-band fusion method significantly improves the detection reliability of weak seismic phase signals (such as PKiKP). The visualization of the probability density distribution not only facilitates manual verification and result evaluation, but also ensures that the main frequency band contributes more weight to the final result.

[0136] Furthermore, step S5 specifically includes the following steps:

[0137] In the embodiment of the present invention, the theoretical first arrival time T is calculated by pre and multi-band cross-correlation time T cc , the final observation first arrival time T is determined manually obs In order to evaluate the signal quality of T cc The SNR and mean parameters are used as key indicators: when the SNR is lower than 1, the signal quality is significantly reduced, resulting in a decrease in the reliability of the first arrival time picking; if the SNR further drops below -10, it indicates that there is a lack of identifiable first arrival information in the signal. At this time, the signal is close to the noise level, and T cc The error increases significantly, and the signal data should be reasonably selected. In addition, the mean parameter of the signal should usually be close to 0. If there is a large deviation, although it reflects the abnormal overall energy of the signal, it has no significant effect on T cc The accuracy is less affected.

[0138] like Figure 3 As shown in the middle column, under high signal-to-noise ratio conditions, press T pre The middle column of the aligned P, PcP, and PKiKP phase waveforms shows the comparison results between the normalized reference phase and the target phase. It can be clearly observed that the waveform consistency between the phases whose first arrival times were not accurately picked is poor. Figure 4 Shows the actual picking first arrival time T obs The waveform effect after alignment is the same as Figure 3 In comparison, after accurate alignment, the matching degree between each seismic phase waveform and the reference waveform is significantly improved, and the initial arrival characteristics of the seismic phase are more consistent. Figure 4 The dashed line on the right indicates the first arrival time T predicted by the multi-band cross-correlation. cc , which is located close to the zero point of the time axis, indicating the actual first arrival time of the observation picked up T obs and the predicted value T ccThis precise correspondence strongly verifies the reliability of the multi-band cross-correlation method in predicting the first arrival time of earthquake phases, especially for weak earthquake phases such as PcP and PKiKP. cc T of prediction results and manual professional judgment obs The difference is generally less than 0.1s, which fully demonstrates the accuracy of the algorithm in practical applications.

[0139] Figure 5 The uncorrected seismic waveform characteristics under low signal-to-noise ratio conditions are shown. The P phase, due to its high energy, maintains a high signal-to-noise ratio and clear first-arrival characteristics, while the PcP and PKiKP phases are significantly disturbed by the coda waves of subsequent phases, resulting in reduced recognition of their first-arrival characteristics. Figure 6 Shows the T obs After the alignment, the first arrival features of the PcP and PKiKP phases were effectively extracted through signal enhancement using multi-band cross-correlation techniques. It is particularly noteworthy that, although the PKiKP phase is still contaminated by coda waves, its key first arrival features can be reliably identified. Figure 6 T marked with a dotted line cc Predicted time and manually determined T obs Good consistency was maintained (average deviation < 0.1s), which confirmed that the method can maintain stable prediction performance under complex waveform interference conditions.

[0140] The program product developed based on this method provides complete interactive functions, including: free zooming of the waveform display area, double-clicking to accurately adjust the picking time, and displaying the relative time of the first arrival of the seismic phase (ΔT = T obs -T pre ), reset the operation, press T obs The program interface intuitively displays the probability density distribution of multi-band cross-correlation (such as Figure 3 The system supports human interaction through visual interaction to complete the whole process from initial prediction to final confirmation. Experiments show that this human-computer collaborative working mode can improve the efficiency of seismic phase picking by more than 40% while maintaining expert-level picking accuracy. It is particularly suitable for the accurate identification of weak seismic phases such as PCP and PKiKP. All picking results are automatically stored in a standardized data format, including T pre 、T cc 、T obs Complete parameters such as the above are provided to facilitate subsequent geophysical research.

[0141] The embodiments described above are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by persons skilled in the art should fall within the scope of protection defined by the claims of the present invention.

Claims

1. A method for picking up earthquake phases and weak signals based on multi-band cross-correlation, characterized in that: include: Obtaining theoretical first arrival times of the reference phase and the target phase, and using the theoretical first arrival times to intercept the reference phase in the waveform data and target waveform data of each target phase or the time period before and after the weak signal; Based on the target waveform data, calculating the cross-correlation function between the reference seismic phase and the target seismic phase, obtaining the multi-band cross-correlation probability density through the cross-correlation function, and determining the final cross-correlation first arrival time; The observed first arrival time of the target seismic phase is picked up using the theoretical first arrival time and the final cross-correlation first arrival time.

2. The method for picking up earthquake phases and weak signals based on multi-band cross-correlation according to claim 1, characterized in that: Obtaining the theoretical first arrival times of the reference seismic phase and the target seismic phase includes: A one-dimensional earth velocity reference model is set, and the theoretical first arrival times of the reference seismic phase and the target seismic phase are calculated using the one-dimensional earth velocity reference model.

3. The method for picking up earthquake phases and weak signals based on multi-band cross-correlation according to claim 1, characterized in that: Acquiring the waveform data includes: Acquiring original seismic data, wherein the original seismic data includes: the longitude and latitude of the earthquake event, the focal depth, and the longitude and latitude of the station; The raw seismic data is preprocessed, wherein the preprocessing includes removing instrument response, removing mean, removing trend and upsampling to obtain the waveform data.

4. The method for picking up earthquake phases and weak signals based on multi-band cross-correlation according to claim 1, characterized in that: Intercepting the target waveform data includes: According to the theoretical first arrival time, a signal analysis time window including the target seismic phase is set, and the signal analysis time window is used to intercept the target waveform data of the reference seismic phase and each target seismic phase or the time period before and after the weak signal.

5. The method for picking up earthquake phases and weak signals based on multi-band cross-correlation according to claim 4, characterized in that: After intercepting the target waveform data, the method includes: Calculate the mean of the phases within the signal analysis time window: Among them, x i is the waveform data point within the signal time window, and N is the total number of data points; According to the theoretical first arrival time, a noise time window is set to obtain a background noise time period before the theoretical first arrival time; Calculate the signal-to-noise ratio of the reference phase and the target phase or the weak signal based on the noise time window and the signal analysis time window: Among them, E signal and E noise are the energies of the signal time window and the noise time window respectively.

6. The method for picking up earthquake phases and weak signals based on multi-band cross-correlation according to claim 5, characterized in that: The energy of the signal time window and the noise time window includes: Among them, x i and y i are the waveform data points in the signal time window and the noise time window respectively, N s and N n are the total number of data points in the signal time window and the noise time window respectively.

7. The method for picking up earthquake phases and weak signals based on multi-band cross-correlation according to claim 1, characterized in that: Calculating the cross-correlation function between the reference seismic phase and the target seismic phase includes: Using multiple signal frequency bands covering the main frequency components of the target seismic phase as a multi-band filter bank to decompose the target waveform data into multiple frequency band signals; Using multiple frequency band signals, the cross-correlation function between the reference phase and the target phase is calculated: Where τ is the time delay parameter, which represents the offset of the target phase relative to the reference phase, l is the length of the cross-correlation window, is the cross-correlation function of the ith frequency band.

8. The method for picking up earthquake phases and weak signals based on multi-band cross-correlation according to claim 1, characterized in that: Determining the final cross-correlation first arrival time includes: Determine the cross-correlation first arrival time: T cc =t·d Where, δ is the signal sampling interval; By weighted fusion of the cross-correlation first arrival times of different frequency bands, the multi-band cross-correlation probability density is constructed: Where M is the total number of frequency bands, w i is the weight of the ith frequency band, weight w i According to T cc Dynamic adjustment of frequency characteristics; The time corresponding to the probability density peak in the multi-band cross-correlation probability density is extracted as the final cross-correlation first arrival time, and the results are screened for reliability based on comprehensive signal quality assessment parameters. Finally, the accuracy of the first arrival time is optimized through manual verification.

9. A pickup device for the seismic phase and weak signal pickup method based on multi-band cross-correlation according to any one of claims 1 to 8, characterized in that: include: A theoretical calculation module is used to obtain the theoretical first arrival time of the reference phase and the target phase, and to intercept the reference phase in the waveform data and the target waveform data of each target phase or the time period before and after the weak signal using the theoretical first arrival time; a cross-correlation analysis module, configured to calculate a cross-correlation function between the reference seismic phase and the target seismic phase based on the target waveform data, obtain a multi-band cross-correlation probability density through the cross-correlation function, and determine a final cross-correlation first arrival time; The manual picking module is used to pick the observed first arrival time of the target seismic phase by using the theoretical first arrival time and the final cross-correlation first arrival time.

Images