CN-121995443-A - First arrival identification method based on spatial attribute change

CN121995443ACN 121995443 ACN121995443 ACN 121995443ACN-121995443-A

Abstract

The invention provides a first arrival identification method based on spatial attribute change, which comprises the steps of 1, inputting a seismic trace set needing to be picked up, 2, eliminating inter-trace time difference based on a small smooth reference surface, 3, reconstructing the trace set by using the spatial correlation of the trace set by adopting a PCA method, extracting a wavelet template on the PCA reconstructed trace set, calculating the spatial attribute of each point of the trace set by using a correlation coefficient method, and 4, identifying first arrival waves according to the spatial attribute change. The first arrival identification method based on the spatial attribute change solves the problem of inaccurate first arrival identification under the complex earth surface condition, fully utilizes the statistical correlation on the first arrival signal space, combines with the meta-mode seismic wavelet of seismic exploration, and realizes the stable identification of the first arrival under the complex earth surface condition.

Inventors

LIAN XIMENG
Cong longshui
WANG XIUMIN
TANG XIANGGONG
WANG RONGWEI
LIU GE
GONG JIAN

Assignees

中国石油化工股份有限公司
中国石油化工股份有限公司胜利油田分公司

Dates

Publication Date: 20260508
Application Date: 20241101

Claims (11)

1. The first arrival identification method based on the spatial attribute change is characterized by comprising the following steps of: step 1, inputting a seismic gather needing to pick up a first arrival; Step 2, eliminating the inter-track time difference based on the small smooth reference surface; Step 3, reconstructing the gather by using the spatial correlation of the gather by adopting a PCA method, extracting a wavelet template on the PCA reconstructed gather, and calculating the spatial attribute of each point of the gather by using a correlation coefficient method; and 4, identifying the first arrival wave according to the spatial attribute change.
2. The first arrival identification method based on spatial attribute change according to claim 1, wherein in step 1, wavelet convolution reflection coefficient is used as linear first arrival data after inter-channel time difference elimination, the elevations of all channels are equal by default, the data randomly apply nonlinear inter-channel time difference, and when a certain surface velocity is given, the elevations are calculated according to the nonlinear inter-channel time difference, and noise is applied at the same time, so that a complex subsurface seismic trace set needing first arrival pickup with a known surface elevation is simulated, and the data is used as an input seismic trace set.
3. The first arrival identification method based on spatial attribute variation as set forth in claim 2, wherein in step 1, when a given surface velocity is 2000m/s, an elevation is calculated according to a nonlinear inter-track time difference while applying noise for making a data signal-to-noise ratio be-3 dB, and a signal-to-noise ratio calculation formula is S is a signal, n is noise, so that a seismic trace set which needs to be picked up at first arrival under the complex subsurface of a known ground surface elevation is simulated, and the data is used as an input seismic trace set.
4. The first arrival identification method based on spatial attribute change as set forth in claim 1, wherein in step 2, the surface elevation is counted and smoothed to generate a smoothed surface elevation, and then the data is corrected from the real surface to the smoothed surface elevation by using a formula, so that the first arrival becomes linear and the spatial correlation between the data is enhanced.
5. The first arrival identification method based on spatial attribute variation according to claim 4, wherein in step 2, the formula for correcting data from a real ground surface to a smooth elevation surface is: The method comprises the steps of elev smooth of a smoothed elevation surface, elev real of a real earth surface elevation, v smooth of a smoothing speed, delta t being a time difference correction amount, calculating according to the formula by using the real elevation of each seismic channel and the smoothed elevation, and enabling each seismic channel to have a time difference correction amount, and enabling the time difference correction amount to act on original data, namely enabling each seismic channel to translate by delta t length on a time axis, so that data acquired on the smoothed elevation surface can be obtained, and arrival time jump between the channels is eliminated.
6. The first arrival identification method based on spatial attribute change according to claim 1, wherein in step 3, spatial attribute extraction is performed, wavelet templates are selected, and then the wavelet templates are matched to obtain attributes representing the similarity between the wavelet templates.
7. The first arrival identification method based on spatial attribute variation according to claim 6, wherein in step 3, first, main component reconstruction data is selected by PCA method on the output result of step 2, wavelets extracted from the PCA reconstructed trace set are used as templates, then, for each seismic trace, the position of each first arrival wave is found on the trace set reconstructed by PCA, windows are opened up and down on a time axis and are used as wavelet templates, and correlation coefficients between the wavelet templates of each trace and the windows of the output data of the corresponding step 2 are calculated by adopting a correlation method.
8. The first arrival identification method based on spatial attribute variation according to claim 7, wherein in step 3, the correlation coefficient calculation formula is: Wherein w represents the length of a time window, a represents a wavelet template, ai represents the amplitude value of each sampling point of the wavelet template, b represents the amplitude value of each sampling point of signals on the corresponding seismic trace of the original data and the wavelet template, bi represents the amplitude value of each sampling point of the signals, a correlation coefficient calculation formula represents the attribute of the similarity of windowing data on the corresponding seismic trace and the wavelet template of the trace, the more similar is the first arrival, the more similar is the similarity of time sequences with two equal lengths represented by the formula, the more similar is the value of the formula, the more similar is the time sequence of windowing on the original data and the wavelet template on a reconstruction trace set, namely the greater is the probability of the position of the central point of the time sequence on the original data.
9. The first arrival identification method based on spatial attribute change according to claim 7, wherein in step 3, a time window around a first arrival wave of a PCA reconstruction gather is used as the wavelet template of the channel, the time window around the first arrival wave of the 1 st channel of the output data of step 2 is traversed to obtain the spatial attribute corresponding to the 1 st channel, the time window around the first arrival wave of the 2 nd channel of the PCA reconstruction gather is used as the wavelet template of the channel, the time window around the first arrival wave of the 2 nd channel of the output data of step 2 is traversed to obtain the spatial attribute corresponding to the 2 nd channel, and the spatial attribute extracted from the output data of step 2 is obtained by calculation channel by channel.
10. The first arrival identification method based on spatial attribute change according to claim 1, wherein in step4, for the output result of step 3, the maximum point of the spatial attribute, that is, the position of the first arrival wave, is found at each pass.
11. The first arrival identification system based on the spatial attribute change is characterized in that the first arrival identification system based on the spatial attribute change adopts the first arrival identification method based on the spatial attribute change as set forth in any one of claims 1 to 10 to identify first arrival waves under complex surface conditions.

Description

First arrival identification method based on spatial attribute change Technical Field The invention relates to the technical field of geophysical exploration, in particular to a first arrival identification method based on spatial attribute change. Background Hydrocarbon seismic exploration has entered the stage of complex earth's surface, complex formations, complex reservoirs and deep target exploration, and "two wide-one high" seismic data acquisition has also been known as industry consensus. The mountain front zone/the mountain front zone detection zone of each oil-gas basin in the west of China becomes a strategic take-over zone of oil-gas resources in China, and good exploration effects are obtained. However, this does not illustrate that we have completely solved the core technical problem of "dual complex" probe zone seismic exploration. The fundamental problem of the imaging process of the pre-mountain zone seismic data is how to build a velocity model meeting the requirement of offset imaging based on inter-theatrical moveout and low signal-to-noise ratio data. The identification and travel time detection of the first arrival wave plays a central role in medium speed estimation and modeling of the surface layer and the shallow layer. For in front of the mountains zones of complex near-surface conditions, conventional first-arrival pick-up methods would be difficult to meet, and thus, development of first-arrival pick-up methods under complex surface conditions would be required. First-arrival picking is an indispensable step in seismic data processing. The first arrival is the first signal generated by the seismic source recorded in the gather of seismic traces. Typically, the first arrival is related to the direct wave at the near offset and the refracted or return wave at the far offset. The first arrival picked up is often used in a tomographic method to estimate near-surface velocity structures and for static correction. The accuracy of the chromatographic method and the effect of static correction are both dependent on the accuracy of the first arrival pick-up. For the 'two wide one high' seismic data acquisition, massive seismic data can generate a large amount of first arrival pickup workload. Thus, as the number of channels in a seismic acquisition system increases, automatic or semi-automatic pick-up methods are essential in order to reduce the time for seismic data processing. Over the past several decades, a number of methods have been developed to improve the effectiveness and efficiency of first-arrival pick-up, which take into account different seismic physics and attributes. In recent years, many machine learning methods have also been used for first-arrival automatic picking, including artificial neural networks (Mai ty et al, 2014), fuzzy clustering (Chen, 2017), and support vector machines (dutan and Zhang, 2019). On the basis of CNNs (Hol lander et al., 2018), a wide variety of new networks have been applied to first-arrival pick-ups, such as SegNet (Badrinarayanan et al.,2017; wu et al., 2019), phaseNet (Zhu and Beroza, 2018) and U-Net (Hu et al., 2019); ma et al., 2020). However, machine learning methods rely heavily on labels, which are time consuming to produce during training and to compute on a network. In addition, the over-fitting problem and lack of diversity in the training data also reduces the stability of these machine learning methods, making them good on training data, but poor on new data. Although successful applications of these methods have demonstrated their effectiveness, they are almost always sensitive to noise and produce incorrect first-arrival pick-up results if the gather signal-to-noise ratio is low. In application number: the invention provides a first arrival picking up process for mountain area data, which comprises the steps of preprocessing, first arrival wave filtering by an anisotropic Gaussian filter, high-dimensional and multi-attribute extraction, determining the approximate range of first arrival picking up by a clustering algorithm, and finally picking up the first arrival by multi-attribute Markov decision to obtain the final picking up result. Compared with the traditional method, the method realizes automatic and intelligent pickup, and aims at seismic data with low signal-to-noise ratio and severe inter-channel time difference change in mountain areas in China. But the method extracts the first arrival structure based on three properties of energy ratio, kurtosis and edge strength. The energy ratio attribute and the kurtosis attribute are all single-channel attribute, the spatial correlation of the first arrival signal is not utilized, the edge intensity attribute only uses the information of 2 adjacent channels, the method is a method in the image field, the seismic image needs to be grayed, and the graying mode can influence the extraction effect of the edge attribute. Therefore, all three attributes do not fully utilize the statist