基于Jitter采样的压缩感知地震勘探数据重构

(1.云南省地震局,云南 昆明 650224; 2.玉龙县地震局,云南 玉龙 674100)

压缩感知; Jitter采样; 地震勘探; 数据重建

Compressed Sensing Seismic Exploration Data Reconstruction Based on Jitter Sampling
ZHANG Shuai1,YANG Runhai1,JIANG Jinzhong1,ZHANG Yan1,ZHENG Dingchang1,DENG Yuefei1,YANG Runping2,WANG Zhihao1

(1.Yunnan Earthquake Agency,Kunming 650224,Yunnan,China;2.Yulong Earthquake Agency,Yulong 674100,Yunnan,China)

compressive sensing; the Jitter sampling; seismic exploration; data reconstruction

DOI: 10.20015/j.cnki.ISSN1000-0666.2025.0009

备注

受野外复杂环境影响,地震勘探采集的地震数据往往不完整且有坏道。为提高原始地震数据的完整度,基于压缩感知稀疏反演理论,构建了一种基于Jitter采样的压缩感知地震勘探数据重构方法。从采样模型和采样信号频谱分析等方面详细对比3种采样方法的优缺点。通过合成地震数据测试,从信噪比、均方根误差、互相关系数3个方面进行重建效果综合评价并应用于实际数据中。结果表明:与传统随机采样方法相比,基于Jitter采样方法重构前后的地震数据形态振幅一致性更强,信噪比更高、误差更小、互相关系数更高。实际数据应用结果显示:重构后的叠后地震成像数据同相轴清晰,连续性更强,振幅一致性强,对噪声压制较好。这表明Jitter采样在保持随机性采样的同时,可以有效控制采样间隔,解决了采样点过于分散或者过于集中的问题,更有利于数据的恢复。综上,Jitter采样方法能够从稀疏不均匀数据中重建出密集规则化的地震数据,可以为后续的高质量偏移成像、速度建模等研究提供完整地震数据支撑。
In the process of seismic exploration data acquisition,due to the complex field environment,the collected seismic data are often incomplete and have bad traces,but the completeness of the original seismic data is prerequisite for seismic data processing and interpretation. Based on the theory of compressive sensing sparse inversion,we constructed a compressive sensing method to reconstruct seismic exploration data by Jitter sampling. Then we compared our proposed sampling with the Gaussian sampling method and the regular undersampling method from the aspects such as their sampling model,spectrum analysis of sampled signal. Further,we tested the synthetic seismic data and,aiming at root-mean-square error,signal to noise ratio,cross correlation coefficient,we made a comprehensive evaluation of the reconstruction effect of the three methods. Comparing the three sampling methods,we found that the original seismic data and the reconstructed data by Jitter method have better consistency in the form and the amplitude,higher signal to noise ratio,less error,and larger correlation coefficient. The practical application of the reconstructed seismic data by Jitter method proves that these data have cophasal axes,good continuity,good consistency in amplitudes,and good effect for noise suppression. This indicates that the Jitter sampling method is able to do random sampling and in the meantime control the sampling interval and in this way make the sampling points not too scattered nor too concentrated. This is helpful for data recovery. In conclusion,the Jitter sampling method is able to reconstruct the dense and regularized data from the original,sparse,and uneven seismic data,thus provides the complete seismic data for the follow-up migration imaging and velocity modeling.