在被动源面波成像中低质量的频谱图会直接导致后续频散曲线的拾取出现较大的不确定性,为了进一步提高被动面波频谱成像图的质量,同时提高频散曲线的拾取的准确性。文章提出了将在原始数据重构得到的虚拟道集上应用CBDNet进行被动源面波...在被动源面波成像中低质量的频谱图会直接导致后续频散曲线的拾取出现较大的不确定性,为了进一步提高被动面波频谱成像图的质量,同时提高频散曲线的拾取的准确性。文章提出了将在原始数据重构得到的虚拟道集上应用CBDNet进行被动源面波信号增强,降低频散成像能量谱中的噪声、增强频散曲线连续性。文中首先通过数值模拟的方式,验证了网络的有效性;然后将其用于北京市广渠门和通州两个地区的实际被动源数据处理中。结果表明,文章提出的网络和被动源面波去噪方法能够很好的解决实际工程中频谱图质量低,频散曲线错位和不连续的问题。In passive-source surface wave imaging, low-quality spectrograms directly lead to significant uncertainties in the subsequent picking of dispersion curves. To further enhance the quality of passive surface wave spectral imaging and improve the accuracy of dispersion curve picking, this study proposes applying CBDNet to virtual shot gathers reconstructed from the raw data. This approach enhances passive-source surface wave signals, reduces noise in the dispersion imaging energy spectrum, and improves the continuity of dispersion curves. First, the effectiveness of the network is validated through numerical simulations;then, it is applied to real passive-source data from two regions in Beijing, Guangqumen and Tongzhou. The results demonstrate that the proposed network and denoising method for passive-source surface waves effectively address the issues of low spectrogram quality, misaligned, and discontinuous dispersion curves in practical engineering applications.展开更多
文摘在被动源面波成像中低质量的频谱图会直接导致后续频散曲线的拾取出现较大的不确定性,为了进一步提高被动面波频谱成像图的质量,同时提高频散曲线的拾取的准确性。文章提出了将在原始数据重构得到的虚拟道集上应用CBDNet进行被动源面波信号增强,降低频散成像能量谱中的噪声、增强频散曲线连续性。文中首先通过数值模拟的方式,验证了网络的有效性;然后将其用于北京市广渠门和通州两个地区的实际被动源数据处理中。结果表明,文章提出的网络和被动源面波去噪方法能够很好的解决实际工程中频谱图质量低,频散曲线错位和不连续的问题。In passive-source surface wave imaging, low-quality spectrograms directly lead to significant uncertainties in the subsequent picking of dispersion curves. To further enhance the quality of passive surface wave spectral imaging and improve the accuracy of dispersion curve picking, this study proposes applying CBDNet to virtual shot gathers reconstructed from the raw data. This approach enhances passive-source surface wave signals, reduces noise in the dispersion imaging energy spectrum, and improves the continuity of dispersion curves. First, the effectiveness of the network is validated through numerical simulations;then, it is applied to real passive-source data from two regions in Beijing, Guangqumen and Tongzhou. The results demonstrate that the proposed network and denoising method for passive-source surface waves effectively address the issues of low spectrogram quality, misaligned, and discontinuous dispersion curves in practical engineering applications.