Quantifying Similarity in Seismic Polarizations
Quantifying Similarity in Seismic Polarizations
复制标题
量化地震偏振的相似性
DOI：
--
复制
发表时间：
2015
期刊：
影响因子：
--
通讯作者：
E. Caffagni
中科院分区：
文献类型：
--
作者：
D. Eaton;Joshua P. Jones;E. Caffagni
评估地震属性的相似性可以帮助识别震颤、低信噪比（S/N）信号和转换或反射相位，此外还可以诊断现场噪声和阵列中的传感器未对准。极化分析是一种被广泛接受的方法，用于通过计算属性来研究地震相位的方向和方向特征，但通常使用与参考值或已知震源的定性比较来讨论相似性。在这里，我们介绍了一种技术的定量极化相似性，使用加权直方图计算在短，重叠的时间窗口，借鉴的方法，从图像处理和计算机视觉文献。我们的方法占模糊的方位角和入射角和S/N比的变化。测量偏振相似性可以很容易地识别站点噪声和传感器未对准，并可以帮助识别相干噪声和紧急或低S/N相位到达。相位到达期间的不同方位角指示未对准的水平分量，相位到达期间的不同入射角指示未对准的垂直分量，并且不同的线性偏振可以指示次级噪声源。使用记录的Mw = 8.3鄂霍次克海地震，从加拿大国家地震台网宽带传感器在不列颠哥伦比亚省和育空地区，加拿大，和一个垂直钻孔阵列Hoadley气田，加拿大中部阿尔伯塔，我们证明了我们的方法是强大的站间距。离散小波分析以一种直接的方式将极化相似性扩展到时间-频率域。钻孔数据的时间-频率极化相似性表明，相干噪声源可能在从“回流”型水力压裂的峰值资源提取后几个月持续超过8 Hz。
S U M M A R Y Assessing the similarities of seismic attributes can help identify tremor, low signal-to-noise (S/N) signals and converted or reflected phases, in addition to diagnosing site noise and sensor misalignment in arrays. Polarization analysis is a widely accepted method for studying the orientation and directional characteristics of seismic phases via computed attributes, but similarity is ordinarily discussed using qualitative comparisons with reference values or known seismic sources. Here we introduce a technique for quantitative polarization similarity that uses weighted histograms computed in short, overlapping time windows, drawing on methods adapted from the image processing and computer vision literature. Our method accounts for ambiguity in azimuth and incidence angle and variations in S/N ratio. Measuring polarization similarity allows easy identification of site noise and sensor misalignment and can help identify coherent noise and emergent or low S/N phase arrivals. Dissimilar azimuths during phase arrivals indicate misaligned horizontal components, dissimilar incidence angles during phase arrivals indicate misaligned vertical components and dissimilar linear polarization may indicate a secondary noise source. Using records of the Mw = 8.3 Sea of Okhotsk earthquake, from Canadian National Seismic Network broad-band sensors in British Columbia and Yukon Territory, Canada, and a vertical borehole array at Hoadley gas field, central Alberta, Canada, we demonstrate that our method is robust to station spacing. Discrete wavelet analysis extends polarization similarity to the time–frequency domain in a straightforward way. Time–frequency polarization similarities of borehole data suggest that a coherent noise source may have persisted above 8 Hz several months after peak resource extraction from a ‘flowback’ type hydraulic fracture.