基于结构导向的层位自动追踪

层位信息被广泛应用于地震资料的处理和解释中。常用的自动追踪层位算法,如相关和神经网络等,通常存在稳定性弱和适应性低等问题,为此提出一种基于结构张量算法的自动追踪层位方法。首先利用希尔伯特变换对原始剖面进行预处理,增强同相轴的连续性,降低假频和噪声影响,提高方向求取的稳定度;然后利用结构张量算法,获取层位发育角度信息,运用平滑处理,进一步提高其稳定性;最后选定一个种子点作为层位追踪起始点,以角度信息为导向,追踪邻近层位点,并在该点周围的窗口范围内进行优选,确定最终层位点。理论沉积模型和实际二维地震剖面的处理结果证明了该方法的可行性和有效性,与常规方法的对比验证了其适应性和稳定性。

Horizon information is important for accurate seismic processing and interpretation.Traditional automatic horizon tracking algorithms such as correlation and neural networks mostly show instability and inadaptability to complex underground structures.An automatic horizons tracking method combined with a structure tensor algorithm is therefore proposed.Firstly,the seismic profile is preprocessed with Hilbert transform to obtain the envelope,thereby to enhance the continuity of events,reduce the influence of aliasing and noise,and improve direction extraction.Then,the structure tensor is adopted to obtain the angle of horizons,with further smoothing to improve its stability.Finally,a seed point is selected as the tracking starting point for tracking the points of adjacent horizons,using the angle information as a guide.The pre-tracked horizon points are further selected optimally within a window around the seed point to determine the best matching horizon point.Tests on field data show the adaptability and stability of the proposed method.

国家自然科学基金项目“基于地震相干层析成像的四川盆地深层天然气储层预测理论方法研究”(U1562219)资助。