基于道集记录的地震道内插算法研究

【作者】 平利姣；

【导师】 朱德兵；

【作者基本信息】 中南大学 ， 地球探测与信息技术， 2010， 硕士

【摘要】 浅层地震勘探在工程勘查中能够提供介质的弹性动力学相关参数而受到工程界的重视。由于浅部地质条件、施工环境的影响,浅层地震记录经常存在空间采样不足的情况,不利于后续的信号处理和精细化解释,地震道内插是目前广泛采用的解决空间采样不足的方法。现有的地震道内插算法,多是沿用深部油气地震勘探的处理模式,未考虑到浅部地震记录的复杂性。浅层地质结构对浅层地震信号采集的影响程度远大于深部信号,并且浅部地震记录受直达波、面波、人文噪声、工业噪声、地表起伏等诸多因素的影响,因此,浅部地震信号比深部记录要相对复杂,研究适合浅部地震记录的插值算法有实践意义。图像处理手段在地球物理资料的数字图形处理中有着广泛的应用,但目前还没有将其运用到地震道内插中。本文基于浅层地震记录整体(多道)规律性不明显、相邻道相关性强的特殊性,利用图像插值可充分考虑相邻近采样点的相关性特点,将图像插值运用到地震道内插中,通过采样率变换和方向插值算法,较好地实现了浅层数据的内插。BP人工神经网络有着很强的泛化能力,能够模拟复杂的函数关系,本文利用其这一特点模拟浅层地震记录道间复杂的内相关关系,以原有的地震记录为学习样本,用训练成熟的网络对有假频的浅层地震数据进行内插,该算法可达到较好的内插效果。模拟记录和实际地震记录的内插结果表明,方向插值的算法和基于BP人工神经网络的插值算法在浅层记录的地震道插值中有着良好的实用性。更多还原

【Abstract】 Shallow seismic exploration attracted great attention from the engineering as it can provide the parameters of Elastodynamics in the engineering investigation. As the impact of the shallow geological conditions and the environment of construction, insufficient space sampling often exists in the shallow seismic records, which is not conducive to the subsequent signal processing and detailed interpretation, so seismic trace interpolation is being used widely to solve this problem.The existing seismic trace interpolation mostly follow the processing mode of deep oil and gas seismic exploration, which doesn’t take into amount of the complexity of shallow seismic records. The affects of shallow geological structure to the shallow seismic signal is much stronger than the deep, and shallow seismic records are affected by the direct wave, surface wave, human noise, industrial noise, surface relief, and many other factors. So the shallow seismic records are much more complex than the deep records, and the research of interpolation algorithm which is fit for the shallow seismic records has great practical significance.The image processing is widely used in the digital image processing of Geophysics, but until now, it has not been applied to the seismic trace interpolation. This article uses image interpolation into the trace interpolation of shallow seismic base on the specificity of shallow seismic records, which doesn’t have the whole (multi-channel) regularity but with strong correlation between the adjacent channels, and makes use of the image interpolation can fully consider the proximity of the relevant characteristics of sampling points. Through the sampling rate conversion and directional interpolation algorithm, this manner can be used well in shallow seismic data interpolation.The BP artificial neural network has strong capacity of generalization, and it can simulate the complex function. So, in this paper, this characteristic is used in simulating the complex internal relationship among the channels of shallow seismic records, with using the original seismic records as learning samples. Then use the mature network in trace interpolation of the shallow seismic data with alias. This interpolation algorithm can achieve good results.The results of trace interpolation used in analog and practical shallow seismic records indicate that the directional interpolation algorithm and the interpolation algorithm based on the BP artificial neural network have good practicability.更多还原