论文详情
压缩感知思想在油气勘探领域已经受到了充分的关注。主要体现在两个方面,一个是与随机采样有关的随机地震数据采集系统的设计与评价、同时源激发的混叠数据分离、空间非规则数据的规则化、地震数据的去噪和地震数据压缩(传输)等;另一方面是地震数据的特征表达、模型参数(如速度模型参数)的特征表达、模型特征表达情况下的地震波正演模拟以及对应的稀疏反演方法(包括多尺度反演方法)。前者可以直接借鉴信号和图像处理领域发展的各种方法;后者需要勘探地球物理学家及相关领域研究人员提出合适的思想和方法。来自以沉积地层为主的地下介质中的反射(绕射)波场是有特征的或可压缩的;以层状地层为主的模型参数也是可压缩的。这是压缩感知方法能用于地震勘探的基础。首先简述了压缩感知的基本思想，指出它是一套随机采样理论。然后阐述了无论是叠前地震数据或是叠前偏移成像结果都是可以稀疏表达的，压缩感知理论可以在地震勘探中得到有效应用。接着从随机采样、同时源激发的混叠数据的分离、高维数据规则化方面展示了压缩感知思想的应用。关于随机采样，指出当前地震数据采集的技术重点应该是更高效地实现“两宽一高”的数据采集。最后列举了压缩感知思想在特征数据提取、编码炮集成像和特征波场成像中的应用。
Compressed sensing (CS) has been given full attention in the domain of oil and gas seismic exploration.The idea of CS is mainly implemented in two aspects,one is random-sampling based applications,such as seismic acquisition design and evaluation,de-blending of simultaneous-source data,regularization of seismic data,denoising and compression of seismic data; the other is the characteristic expression of data and model parameters,and the corresponding forward modeling and sparse inversion methods (including multi-scale inversion).The former one can be solved by using various methods developed in signal and image processing fields while the latter one need geophysicists to develop new ideas and methods.Both the reflection scattering waves and layered models are compressible,which make is possible for the successful application of CS in exploration seismology.The basic idea of CS is concisely sketched and it is pointed out that CS is a theoretical frame of random sampling.And then we demonstrate that both the prestack seismic data and its migration results can be sparsely expressed.CS can be reasonably used in seismic exploration.We show its applications in random sampling,de-blending of simultaneous-source data and regularization of multi- dimensional seismic data.It is indicate that the current seismic data acquisition technology development should be focused on how to truly implement the so-called “broadband and wide-azimuth and high sampling density” seismic data acquisition.At last,the application of CS in characteristic data extraction,encoding shot gathers imaging and characteristic wavefield imaging are enumerated.
国家自然科学基金项目(41374117)和国家科技重大专项(2011ZX05005-005-008HZ,2011ZX05006-002,2011ZX05023)联合资助。