西南印度洋脊海底地震仪缺失数据稀疏约束重建方法研究

The Southwest Indian Ridge is a typical ultraslow-spreading ridge. The crustal structures detection by applying Ocean Bottom Seismometer (OBS) is very important for the seabed evolution process study and polymetallic sulphide exploration. However, the phenomena of OBS station data and shot data losses are very serious caused by failure instruments retrievement, data recording problems, void airgun shooting and other problems in recent OBS cruises, which are seriously affecting the structural imaging and geological interpretation. To the above problem, the latest developed Compressed Sensing theory in the fields of signal processing and mathematics is introduced to study the sparsity principles, random missing model and missing data reconstruction method for OBS data in this project. This project focuses on breaking the key scientific problems, such as the sparse representation based on dictionary learning and the high-precision reconstruction with rare receiving points in relation to OBS wavefields. A set of OBS missing data reconstruction method based on Compressed Sensing is finally established. It is the innovation to establish a new data reconstruction method with independent intellectual property rights by introducing Compressed Sensing theory into the reconstruction of OBS missing data at Southwest Indian Ridge. The research results can provide theoretical supports for solving the problem of OBS data missing problems at Southwest Indian Ridge, which is of great significance to improve the crustal structure detection level, and to promote the submarine scientific research and resources exploration.

西南印度洋脊属于典型的超慢速扩张洋脊，利用海底地震仪（OBS）探测其地壳结构对海底演化过程认知和多金属硫化物勘查具有重要意义。然而，近年来开展的OBS探测中，设备丢失、记录障碍和废炮等引起站位数据和炮数据缺失问题突出，严重影响构造成像和地质解释。针对上述问题，本项目引入信号处理和数学领域最新发展的压缩感知理论，开展OBS数据稀疏性基本理论、OBS数据随机缺失模型和OBS缺失数据重建方法研究。重点突破基于字典学习的OBS波场稀疏表示和接收点稀少的OBS波场高精度重建等关键科学问题，最终建立一套基于压缩感知的OBS缺失数据重建方法理论。将压缩感知理论引入西南印度洋脊OBS缺失数据重建中，建立一套全新的具有自主知识产权的数据重建方法理论是本研究的创新之处。研究成果可为解决西南印度洋脊OBS数据缺失问题提供理论支撑，对提高超慢速扩张洋脊地壳结构探测技术，进而推动海底科学研究和资源勘查具有重要意义。

开展西南印度洋脊海底地震仪缺失数据重建方法研究可为解决当前OBS数据采集中存在的数据缺失问题提供新方案。针对上述问题，本项目首先基于数值模拟和波场传播理论揭示OBS数据具备稀疏性的数学物理依据，选择多尺度多方向曲波变换作为稀疏变换。在分析带限信号Nyquist采样局限性的基础上，构建OBS随机缺失理论模型和评价准则，为数据重建提供前提和指导。提出基于迭代阈值的OBS缺炮数据重建方法以及适用于复杂数据的重建的多域稀疏约束重建方法，分析了震源频率和缺失模型对重建的影响。提出了基于压缩感知和地震干涉理论的OBS站位数据重建方法，一定程度上适用于大间距站位OBS缺失数据重建。通过上述研究，本项目突破了OBS随机缺失模型构建、稀疏约束反演高精度数据重建和稀疏约束-地震干涉联合数据重建技术，初步构建了一套OBS数据重建方法理论，通过数值试验证实了方法的有效性，为解决OBS数据缺失问题、推进海底探测技术发展提供理论支撑。