稀疏反演求解基追踪降噪问题的地震谱分解方法

摘要

Spectral decomposition was described as a linear inversion problem.As this problem is underdeter-mined,it needs sparse inversion algorithm to solve the problem of Basic Pursuit De-noising (BPDN).The authors use Spectral Projected Gradient --L1 (SPGL1 )to increase the resolution of Inversion Spectral Decomposition (ISD),and further study its potential advantages.The results show that ISD based on BPDN (ISD--BPDN)has higher resolution in time-frequency representation,which can accurately distinguish the formation,and can detect precisely the presence of hydrocarbons.%谱分解描述为一线性反演问题，由于该问题欠定性，需使用稀疏反演算法，然后将该问题转化为基追踪降噪问题（BPDN），引入谱投影梯度（SPGL1）稀疏反演算法提高反演谱分解（ISD）分辨率，并进一步研究其潜在优势。结果表明：基于基追踪降噪问题的反演谱分解方法（ISD －BPDN）在烃类检测中有更高的时频分辨率，分层更准确，精细地指示了烃类的存在。