基于压缩感知的反射系数域沿层L2范数约束去强屏蔽方法

常规的压缩感知方法均是基于反射系数稀疏反演提高地震资料分辨率,而对于强屏蔽层信息和储层弱信息耦合叠加情况而言,效果并不理想,原因在于在剥离强反射层的同时往往会损失弱反射层的信息,从而不能有效预测储层。为此,基于压缩感知稀疏表示能力强的特点,利用反射系数域分辨率高和不存在子波重叠的优势,提出了基于压缩感知的反射系数域沿层L₂范数约束去强屏蔽方法。该方法基于压缩感知理论,首先根据时域反射系数域稀疏特性,利用沿层信息分离强屏蔽层与储层,再进行稀疏反演,最后将去强屏蔽后的反射系数与原子波褶积,从而获得去强屏蔽层的高分辨率结果。其优势在于高分辨率的反射系数能够分离强屏蔽层与储层的信息,有利于拾取和剥离强屏蔽层。模型测试和实际地震资料的应用表明:利用所提方法可以精准分离强屏蔽层反射信息与储层弱反射信息,提高储层识别精度;在剥离强屏蔽层的时频切片上,可以看见表征储层的弱能量团,并出现低频伴影现象;在剥离强屏蔽层的沿层能量半时属性切片上,能量半时属性与滩坝砂体储层的相关性较好,可以有效识别有利储层区域。

A strong shielding removal method of reflection coefficient domain based on compressed sensing with L2 norm constraint along layer

Conventional compressed sensing methods are based on the sparse inversion of reflection coefficients to improve the resolution of seismic data. However, if the strong shielding layer information and the weak reservoir information are superimposed, the reservoir cannot be effectively predicted, because the information of the weak reflection layer will be lost when the strong shielding layer is removed. For this reason, taking advantage of the strong sparse representation ability of compressed sensing and the high resolution of the reflection coefficient domain without wavelet overlap, this paper proposes a strong shielding removal method of reflection coefficient domain based on compressed sensing with L₂ norm constraint along the layer. This method is based on the theory of compressed sensing. First, according to the sparse characteristics of the reflection coefficient in the time domain, the strong shielding layer and the reservoir are separated using the information along the layer. Then the sparse inversion is performed, and finally the reflection coefficient after strong shielding removal is convolved with the original wavelet to obtain the high-resolution results in the absence of a strong shielding layer. The advantage of the method is that the high-resolution reflection coefficient can separate the information of the strong shielding layer and the reservoir, which is beneficial to extract and remove the strong shielding layer. The model tests and actual seismic data application show that the proposed method can accurately separate the reflection information of the strong shielding layer from the weak reflection information of the reservoir and thus improve the accuracy of reservoir identification. On the time-frequency slices after strong shielding layer removal, the weak energy of the reservoir can be seen, and low-frequency accompanying phenomena appear. On the energy half-time attribute slices along the layer after strong shielding removal, the attribute has a good correlation with the beach-bar sand reservoir, which can effectively identify favorable reservoir areas.