外推算子预测去噪方法

The extrapolation operator prediction and noise removal method

河南石油勘探局地调处物探研究所,南阳473132

Geological Survey Division, Henan Petroleum Exploration Bureau, Nanyang 473132

由于地震数据不同频段有不同的信噪比,常规的去噪方法对信噪比低于一定程度的频率成分就起不到压制噪声的作用,相反,在这样的频率点可能还会降低信噪比。这是由于信号估计的有效性过分依赖于输入数据的信噪比本身所致。实际上,有效反射信号不同频率成分间存在着一定的关系,即空间延续方向是不随频率而变化的,故尔可用高信噪比频段的信息来推算低信噪比频段的信号,从而得到更准确的信号估计结果。理论数据和实际资料处理结果表明,将这种方法用于时间域和频率域的去噪均能起到压制噪声的效果,有利于展宽有效频带,提高分辨率。

As S/N ratio is different in varied bands for seismic data. The traditional noise removal method cannot depress noise for some frequence which S/N ratio is lower some level. On the contrary, in this frequence point the S/N ratio maybe lower. The result occurred because signal estimation effectiveness depends on the S/N ratio of input data too much. In fact,among varied frequencies in effective reflected signal, some relationship exists. The spatial continuation direction does not change with varied frequencies. Lower S/N ratio band can be estimated using high S/N ratio band information to obtain the accurate signal estimation result. Theoretical calculate and real data processing show the method can depress noise in both time and frequency domains, spread out the effective frequency band, and improve the resolution factor.