【摘要】 进行保边滤波需要已知反射倾角和边界信息,通常这些信息由全频率地震资料估算得到,但噪声的存在往往会降低估算的准确性。地震资料的各个频率成分具有不同的信噪比,Morlet小波具有良好的局部性能,所以用Morlet小波可以将地震数据按照倍频程分为几个分频体。对于信噪比不是太低(一般指大于2.5)的地震资料,会有2～3个分频体具有较高的信噪比。扫描上述分频体,确定它们的倾角以及边界信息,然后利用倾角及边界信息对上述分频体以及与其频率接近的分频体进行平滑,并根据是否存在边界采取不同的滤波手段,保证滤波算法的稳定性,最后将平滑后的各频段地震记录合成,得到去噪后的地震记录。理论模型和实际资料处理表明,对于信噪比大于2.5的地震资料,该方法能有效去除噪声。更多还原

【Abstract】 Boundary-preserving filtering requires the information of reflection dip and boundary.The information is usually estimated from seismic data with full frequency,but noise usually decreases the accuracy of estimation.Every frequency component of seismic data has different SNR.Morlet has good local performance,so which can divide seismic data into several division-frequency data volume in terms of octave.For seismic data with SNR higher than 2.5,at least 2—3 frequency-division data volume have higher SNR. If we scan these frequency-division data volumes,their dip and boundary can be identified which is used to smooth these data volumes and the ones with almost the same frequency.Meanwhile,in terms of whether existing boundary,different filtering means are used to ensure the stability of filtering algorithm.Finally,the seismic record of different frequencies after smoothing was synthesized to obtain seismic data with noise suppression.Theoretical model and actual data processing indicates the method can effectively remove nosie for the seismic data with SNR higher than 2.5.更多还原