基于ARMA模型的剩余子波反褶积方法

ARMA MODEL-BASED DECONVOLUTION

1. 清华大学,北京100084;2. 大庆石油管理局,大庆163357

1. Qinghua Univerzity, Beijing 100084;2. Daqing Oil Field Administration Beauru, Daqing161551

反褶积是提高地震记录分辨率的主要手段.影响反褶积效果的主要因素是:(1)子波模型假设是否合理;(2)子波幅度谱估计是否准确;(3)子波相位谱估计是否准确.本文针对这三个方面提出了基于ARMA模型的剩余子波反褶积方法,可以较好地解决上述前两个问题.在叠前反褶积中,本文假设剩余子波为最小相位.在叠后反褶积中,本文论证了剩余子波AR部分必为最小相位,而子波MA部分可以用零相位近似.实际数据验证表明,本文的方法是有效的.叠后反褶积效果更为明显.

Dcconvoluton is one of the main tools in enhancing resolution for seismic records. The factors that affect the dcconvolution performance arc: 1) whether the wavelet model is appropriate, 2) whether the wavelet's power spectrum is accurately estimated, and 3) whether the wavelet's phase charctcr is correctly specified. In this paper, we present an ARMA model-based dcconvolution method for solving the above three problems. In prcstack dcconvolution, we assume that the wavelet is of minimum phase, While in post-stack dcconvolution, we have shown that the AR part of wavelets must be of minimum phase and the MA part of wavelets can be approximated by a zero-phase sequence. We have tested the method on both prcstack and post-stack data and achieved good results in either case especially the latter one.