【摘要】 本文提出了一种新的反褶积方法——谱模拟反褶积方法。此法基本上克服了统计性反槽积方法中反射系数序列在计算子波自相关过程中带来的影响。在地震子波具有光滑报幅谱的假设条件下，首先对地震数据振幅谱进行模拟，得到比较精确的自相关估计值；然后在不改变常规统计性反褶积方法对地震子波相位假设的条件下，得到更好的子波估计值；再通过对反射系数有色成分的模拟，得到与反射系数有色成分有关的算子；最后通过对地震数据进行褶积运算，就能获得比常规反褶积方法好得多的结果。更多还原

【Abstract】 Spectral-modeled deconvolution, a new method, is advanced here. The methodalmost removes influence resulting from reflection coefficient sequence in waveletautocorrelation estimation in statistical deconvolution. The method works in thefollowing steps:. Estimate quite accurate autocorrelation values by modeling amplitude spec-trum of seismic data on the presumption that seismic wavelet has smooth amplitudespectrum;. Obtain better wavelet estimations without changing the presumptive condi-tion of wavelet phase in statistical deconvolution;. Evaluate the operator relating to coloured contents of ref1ection coefficientsby modeling the coloured contents;. Gain the result much better than that of conventional deconvolution by doingconvolution of seismic data.更多还原