多信息融合的近地表速度建模技术及应用

Near-surface model building with multi-discipline information fusion

为了提高复杂区近地表速度模型的构建精度,提出并实现了基于多信息融合的近地表速度模型构建技术。采用协同克里金的插值方法,根据初至反演的近地表速度与近地表调查速度的相关性及变化趋势,确定对应的变差函数和用于插值的权系数,对速度变量进行协方差的空间估计,使横向的速度变化趋势既与大炮初至层析反演的结果较吻合,数值上又与近地表调查的结果基本一致。该方法可以提高复杂近地表速度模型的精度。在中国西部复杂区地震资料的应用取得了较好的效果,进一步验证了该方法的有效性。

In order to build a relatively accurate near-surface velocity model,we put forward in this paper a new method based on multi-discipline information fusion.First we calculate corresponding variogram functions and weights for the interpolation with Cokriging interpolation based on velocity correlation and change trend between first break tomography inversion and near-surface survey. Then we use covariance to estimate spatial velocity variables,and obtain a final velocity model in which lateral velocity variation trend is consistent with both first break tomography inversion and near-surface survey velocity. So this method improves velocity model accuracy. Good results have been achieved in rugged surface area,Western China,which verifies the effectiveness of the proposed method. © 2017, Editorial Department OIL GEOPHYSICAL PROSPECTING. All right reserved.