论文部分内容阅读
本文讨论根据实测地震数据用一维地震脉冲响应模型估计反射系数的一种非线性反演算法的应用。已经用合成地震剖面作过关于这种算法的稳定性和唯一性的检验。由于目标函数出现一些局部极小值,模型参数的估计值表现出对初值的依赖性。这种依赖性似乎随噪音水平而增大。我们曾利用叠加且作过时间偏移的三维地面地震剖面和测井数据实施这种算法,用以估计出一口井附近深约2400m 处的高分辨率反射系数剖面。我们有幸用采样间隔4ms 的地震数据取得了采样率为2ms 的反射系数集,在此深度这相当于约2.5m 的垂向厚度,对于进行详细的储集层研究来说,需要这样高的分辨率。
This paper discusses the application of a nonlinear inversion algorithm that estimates the reflection coefficient using a one-dimensional seismic impulse response model based on measured seismic data. Synthetic seismic profiles have been used to test the stability and uniqueness of this algorithm. Due to some local minima of the objective function, the estimated values of the model parameters show a dependence on the initial value. This dependence seems to increase with noise level. We have implemented this algorithm using superimposed, time-migrated 3D surface seismic profiles and well logs to estimate the high-resolution reflection coefficient profile at a depth of about 2400 m near a well. We were fortunate to have a set of reflection coefficients with a sampling rate of 2 ms with seismic data sampled at intervals of 4 ms, at which depth this corresponds to a vertical thickness of about 2.5 m, which is required for detailed reservoir studies rate.