Inversion of Seismic Refraction and Reflection Data for Long-wavelength Velocity Model
Inversion of Seismic Refraction and Reflection Data for Long-wavelength Velocity Model
复制标题
DOI：
10.3997/2214-4609.20141088
复制
发表时间：
2014-06
期刊：
影响因子：
--
通讯作者：
Haiyang Wang;Satish C. Singh;F. Audebert;H. Calandra
中科院分区：
文献类型：
--
作者：
Haiyang Wang;Satish C. Singh;F. Audebert;H. Calandra
我们已经应用波动方程层析折射和反射数据检索的长波长地下速度模型。为了模拟反射波，特别是，长波长和短波长结构被分离并分别映射到速度和密度。为了匹配真实的反射波的振幅，将全波形反演应用于近偏移距数据，其中所有短波长阻抗对比度被映射到密度。然后将短波长密度模型转换为零炮检距走时域，在该走时域中，它独立于长波长速度模型。当使用波动方程层析成像更新速度时，密度被转换回深度域以用于波场建模。为了减轻反演问题的非线性，采用偏移距延拓策略，首先使用折射波提供对速度模型的垂直约束，然后包括反射波提供横向约束。由于波动方程层析成像已经得到了较好的长波长速度模型，因此初始化FWI反演所有数据以恢复短波长速度结构。该策略被应用到一个海洋真实的数据集的例子与15公里长的拖缆。我们的研究结果表明，可以恢复详细的速度信息，从一个约束条件差的模型与适当的正则化。
We have applied wave-equation tomography to both refraction and reflection data to retrieve the long-wavelength subsurface velocity model. To simulate reflection waves, especially, long- and short-wavelength structures are separated and mapped into velocity and density, respectively. To match the amplitude of the real reflection waves, full waveform inversion is applied to the near-offset data, where all the short-wavelength impedance contrasts are mapped into density. The short-wavelength density model is then converted into zero-offset travel time domain where it is independent of the long-wavelength velocity model. As velocity is updated using wave-equation tomography, density is converted back into depth domain for wave-field modeling. To alleviate the nonlinearity of the inversion problem, an offset continuation strategy is adapted that refraction waves are used first to provide vertical constrains on the velocity model, after which, reflection waves are included to provide lateral constrains. As wave-equation tomography has derived the satisfactory long-wavelength velocity model, FWI is initialized to invert all the data to retrieve the short-wavelength velocity structures. The strategy is applied to one marine real dataset example with 15-km-long streamer. Our results shows that one can recover detailed velocity information starting from a poorly constrained model with appropriate regularization.