Model-based layer stripping FWI with a stepped inversion sequence for GPR data
Model-based layer stripping FWI with a stepped inversion sequence for GPR data
复制标题
DOI：
10.1093/gji/ggz210
复制
发表时间：
2019-08
影响因子：
2.8
通讯作者：
N. Huai;Z. Zeng;Jing Li;Yingwei Yan;Q. Lu
中科院分区：
文献类型：
--
作者：
N. Huai;Z. Zeng;Jing Li;Yingwei Yan;Q. Lu
探地雷达(GPR)数据的全波形反演是一种很有前途的近地表参数成像技术。在本文中，我们提出了一种新的基于模型的层剥离FWI(MLS FWI)方法来重建特定适用于雷达波传播的介电常数。在模型域中，从上到下，相对介电常数逐层更新。与使用偏移和/或时间窗来提取部分观测数据以计算局部梯度的传统剥层FWI不同，模型域中的空间汉宁窗被应用于从整个道集计算的总体梯度以获得层的梯度。我们将传统的层剥离定义为基于数据的，而将提出的层剥离定义为基于模型的。最重要的是，当与信源编码相结合时，这种MLS FWI从根本上解决了在基于数据的层剥离FWI中添加窗口时需要正演所有信源位置的问题，从而大大提高了计算效率并节省了内存。此外，在这种方案中，我们还提出了一种新的阶梯式反演序列来改善深度反演。在这里，我们用两个基于地面多偏移距数据的合成例子来说明我们的方法。与频率多尺度快速傅里叶变换和基于数据的剥层快速傅里叶变换的结果相比，阶梯式反转序列的最小二乘快速傅里叶变换获得了更高分辨率的图像，并准确地重建了相对介电常数模型，有效地避免了严重的局部极小值问题。这种探地雷达快速傅里叶变换过程可以帮助反演非均匀介质中的近地表模型参数。
Full-waveform inversion (FWI) of ground-penetrating radar (GPR) data is a promising technique for the parameter imaging of near subsurface. In this paper, we present an innovative model-based layer stripping FWI (MLS FWI) approach for reconstructing dielectric permittivity that is specifically applicable to the propagation of radar waves. From top to bottom in the model domain, relative permittivity is updated layer by layer. Differing from the traditional layer stripping FWI, which uses an offset and/or temporal window to extract partially observed data for calculating a partial gradient, a spatial Hanning window in the model domain is applied to the overall gradient calculated from the entire gathers for getting the gradient of a layer. We designate the traditional layer stripping as data-based and the proposed layer stripping as model-based. Most importantly, when combined with the source encoding, this MLS FWI fundamentally solves the problem that forward simulation of all source positions is required when adding the window in a data-based layer stripping FWI, and thus greatly improves the calculation efficiency and saves the memory. Furthermore, in such a scheme, we also propose a new stepped inversion sequence to improve the inversion at depth. Here, we illustrate our approach with two synthetic examples based on on-ground multioffset data. Compared to the results of the frequency multiscale FWI and that of the data-based layer stripping FWI, the MLS FWI with the stepped inversion sequence yields substantially higher resolution images and accurately reconstructs a relative permittivity model, effectively avoiding severe local minimum problems. This GPR FWI process can help retrieve near-surface model parameters in heterogeneous media.