Improving the imaging capabilities of modern portable loop-loop electromagnetic induction (EMI) systems using ground-penetrating radar (GPR) data

在近地表地球物理应用中，便携式环回电磁感应（EMI）传感器越来越多地用于快速成像相当大区域（几公顷）地下最上层的电导率。由于许多岩石、土层和人为材料的电导率存在差异，因此由此产生的电导率3D模型可以用来描述大量目标的特征。然而，由于地下材料的电导率受到许多不同的土壤和岩石性质的影响，电磁干扰电导率模型的解释往往是复杂和非独特的；特别是在没有可靠的地下成像背景信息的情况下。此外，即使已知目标的性质，EMI数据在结构分辨率方面也受到限制，因为每次测量都对地下的整体体积很敏感。探地雷达（GPR）方法是另一种流行的近地表成像方法。探地雷达作为一种对介质介电常数对比敏感的波基成像技术，通常被认为是提供最高结构分辨率的地球物理方法。然而，基于物理性质模型的更定量的分析往往局限于典型的2D/3D GPR反射数据。虽然电磁干扰和探地雷达方法都用于探测相似的深度范围，但它们从未在定量综合成像/反演程序的框架中结合起来。考虑到每种方法的优缺点以及它们可以提供互补的信息，我们假设定量集成可以改善地下结构和性质的表征。在本项目中，我们建议开发和评估定量结合EMI和GPR数据的方法，以减少仅使用EMI方法时遇到的经典模糊性和分辨率限制。为此，我们将首先通过在几种类型的受控目标上比较三种不同的电磁干扰数据反演策略来研究电磁干扰方法的典型非唯一性。然后，我们将专注于将从探地雷达数据中导出的结构纳入EMI数据的反演，并研究这种策略如何有助于减少反演EMI模型的非唯一性。在这方面，我们将开发和评估两种约束反演策略：一种基于确定性网格的方法，最近在文献中报道了用于更大规模问题的方法，另一种是随机参数方法。因此，我们期望从这个项目中得出关于结合电磁干扰和探地雷达数据的可能性的一般性结论，以及关于电磁干扰数据反演的方法创新，进一步提高成像能力和电磁干扰方法的适用性。

In near-surface geophysical applications, portable loop-loop electromagnetic induction (EMI) sensors are increasingly used to rapidly image the electrical conductivity of the uppermost meters of the subsurface across rather large areas (several hectares). The resulting 3D models of electrical conductivity can serve to characterize a large selection of targets because many rocks, soil layers and anthropogenic materials show contrast of electrical conductivity. However, because electrical conductivity of subsurface materials is influenced by many different soil and rock properties, the interpretation of the EMI electrical conductivity models is often complex and non-unique; especially, in contexts where no reliable background information about the imaged subsurface is available. Moreover, even if the nature of the targets is known, EMI data are limited in terms of their structural resolution capabilities because each measurement is sensitive to an integrated volume of subsurface.The ground-penetrating radar (GPR) method is another popular near-surface imaging method. As a wave-based imaging technique, which is sensitive to dielectric permittivity contrasts, GPR is typically considered as the geophysical method providing the highest structural resolution. However, a more quantitative analysis in terms of physical property models is often limited with typical 2D/3D GPR reflection data. Although both EMI and GPR methods are used to explore similar depth ranges, they have never been combined in the framework of a quantitative integrated imaging/inversion procedure. Considering weaknesses and strengths of each method and that they can provide complementary information, we hypothesize that a quantitative integration results in an improved characterization of subsurface structures and properties. In this project, we propose to develop and evaluate approaches for quantitively combining EMI and GPR data in order to reduce the classical ambiguities and resolution limitations encountered when using the EMI method only. In doing so, we will first study the typical non-uniqueness of the EMI methods by comparing three different EMI data inversion strategies on several types of controlled targets. Then, we will focus on incorporating the structures as derived from GPR data into the inversion of EMI data, and study how such a strategy helps to reduce the non-uniqueness of the inverted EMI models. In this respect, we will develop and evaluate two constrained inversion strategies: one deterministic grid-based approach, which was recently reported in the literature for larger scale problems, and one stochastic parametric approach. Thus, we expect from this project general conclusion regarding the possibilities of combining EMI and GPR data as well as methodological innovations regarding EMI data inversion further improving the imaging capabilities and the applicability of the EMI method.