Search

The full-waveform inversion (FWI) of ground-penetrating radar (GPR) data yields promise for quantitatively characterizing the parameters of the Earth’s shallow subsurface. However, conventional FWI is highly nonlinear and suffers from cycle skipping once the low-frequency data are missed or the initial model is poor. Furthermore, having limited prior knowledge of the subsurface in GPR measurements increases the ill-posedness of the inverse problem. Wavefield reconstruction inversion (WRI), which mitigates cycle skipping, extends the FWI search space by relaxing the wave equation constraint, reduces the nonlinearity, and is less sensitive to the initial model. In this article, we extend WRI to the 2-D frequency-domain imaging of on-ground GPR data. To improve the inversion stability and mitigate the ill-posedness, we utilize modified total variation (MTV) regularization to constrain the inverted models. With a simple numerical example, we first investigate the effects of the penalty parameter, initial models, and MTV regularization on WRI and further discuss the differences between WRI and FWI. Then, we analyze the sensitivity of the proposed approach to the frequency component and noise in a multiple-targets model. Furthermore, we assess our method with a complex synthetic example containing noise-contaminated data, showing that our proposed approach works efficiently even given noisy GPR data. Therefore, the reasonable combination of WRI and MTV regularization can improve the accuracy and efficiency of imaging for on-ground GPR data. This joint approach for the multiparameter quantitative reconstruction of GPR data ultimately exhibits good applicability and strong robustness and is worthy of promotion. [ABSTRACT FROM AUTHOR]

Full waveform inversion (FWI) of ground-penetrating radar (GPR) data is becoming a promising technique to facilitate the interpretation of surface-GPR data and the mapping of the subsurface. However, more general FWIs still require a sufficient amount of RAM memory, and it is difficult to produce an accurate and representative reconstruction result due to a large amount of the Hessian matrix calculations and singular value decomposition (SVD). In this article, we developed a novel full-waveform approach of the limited-memory Broyden–Fletcher–Goldfarb–Shanno (L-BFGS) algorithm for surface-GPR data that is based on a quasi-Newton framework and the total variation (TV) regularization. The proposed approach uses an L-BFGS algorithm and combines a scale-transformation regularization technique to mitigate the ill-posed problem of inversion, which can impose a biparameter preinformation constraint to ensure the stability of inversion, and adaptive regularization weights are applied to improve the convergence efficiency of inversion. To demonstrate the novelty and effectiveness of the proposed scheme, we tested our FWI algorithm using synthetic data and in-site field data. In the testing, we focus on analyzing the influence of different aspects of the FWI results, including different scale factors, regularization weights, inversion strategies, acquisition configurations, initial models, and the noisy data set. In particular, the FWI experiment is performed to demonstrate the applicability of the proposed algorithm. The results show that the proposed algorithm can effectively reconstruct the biparameter near the subsurface with high accuracy, which makes our approach very attractive for attribute analysis applications and makes the surface-GPR FWI commercially viable. [ABSTRACT FROM AUTHOR]

To address problems, such as the computationally intensive inversion requirements, low inversion efficiency, and inadequate inversion accuracy caused by multiparameter crosstalk in a synchronous inversion, a new dynamic stochastic source encoding strategy combined with minmax-concave total variation (MCTV) regularization model constraints was proposed. This strategy avoids crosstalk noise between shots caused by the algorithm and greatly improves the inversion efficiency without affecting the inversion accuracy. By comparing a “cross”-shaped model with the multiparameter inversion results, we found that the MCTV regularization strategy boasts the best inversion effect. We further showed that dynamic stochastic source encoding can increase the inversion efficiency threefold by applying the 1994 British Petroleum (BP) migration international standard topography model and establishing a function to evaluate the most efficient inversion strategy from among seven options. Compared with the traditional stochastic source encoding strategies, dynamic stochastic source encoding was shown to better suppress crosstalk noise. The proposed strategy also presented a higher acceleration ratio; additionally, combining this strategy with MCTV regularization model constraints provided the clearest reconstructed image with the highest inversion precision and obtained the best evaluation score among the considered inversion strategies, albeit with a slight reduction in the total elapsed time-acceleration ratio. [ABSTRACT FROM AUTHOR]