12 October 2022 in IEEE Transactions on Geoscience and Remote Sensing
Magnetic resonance tomography (MRT) has the advantages of direct, quantitative and unique interpretation in the field of groundwater detection. Currently, the inversion of MRT data primarily uses the QT (Q-time) inversion method based on Tikhonov regularization. However, when the heterogeneity of an aquifer is high, and the water content distribution is markedly uneven, this method removes many details in the model and cannot perform uncertainty analysis on the results. To solve these problems, we propose a Bayesian inversion for MRT data based on geostatistics. The method uses previously known geological data, such as drilling, to determine prior information model containing variograms and mixture Gaussian probability distributions for generating many stochastic realizations. Under the Bayesian framework, a modified Markov chain Monte Carlo strategy (MCMC) is used to obtain the posterior probability distributions of subsurface aquifers and hydraulic conductivity, and the results of quantitative uncertainty analysis. By comparing the inversion performance for simulated models, the imaging result of the Bayesian method is found to be markedly more accurate than that of the QT method for subsurface two-dimensional aquifers, particularly in explaining the stochastic model (i.e., the water-bearing model with an uneven distribution of water content). This method can also intuitively quantify the uncertainty of the imaging results, which mitigates the shortcomings of existing inversion methods. This paper also discusses the effects of prior information, number of chains and noise levels on the results, and also validates the effectiveness and practicability of the proposed method using field-measured data.
14 June 2024 in IEEE Transactions on Geoscience and Remote Sensing
Magnetic resonance tomography (MRT) is a geophysical exploration technique that enables the imaging of two- or three-dimensional water-bearing structures, offering distinct advantages, including noninvasiveness, quantifiability, and unique interpretability. Currently, MRT data inversion mainly relies on the Q-time (QT) inversion method. Since this method utilizes the Gauss-Newton iteration to seek the optimal solution, it involves a considerable amount of computational workload, thus consuming a significant amount of time. To overcome this challenge, this study introduces an accelerated imaging method by combining singular value decomposition (SVD) pseudo-inversion algorithm and the deep neural network algorithm. The SVD pseudo-inversion algorithm transforms MRT data into a water-bearing feature matrix containing only water content and relaxation time information by introducing a priori forward kernel function. Subsequently, neural network establishes a nonlinear mapping relationship between the water-bearing feature matrix and the spatial distribution of the water content and relaxation time in the subsurface. The SVD pseudo-inversion algorithm, by incorporating prior information, mitigates the distribution differences in MRT data caused by geological and measurement parameters. This addresses the limited applicability of deep learning methods under complex geological conditions and multiple measurement schemes. The experimental results demonstrate that the method achieves precise and rapid imaging, while also possessing effectiveness and practicality.
9 January 2025 in IEEE Transactions on Instrumentation and Measurement
Nuclear magnetic resonance (NMR) has various applications in underground engineering, particularly in tunnels, where it plays a crucial role in the early detection of potential disaster-inducing water sources. Water bodies that pose a threat in tunnels often manifest as fissure water nestled between rock masses. However, the current application of tunnel NMR detection is limited to the one-dimensional (1-D) layered models, specifically magnetic resonance sounding (MRS), which fails to capture the spatial structure of such fissure water. To precisely locate disaster-inducing water bodies, the MRS method needs to be extended to two-dimensional (2-D) structural detection. Thus, we propose a method for imaging the structure of tunnel fissure water based on multipoint geostatistical Bayesian framework. This method relies solely on data acquired from 1-D MRS measurements, complemented by borehole data obtained during the construction process, to detect 2-D structures of fissure water, without the need of complicated 2-D measurement process. Furthermore, we enhance the Markov chain Monte Carlo (MCMC) algorithm to improve the sampling performance of Bayesian method through a biased-estimation constraint strategy. Through synthetic experiments, we demonstrate the effectiveness of this method in predicting the intricate distribution of fissure water and its interconnected structure. Practical validation is provided on the basis of a series of field-measured data collected in an actual tunnel.
11 February 2025 in IEEE Transactions on Geoscience and Remote Sensing
Surface magnetic resonance sounding (MRS) offers the advantages of direct, quantitative and unique interpretations in the field of groundwater detection. The time-lapse inversion method, with its temporal continuity, has been applied to monitor the time-varying trends of the hydrological parameters of groundwater. However, the ambient noise levels in MRS data fluctuate significantly over time (daily), and the presence of low signal-to-noise ratio (SNR) data can lead to a deterioration in the results of time-lapse inversion. Thus, we propose a new time-lapse inversion of MRS data weighted by noise level estimation in this paper. Noise-weighted time-lapse inversion (NW-TLI) quantifies the reliability of each MRS dataset on the basis of noise estimation residuals and incorporates time-lapse reference weights into the inversion process, thereby ensuring that the hydrological trends are more reasonably constrained by high-SNR data. In synthetic data experiments, we demonstrate that the NW-TLI method effectively mitigates interference from adjacent low-SNR data under various complicated noisy cases. Even with multiple sets of low-SNR MRS data, NW-TLI can provide more accurate hydrological time-varying trends than conventional TLI. Additionally, we assess the impact of the temporal variability of the water-bearing model and the degree of data weighting on the interpretative accuracy and ultimately validate the practicability of the NW-TLI method via field-measured data.
3 November 2025 in Computers & Geosciences