An interpolating scaling functions method with low-storage five-stage fourth-order explicit Runge-Kutta schemes for 3D ground penetrating radar simulation

Xun Wang,Deshan Feng,Daoxue Yang,Xiangyu Wang
DOI: https://doi.org/10.1016/j.jappgeo.2020.104128
IF: 1.845
2020-09-01
Journal of Applied Geophysics
Abstract:Accurate and efficient 3D simulation is extremely important for both grasping the features of radar wave propagation in realistic models and the high-resolution inversion. In this paper, we propose a new numerical algorithm called the low-storage five-stage fourth-order explicit Runge-Kutta interpolation scaling function method (LSERK-ISFM) and apply it into the 3D simulation of ground penetrating radar (GPR). Firstly, we derive the auxiliary differential equation (ADE) of the 3D Maxwell's equations based on the complex-frequency shifted perfectly matched layer (CFS-PML), discretize the spatial derivative of the Maxwell equation by ISFM, and integrate the temporal dimension through the LSERK scheme. Compared with the traditional finite-difference time-domain (FDTD) method, the LSERK-ISFM algorithm can effectively improve the precision of space-time dispersion of the Maxwell's equations. Specifically, by the homogeneous vacuum medium model, we compare the forward modeling results of the different number of grids per wavelength, the comparison results prove that a smaller number of grids per wavelength can be selected in the LSERK-ISFM scheme, which can greatly contribute to reducing the consumption of memory. Then we build a simple 3D model to properly compare the proposed algorithm with the coarse-mesh FDTD and the fine-mesh FDTD, the results show that, with the same mesh generation, the LSERK-ISFM algorithm outperforms the conventional FDTD algorithms in terms of both accuracy and efficiency. Finally, we test the algorithm on a 3D complex model by a series of 3D forward profiles and the corresponding slices, both simulated reflections and diffractions clearly depict the shape and location of the anomalous bodies, which shows great potential of application in the field of 3D exploration and interpretation for GPR community.
geosciences, multidisciplinary,mining & mineral processing
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: to improve the accuracy and efficiency of three - dimensional ground - penetrating radar (3D GPR) simulations. Specifically, the author proposes a new numerical algorithm - the low - storage five - stage fourth - order explicit Runge - Kutta interpolation scale function method (LSERK - ISFM), and applies it to 3D GPR simulations. ### Main problems 1. **Spatial discretization accuracy**: - The traditional finite - difference time - domain method (FDTD) requires more grid points to ensure accuracy during spatial discretization, which leads to higher memory consumption and computing time. - The newly proposed LSERK - ISFM algorithm improves the accuracy of spatial discretization through the interpolation scale function method (ISFM), enabling high - accuracy results to be obtained with fewer grid points. 2. **Temporal discretization accuracy**: - Errors in temporal discretization also affect the accuracy of GPR simulations. The traditional central difference method has lower accuracy, while the explicit Runge - Kutta (ERK) method can provide higher temporal discretization accuracy. - LSERK - ISFM adopts a low - storage five - stage fourth - order explicit Runge - Kutta scheme, which not only improves the accuracy of temporal discretization but also reduces memory usage. 3. **Numerical dispersion problem**: - Numerical dispersion is an important factor affecting the accuracy of electromagnetic wave propagation simulation. The ISFM method performs better than the traditional FDTD method in terms of numerical dispersion, especially under coarse - grid conditions. ### Solutions - **Auxiliary differential equation (ADE)**: Based on the complex - frequency - shifted perfectly matched layer (CFS - PML), derive the auxiliary differential equation of the 3D Maxwell equations, and use ISFM to discretize the spatial derivatives and LSERK to integrate the time dimension. - **Interpolation scale function method (ISFM)**: Construct the interpolation scale function through an iterative interpolation process to achieve spatial discretization of the Maxwell equations and reduce numerical dispersion. - **Low - storage five - stage fourth - order explicit Runge - Kutta (LSERK)**: Used for integration in the time dimension to improve the accuracy of temporal discretization and reduce memory consumption. ### Experimental verification - **Single - channel radar waveform analysis**: By comparing the single - channel radar waveforms under different grid densities with the analytical solutions, the superiority of the LSERK - ISFM algorithm under coarse - grid conditions is verified. - **3D model verification**: By constructing a complex 3D geological model and comparing the results of LSERK - ISFM with those of the coarse - and - fine - grid FDTD methods, the advantages of the new algorithm in terms of accuracy and efficiency are proven. In summary, this paper aims to improve the accuracy and efficiency of 3D GPR simulations, reduce memory consumption, and effectively control the numerical dispersion problem by introducing the LSERK - ISFM algorithm.