基于Julia加速的射频大地电磁法正演模拟与位移电流作用研究

射频大地电磁法(Radio-magnetotelluric, RMT)是浅地表电磁勘探的重要手段之一。由于勘探频段为10～300 kHz，其电磁场传播受地下介质介电常数的影响较大。传统准静态条件下的电磁响应严重制约RMT正演模拟精度，并进一步影响反演成像分辨率。针对这一问题，提出了一种基于Julia并行加速的全电流RMT电磁响应数值模拟方法，利用Julia的分布式计算将各频点的计算发送到不同进程进行求解，从而达到加速计算的目的，同时在计算中考虑位移电流的影响，提升正演的模拟精度。通过计算几种典型高阻/高介电模型的RMT响应，分析并总结了位移电流对射频段电磁场视电阻率及相位响应的影响规律。数值模拟结果表明：当浅部存在高阻覆盖层时，基于准静态假设条件计算的RMT视电阻率和相位响应偏高，且频率越高、覆盖层电阻率越大，响应偏差越大；对于煤炭采空区模型，RMT法能有效反映异常体位置，但忽略位移电流会在采空区及其附近引起较大计算误差；起伏地形的算例表明地形会覆盖地下异常体的RMT数值响应，尤其是地形拐角处；2种不同规模的并行算例对比证明了并行算法的高效性，且随着求解问题规模增大，并行算法效率也随之...

Forward modeling of radio-magnetotelluric method based on Julia acceleration and study on effect of displacement current

Radio-magnetotelluric method (RMT) is one of the important tools for shallow surface electromagnetic exploration. Since the exploration frequency band is 10?300 kHz, the propagation of electromagnetic fields is greatly affected by the dielectric permittivity of the earth. The electromagnetic response calculated with traditional quasi-static conditions severely restricts the accuracy of RMT forward modeling and further affects the resolution of the inversion imaging. In response to this problem, a numerical simulation method of full-current RMT electromagnetic response based on Julia parallel acceleration was proposed. Then, the calculation of each frequency point was sent through the distributed computing in Julia to different processes for solving, so as to achieve the purpose of accelerating the calculation. At the same time, the influence of displacement current was considered in the calculation to improve the accuracy of forward modeling. Besides, the influence law of displacement current on the apparent resistivity and phase response of electromagnetic field in the radio frequency band was analyzed and summarized by calculating the RMT response of several typical high-resistance/high-dielectric models. The numerical simulation results show that: The calculated RMT apparent resistivity and phase response under the quasi-static conditions were relatively high in case that a high-resistance overburden is present in the shallow surface. Besides, the higher the frequency, the greater the resistivity of the overburden, and the larger the response deviation is. For the coal goaf model, the RMT method can effectively reflect the location of the abnormal body, but ignoring the displacement currents may cause a large calculation error in the goaf and its vicinity. As shown in the calculation example of rugged terrains, the RMT numerical response of underground anomalous body may be covered by the terrain, especially at the corner of terrain. Further, the parallel numerical examples at two different scales comparatively demonstrate the efficiency of the parallel algorithm herein, and the efficiency of the parallel algorithm is improved with the increasing scale of the problem to be solved. Our research improved the computational efficiency and accuracy of RMT forward modeling, laying the foundation for the realization of subsequent rapid inversion algorithm. In addition, the numerical calculation examples also indicate that the RMT method has a good application prospect in the actual exploration of coal goaf.