基于FCM聚类算法的二维RMT电阻率与介电常数联合反演

Joint inversion of resistivity and permittivity for two dimensional RMT data based on FCM clustering

射频大地电磁(RMT)是新兴的浅地表勘探法,其工作频段(10 kHz~1 MHz)处于音频大地电磁(AMT)与探地雷达(GPR)之间,随频率的增加,位移电流对RMT观测的影响逐渐增大,单一电阻率参数反演会影响数据解释的准确性.本文在同时考虑传导电流与位移电流影响的条件下,开展二维RMT数据电阻率和介电常数联合反演研究.为提高电阻率与介电常数之间的联合约束效果,在经典最小结构模型正则化约束的基础上,引入模糊C均值(FCM)聚类算法进行联合约束.模型试算表明,基于FCM聚类的二维RMT电阻率与介电常数联合反演可以在丰富解释参数的同时提高反演效果.

Radio-magnetotelluric(RMT)is a new technique for shallow geologic survey.The frequency range of RMT is from 10 kHz to 1MHz,between the highest frequency of audio-magnetotelluric(AMT)and the lowest frequency of conventional ground penetrating radar(GPR).The effect of displacement current on RMT data increases with frequency and is related to electric permittivity.Therefore,the quality of RMT inversion model is significantly reduced with a high frequency when only considering conductivity in the inversion.A novel RMT inversion method which considers both conduction and displacement currents and recovers resistivity and permittivity properties simultaneously is proposed in this paper.To improve the mutual constraints between resistivity and permittivity,a fuzzy C-means(FCM)clustering constraint term is added to the classic minimum structure regularization function.The synthetic examples show the recovered resistivity and permittivity models are closer to the true model parameters after including fuzzy C-means clustering constraint term.The new method not only improves the effectiveness of RMT inversion but also recovers additional physical parameters which carry valuable subsurface geologic information.