An Accurate 3-D CFS-PML Based Crank-Nicolson FDTD Method and Its Applications in Low-Frequency Subsurface Sensing

An unsplit-field and accurate Crank-Nicolson cycle-sweep-uniform finite-difference time-domain (CNCSU-FDTD) method based on the complex-frequency-shifted perfectly matched layer (CFS-PML) is proposed. It is applied to 3-D low-frequency subsurface electromagnetic sensing problems. The presented CNCSU-FDTD takes advantage of both CFS-PML and unconditionally stable CN method so that it can attenuate evanescent waves, eliminate late-time reflections, and overcome the stability limits of the FDTD method. The time step intervals in CNCSU-FDTD can be 1000 times larger than that in the regular FDTD for the low-frequency sensing centered at 25 Hz while remaining accurate. Several 3-D numerical examples in the airborne transient electromagnetics system have been demonstrated to validate the proposed method. The CFS-PML-based CNCSU-FDTD method not only attains good accuracy but also saves several dozen times of CPU time as compared with the regular FDTD method.

Duke Scholars

Author William T. Joines Electrical and Computer Engineering