Abstract: There is still a theoretical defect in interpretation and judgment of karst morphology and filling state in karst cave detection using electromagnetic wave computed tomography. In order to solve practical problems in the detection, on the basis of finite-different time-domain(FDTD) algorithm, a numerical CT detection model is established using Matlab. A perfectly matched layer(PML) and a perfect electric conductor(PEC) are set at the boundary of computation model to prevent electromagnetic wave getting into the model again after it passed through the boundary. Simulation and experiment results show that: (1) When the electromagnetic wave passes through karst cave, it refracts at the cave boundary. And within the cave, the phenomenon of reflection, diffraction, and obstacle gain occur. Energy attenuation of electromagnetic wave is mainly caused by refraction. (2) When the cave filled with water, the reflection makes a part of the electromagnetic wave remains in the cave inside, this part of wave reflects and diffracts repeatedly inside the cave. (3) After electromagnetic wave passing through the cave, the disturbance of the waveform is mainly caused by the reflection phenomenon and the electromagnetic wave trapped inside the cave. (4) We have to consider the influence of the filling type, water content and other factors when the electromagnetic wave absorption coefficient changes in karst caves. At last，the tomography data obtained in practical detection are judged and interpreted according to the simulation results. By comparing the interpretation of tomography with previous investigation data, the accuracy of the simulation results and the reliability of the application of electromagnetic wave CT technology in detecting karst caves are validated. This research provides a reference and a theoretical basis for interpreting and judging tomography results in the future.

Key words: karst cave, electromagnetic wave computed tomography, FDTD(finite-different time-domain) algorithm, absorption coefficient