Abstract: The different distributions of stratum layer have an important influence on the propagation characteristics of electromagnetic wave through the earth communication. Kriging interpolation method utilizes the finite sampling points of the stratum to establish the layered medium model effectively,in which the parameter optimization of variation func- tion model is the key to the accuracy of the layered medium model. To solve the error problem between the interpola- tion points and the sampling points which is caused by the unevenly distributed sampling points,and make the interpo- lation points more close to the sampling points for establishing a layered stratum model more accurately,a Kriging in- terpolation method based on particle swarm optimization-genetic algorithm ( PSO-GA) jointly is proposed to optimize the parameters of variation function model. Compared to Kriging interpolation method based on the PSO or the GA,the proposed algorithm not only solves the problem of falling into the local optimum and achieves a global optimal sample but also speeds up the convergence through the information exchange among particles twice during the iteration search- ing. Simulation and analysis results show that a better searching performance and optimizing stability are achieved with the proposed algorithm for establishing a layered stratum model for though-the-earth communication than that with Kriging interpolation method based on the PSO or the GA.