Abstract: On the eve of the landslide disaster, bursting, friction and fracture of rock and soil mass in the sliding body will cause low frequency infrasound signals to propagate outward. Advance positioning based on infrasound signals has become a key technology for landslide monitoring and early warning. In this paper, the wave equation was transformed into Helmholtz equation by Fourier transform, and the Kernel function of the wave equation was obtained by Hankel transform. Based on this, we established the normal wave geological sound field model, and simulated the reflection-refraction relationship between the air, soil and rock interface. By matching the measured infrasound signals of the vertical receiving array with the copied infrasound signals of the propagation model, we proposed the advanced matching field location algorithm for infrasound signals of landslide, which searches for the coordinates at the maximum correlation coefficient as the predicted sound source. We compared the performance of three matching field algorithms and found that the variable-coefficient-likelihood matching field algorithm can be applied to landslide positioning technology due to its small sidelobe and accurate positioning, and found that the 12-element vertical array in the soil layer could be accurately positioned, but the error in the air was large. Finally, we applied the algorithm to the advanced positioning of landslides, and obtained the local failure position and predicted the slip line of landslide. This algorithm provides a new idea for the landslide advanced prediction and positioning technology.

Key words: advanced positioning of landslides, infrasound signal, normal wave model, matching field process