Abstract. Jeju Island, located in the southern sea of the Korean Peninsula, is a volcanic island created by the tertiary and quaternary volcanic eruptions. A group of lava tubes formed between quaternary lava flows constitutes one of the most predominant geological contexts owing to its unique and complex network, for which a total of 178 lava tubes is estimated. As a significant portion of lava caves have not been discovered, the threat caused by lava cave collapse has become one of the major concerns in connection with the recent infrastructure construction in Jeju Island. Considering the risk potential, the overall distribution and collapsing risk of the Jeju lava tube network were investigated in this study. Through spatial analysis, we firstly found that the lava tubes distribution is not correlated with specific geological units. Secondly, the risk of collapse is high especially when there are ongoing artificial constructions around the undisclosed lava tube network. We therefore introduced Interferometric Synthetic Aperture Radar (InSAR) techniques to measure the deformation of the ground surface where lava tube networks distributed underground. InSAR results and the proposed machine learning applications identified that the populations of ground deformations was up to 1–2 mm/year and was inferred to be caused by the instability of the shallow lava cavity. Given that underground cavities could pose serious risks, a detailed physical exploration and threat assessment of potential cave groups is required before intensive anthropogenic construction is developed.