Abstract: China frequently experiences geological disasters,resulting in substantial losses. Through multiple rounds of geological hazard surveys and inspections,nearly 300 000 potential hazard sites have been identified. However,approximately 80% of catastrophic geological disasters in recent years occurred outside these identified sites. Detection systems based on the"Space-Sky-Surface"(3S)framework can monitor large-scale disasters with obvious deformation and destruction signatures but are less effective at identifying small,latent landslides that exhibit subtle precursory signs,short gestation periods,or high abruptness. Future disaster prevention and mitigation efforts should adopt a"dual-control" strategy that integrates both known hazard points and risk-prone areas(point-surface integration),with the risk slope serving as the basic analytical unit. This study examines the typical characteristics of latent landslides in southeastern hilly regions and the Loess Plateau,summarizing their formation conditions and rainfall-triggering factors. Using refined rainfall forecasting techniques,spatiotemporal prediction of such landslides is conducted. A rapid regional-scale slope stability evaluation model is then established based on a high-resolution three-dimensional geological model. Through a"front-end-cloud" platform architecture,an intelligent latent-landslide identification model,a rainfall forecasting model,and a stability evaluation model are integrated into a unified software platform—Earth Survey. This enables multi-source information fusion and dynamic disaster assessment for latent landslides across large-scale slope surveys. The proposed research framework and the Earth Survey platform are applicable to rainfall-induced geological disaster-prone regions across eastern,central,and western China,providing theoretical foundations and technical support for the identification and assessment of latent landslides.