Abstract: In light of the accuracy and model construction challenges in county-level meteorological risk early warning for geo-hazards, and considering the prominent characteristics that these geo-hazards mainly occur on the shallow surface of residual slopes, the mechanism of shallow surface slope instability in Guangdong Province was studied through physical simulation experiments and numerical simulations of typical geo-hazards. The results show that the slope is easy to lose stability in the shallow surface layer under the condition of rainstorm, and the main factors are rainfall, rainfall duration, soil type and slope structure. Subsequently, by dividing the study area into slope units, we developed a generalized classification and numerical modeling of these units based on parameters such as slope length, slope gradient, rock and soil type, stratification, and key physical and mechanical parameters of each slope unit, and by combining the Green-Ampt rainfall infiltration model with the infinite slope stability evaluation method, the slope instability dynamics warning model was then constructed. Through the application in the basin of Beiling Town in Longchuan County, and the application of dynamic early-warning technology for slope instability in meteorological risk early warning for geological hazards was preliminarily explored at the scale of slope units, which can provide support for county-level geo-hazards meteorological risk early-warning based on slope unit early-warning in Guangdong Province.