Abstract: This paper presents a design of a geo-hazard early warning model. It can analyze the influence factors including terrain, physiognomy, hydrology, etc. It is based on the features of variety, randomicity and variability of geo-hazards and use the map algebra and GIS technologies. Considering the features of influential factors, the principal components are selected to quantitatively identify and measure the influence of each factor. And then the levels of potential hazards are calculated by the data and information collected from the station. On this basis, using the Thiessen interpolation method, the surface data are generated with the levels of hazards according to the site-based hazards information. Based on the daily effective rainfall and intensity records, the occurrence of probability of geo-hazard is calculated. After overlaying the level map of hazards, the surface data with the levels of hazards and with rainfall map, the occurrence of geo-hazard can be predicted based on the weather confidence method and the functionality of geo-hazard early warning model can be realized. The case study in Guangdong province shows that the geo-hazard early warning model can predict the occurrence of geo-hazards with high spatial resolution. The framework itself is logic while the predicted results are reliable. So in some sense, this method could be extended to other case studies. The research results in Guangdong are bearing more significance for the guideline of preventing and controlling the occurrences of geo-hazards.