In this study, a citywide early warning model capable of predicting landslide and debris-flow disasters was developed, in response to the increasing frequency and intensity of such disasters due to climate change. The model, which considers geo-data with spatial variability in addition to rainfall data with temporal variability as input parameters in real-time, generates and displays a map of five warning levels by applying five thresholds developed using various approaches in a stepwise and overlapping manner through a decision tree. In order to validate the spatial/temporal performance (accuracy and efficiency) of the model, a case study on landslide events in 2009 was modeled for early warning state changes. The model, which was developed by adopting the advantages and overcoming the limitations of existing internal and external early warning systems, is expected to function as an effective and rliable diasaster management tool for decision makers.