<p>This paper attempts to realize the identification and classification of slope failure/landslide patterns in the early warning system (EWS) based on Soil Water Index (SWI), for fuzzy evaluation of the slope failure scale based on meteorological data. For this purpose, the stability analysis and shear strength parametric discussions of a homogeneous slope model composed of two kinds of soil, i.e., volcanic soil and Toyoura sand, are performed under 22 kinds of designed rainfall conditions. In a total of 8,976 simulated slope stability scenarios, 374 slope failures with a factor of safety (FOS) less than 1.0 for the first time were identified. After that, the depths of the potential slip surface of the slope failure patterns were collected and analyzed. Results indicate that the SWI-based EWS can identify and classify the four landslide patterns. As rainfall intensity rises, the slope failure pattern gradually changes from Pattern I (Sliding) during long-term low-intensity (LL) type rainfall, to Pattern II (Buckling), to Pattern III (Toppling), and finally to Pattern IV (Crumbling) during short-term high-intensity (SH) type rainfall. In addition, the correlation between the slope failure pattern and the potential slip depth and SWI is very poor, but there is a strong correlation between the landslide pattern and the potential slip depth and water storage height (<i>H</i>₂) in the second tank layer. Therefore, in the SWI-based EWS, the water storage height (<i>H</i>₂) in the second tank layer might be used to evaluate the scale of slope failure.</p>