A new method, which is based on Qualitative Rea - soning, for severe rainfall forecasting which occurs on a relatively small scale is presented in this paper. First, the overview of existing numerical weather forecasting methods are shown with their relating scales. Second, qualitative inference of local wind is presented. In this scheme, the partial differential equation is modeled qualitatively and diagnostically. Finally, a qualitative cumulus model is discussed. This cumulus model is designed to simulate cumulus development process with consideration of cloud microphysical processes and is expected to infer the time series variation and distribution of severe rainfall.