EXPLORATION AND EVALUATION OF SHORT-TERM RAINFALL FORECASTING BY FRICS DATA

Rainfall is the most important and direct agent that causes flood disaster. It is therefore the flood forecasting essentially depends on forecasting of rainfall. With radar rainfall data remotely observed by the Foundation of River & Basin Integrated Communications, i. e. FRICS, a new methodology is developed for accomplishing short-term rainfall forecasting, wherein two main components are included : (1) settlement of rainfall vector movement with modified correlation method and Fuzzy rule, and (2) determination of spatial and temporal distribution of rainfall intensity using neural network (NN) approach. Reasonable results have been derived with a high accuracy through rainfall data on a real time basis.