A new model of synthetic integration for meteorological forecast based on neural networks and fuzzy logic

Combination methods of neural networks and fuzzy logic are briefly surveyed. Then, a novel combination model is presented for synthetic integration of rainfall. The presented model is composed of four network layers: input layer, membership function construction layer, inference layer and defuzzification layer. The combination model is applied to synthetic integration of forecasted rainfall data produced by gradual regression method, periodic analysis plus multi-layer method and model output statistics method. The model is trained by short-term rainfall data of Zhejiang Province from 1980 to 1997. The synthetic integration (forecast) results from 1998 to 2000 show that the presented model can obtain satisfactory forecast performance.