Hailstone Detection Based on Image Mining

Hailstone is one of the main meteorological disasters and it is very difficult to be forecasted effectively. A novel method of automatic recognition hail cloud is proposed base on association rules in this paper. Firstly, the method preprocesses the original reflectivity image. Then the color features and texture features are extracted from the radar echo image. Lastly, the association rules, which are used to distinguish hailstorm from rainstorm and super refract, are constructed, namely hail echo ensemble forecast model based on image mining is constructed. The results show that hail echo ensemble forecast model based on image mining has higher accuracy compared with the PUP products.