根据广西低温雨雪冰冻过程各站逐日的降水、温度情况构建代表过程强弱的冷湿指数，利用NCEP/NCAR再分析资料和ECMWF、T639等数值预报产品，识别影响广西低温雨雪冰冻过程的主要天气系统和影响因子，并采用信息熵计算方法与粗集理论相结合的属性分析研究技术，构建广西低温雨雪冰冻过程的模糊神经网络集合预报模型。进一步利用蚁群算法和非线性降维技术对模型输入矩阵的属性差异进行分析，开展集合预报个体差异性预报试验研究，分别建立一般性和严重持续性的低温雨雪冰冻过程的统计集合预报模型，提高分类预报模型的确定性和稳定性，为利用数值预报产品进行低温雨雪冰冻过程预报提供一种新的思路。

A wet-cold index will be developed based on the daily precipitation and temperature datasets about freezing rain and snowstorms over Guangxi. Dominant weather systems influencing the freezing rain and snowstorms will be identified based on the NCEP/NCAR reanalysis data and Europe center datasets and T639 numerical forecast products. A Fuzzy Neural Network Ensemble Forecast Model of freezing rain and snowstorm over Guangxi will be constructedusing information entropy calculation method with the combination of rough set theory attribution analysis technology. In order to improve the classification of the certainty and stability of the forecast model, the difference of model input matrix attributions will be analyzed using ant colony algorithm and nonlinear dimension reduction techniques. The individual differences of ensemble forecast experiments will be developed in this project. Based on all the above studies, a statistical ensemble forecast model which can be applied in both general-degree and severe-degree of the freezing rain and snowstorms process over Guangxi will be constructed.