基于不确定性理论的降水天气分类研究

首先介绍了云模型这种新兴的处理不确定性问题的工具,提出了一种基于云模型的层次聚类方法.该方法将原始数据集划分为许多小簇,然后分别用云模型来表示这些簇,再通过云综合的方法对这些小簇进行逐层合并,实现对数据集的聚类.将该方法应用于2006-2007年中国及周边亚太地区FY-2C卫星云图的聚类分析,得出不同降水类型的云图特征.最后以2006年7月6日FY-2C中国分区图所示地区对应的FY-2C卫星云图进行降水类型分类为实例,得到不同降水天气对应的云图辐射亮度值特征,并利用分类结果对实况云图进行降水天气的判别,结果表明该方法对大暴雨天气的判别效果良好.

On the precipitation classifications based on uncertainty theory

A new-Dome uncertainty theory named "Cloud Model" was introduced in this paper, to solve the problems concerning uncertainty issues. A hierarchy clustering method based on cloud model theory was proposed, which divided the data sets into a number of different small dusters, denoted by corresponding cloud model. Combining the dusters level by level through a synergistic method, the objective of the data sets clustering had been achieved. Having recourse to the cloud model, baaed on the large quantitative of FY-2C images, consisting of five wavelength of spectral data, which were acquired during the period 2006 to 2007 over Asia-Pacific area, the main radiance features of different rain types such as rainstorm, light rain. can be determined in advance. Taking the FY-2C image acquired on 6 June 2007 as an example, we got the corresponding radiance features consistent with a variety of rainfall types, then with a prior knowledge of rainfall from above mentioned data and methods, different rainfall types were determined accordingly. By validating the classification of rainfall types against the rain data from the ground-based meteorological stations, it is obvious that the determination of rainstorm is more precise than other ones, demonstrating as well that the cloud model can be potentially used widely in the meteorological fields.