摘要
根据日本地球静止气象卫星（ＧＭＳ－５）云图的新特点，运用动态分类方法对ＧＭＳ－５四通道卫星云图进行分类，得到各种云类及地表。并由分类结果，根据一维云模式得到的对流云对流核心云顶温度与降水之间的关系，对层云和对流云做定量降水估算。并用１９９５年８月３１日的云图资料进行对流云和层云的降水估计试验，将估算出的降水率和降水面积与地面１ｈ的观测降水资料进行比较，结果表明：假如设置４０％为降水的允许误差，那么降水估计的准确覆盖率将达到７０％。能在业务应用中推广，并且该方法可以应用到即将发射的风云２号气象卫星资料处理中。
This paper studies the cloud classification based on the spectral characteristics of the Geostationary Meteorological Satellite (GMS-5) observations at both day and night time. The difference between 115 μm window and 67 μm water vapor channels replace the visible one in the nighttime cloud classification With such a “new channel”, the accuracy of the classification is very close to that of daytime one with both infrared and visible channels, which shows that the water vapor channel plays an important role in the nighttime cloud classification when only infrared channels are available. Then the convective and stratiform rainfall estimation from classified cloud image was made using a one-dimensional cloud model. The statistically accurate rainfall estimation from one case reaches 70%, which shows the potential application of this method in the operational use of GMS-5 and FY-2 data.
出处
《大气科学》
CSCD
北大核心
1998年第3期371-378,共8页
Chinese Journal of Atmospheric Sciences
基金
国家"八五"攻关课题
关键词
卫星云图
自动分类
降水估计
定量估算
satellite cloud image classification rainfall estimation
