Study On The Construction Of Satellite Nephogram Feature System For Severe Convective Weather

Severe convictive weather is one of the most dangerous disaster climates, which heavily affect routine work and daily life. Severe convective weather is a small-scale weather system with small spatial-scale and a short-time lifecycle. So a good many attentions is paid to the severe convictive weather forecasting. There is a lot of valuable information in Weather satellite nephograms, and the area, gray value, texture and contour features of the cloud can tell us a lot of things about the severe convictive weather. So with the development of technology of the pattern recognition and spatial database, it is a useful to forecast the severe convective weather by the mean of the satellite nephograms.In this paper, the difference of severe convective cloud and non-severe convective cloud are highly regarded, and the clouds of east Gansu in July are the point of our experiment. In order to describe and quantify these differences, we extract eigenvector from the severe convective cloud and construct the feature system of time series satellite nephograms by making use of technology of pattern recognition and spatial cloud database.Firstly, the use of mathematical morphology and pattern recognition methods for pretreatment, feature extraction and classification on the FY-2 satellite nephograms help us extract the eigenvector effectively.Secondly, several static features are extracted according to the characteristics of severe convective cloud, such as textural, area, average gray value, and centroid and so on. Also, the temporal variation features of each feature are also extracted. Additional, the result of the experiment which is based on the eigenvector group and classifier of Support Vector Machine (SVM) proved the efficiency of the method.Finally, having study the recognition of the severe convective cloud, a feature system on the severe convective weather forecasting is put forward for the first time, which can improve the veracity of weather forecasting.In one word, a feature system based on the extracted eigenvector, both static and dynamic, and their temporal variation features are constructed, which provides the foundation for the recognition model, the satellite image database and the construction of the severe convective weather forecasting.