Analysis And Mining For The Relationship Between Satellite Nephograms And Rainfall

Satellite nephograms are observational images of earth atmosphere, which are obtained through remote-sensing and telemeter by meteorological satellites. Information directly reflected by nephograms can be classified into two types: one is temperature information of clouds on top, the other is the distribution information of cloud clusters. There are some advantages to utilize nephograms to analyze rainfall conditions, such as good intuitiveness, control easiness, fast speed and wide range. Nephograms has already become an important observed factor for the decision-making process of preventing flood and fighting a drought.At present, the research of relation between nephograms and rainfall mainly relies on expert's experiences and observation of nephograms with the unaided eye. So it still belongs to a qualitative category. This method is in a complete got-up mode, in which some problems exist, including excessive subjectivity, lack of quantitative analysis and low precision. Therefore the result of analysis is always short of stringency in the decision-making process.Image mining technique is a front of Data Mining. As an effective tool of analysis, image mining can not only dig out knowledge from simple image sets, but also discover knowledge from related non-image data sets. This paper takes about 2000 nephograms of Hunan area from April to August in 1996 and hyetal records from 262 ground hyetal stations as training data. Through image mining analysis on them, we acquire valuable knowledge. The knowledge has quantitative explanations, and has passed meteorologists' assess. Therefore, it is of great significance to research on relation between nephograms and rainfall.In detail, the content and achievement of this paper are listed as follows:1 Using present achievement in the field of multimedia research and combining ripe mining technology, a new method used for researching connection between satellite nephogram and rainfall, utilizing image mining technology, is presented in this paper. This paper use the first strategy to preprocess data of satellite nephogram so as to represent image data by a new means and improve several mining cs, and ultimately realize the task of image mining.2 Mining of association regulation based on nephograms. A great number of satellite nephograms and abundant record of rain quantity present original material for mining of association regulation. Nephogram belongs to image data and record of rain quantity belong to character data. It is the first problem to deal with that how to choose a appropriate organization form of data to combine two kings of data. So we introduce concept of multidimensional data cube, image character of nephogram, data of rainfall and so on, as component of data cube.The other problem is greatness of image data. Apriori algorithm, a classical algorithm, don't meet the need of efficiency. In order to deal with the problem, a new concept of transaction is introduced. Basedon the concept, Apriori algorithm can be improved so to meet the need of efficiency.3 The clustering analysis of satellite nephogram. The first problem that nephogram clustering faces is how to construct eigenvector. Nephogram grey information, texture and so on can be regard as one dimension of eigenvector. But if the dimension of eigenvector is so high, it will aggravate burden of clustering algorithm, so it must chose the most representative character to construct eigenvector. At present, there are a great kinds of clustering algorithms. So the second problem that nephogram clustering faces is how to select a effective clustering algorithm.Based on comprehensive understanding of all kind of algorithms and combining self character of satellite nephogram, this paper select Self Organized Map network as clustering algorithm. This method is valid via result of clustering analysis.Full text is based on application background of study of relation between satellite nephogram and rainfall. Mining result has reference merit to meteorologists. Although there has some faultiness, this paper has it's ow...