Research And Implementation Of Data Mining Application In Meteorology

With the rapid development of modern meteorological concept explorationtechnology, the meteorological departments at all levels can acquire more and moremeteorological data that supplies them with important information. In traditionalapplication system of meteorological database, it is difficult to find relations and rulesamong meteorological data, as well as methods to acquire information behind it, thusthere appear the explosion of data and the limitation of knowledge. It is important formeteorologists to acquire large amount of data, analyze them, and generatemeteorological data closely relevant to weather forecast, which would be helpful tometeorological operations and decision-makings.Based on the application of data mining technology in meteorology, the latestdevelopments in data mining technology were analyzed，and a model about an in-depthstudy of meteorological data mining, data preprocessing, association rule mining anddecision tree generation algorithm were carried out.The main work of this thesis accomplished is as follows:1. Based on the existed data mining process model and combined with processesand business needs, a meteorological data mining process model is proposed.2. Based on the existed problems in meteorological data storage and processing,the common data pre-processing technology and application are summarized.3. A variety of association rules in mining algorithm have been designed andcarried out, where Apriori algorithm and CP algorithm were focused on, and hashtechnology for meteorological data characteristics, clustering algorithms to divide theinterval discrete continuous data and introduced interesting conception in methods foralgorithm improvement were studied.4. Decision tree to generate subsystem has been studied, where the choice ofattributes, the cut of branches and the access to rules have been designed and carriedout, and the dichotomy, merging mapping and clustering were used to processcontinuous meteorological data.The test of Weather Forecast Center showed that the system worked well. The database can well support weather forecasting, help to analyze weather causes andstatistics in weather forecasts screening predictor and to build a knowledge base.Therefore, it can improve the accuracy of weather forecasting, upgrade meteorologicaldata from simple low-level application to advanced application of knowledge miningand decision support.