Study On Application Of Spatio-Temporal Data Mining In The Enviroment Protection

In recent years, the environment dymatic monitoring systems had collected voluminous data. The traditional data analysis tools could hardly analyze these data. To detect the knowledge in the vast data is a important tast of spatio-temporal data mining, and is very significent for a environment-protection worker to make environment plans and other decision.Data Mining (DM) is a procedure of extracting implicit knowledge hiding in the large databases, which includs association rules detection, clustering, classification and so on. Spatio-Temporal Data Mining (STDM) is to extract knowledge from spatial-temporal Data, which contains spatial-temporal association analysis, spatial-temporal clustering, spatial-temporal predictection etc.The work of this thesis is to employ theory of STDM in the environment field and to mine the pertinent knowledge about the air pollution. Overviewing achievements of STDM, theory, process and technique frame of STDM are expounded. Since the traditional environment statistic analysis methods can only analysis relationship of environment varying and infection factors, which dosn't intuitively express the pollution diversty, this paper uses a classical algorithm to mine association rules between air pollutants and meteorological factors, and quantificationally expresses their relationship.In order to finding relationship between the spatial distribution of pollution sources and air pollutants indices, a Voronoi-based spatial association rules mining methods is proposed. First, Voronoi diagram is created according to monitor stations, which discretes the study regions. And then, a spatial transaction database is build depending on the spatial prediction. In the spatial transaction database, the classical alogorithm is carried out to extract spatial association rules between air quality and pollution sources.A spatio-temporal event (STE for short) based spatio-temporal association rules mining method is studied. According to the spatio-temporal domain of STE, the research spatio-temporal field divided into lots of spatio-temporal transaction cells. Next, depending on spatio-temporal transaction cells, the spatio-temporal transaction database is build utilizing time and spatial prediction words. Finally, association rules between air pollution and spatio-temporal events are mined in the transaction database.In the last charter, the thesis concludes the contributions and somelimitation, and points out that in the representation and evaluation ofspatio-temporal association rules are the future works.