The Application Of Time Series Data Mining In Glacier Mass Balance Prediction

Data mining is the complex process of extracting and digging out the unknown and valuable mode or the rules from the large amounts of data. Time series data mining is one of the popular researches in the field of data mining. Because the time series data always has high dimensions, knowledge discovery based on multidimensional time series data is a especially hot topic. At present, much progress has been made in many aspects of time series data mining such as dimension-reduction, similarity measurement, periodic pattern discovery,feature representation and so on, at the same time, the research of time series data mining has promoted the development of finance, hydrology, stock, life sciences and others. In the field of glaciers, Glacier scientists encounter tremendous difficulties and challenges when they studies the Glacier Mass Balance. The traditional method, according to altitude gradient, is to study material balance by digging holes from the surface of glacier and inserting slider into it,but this method is very difficult to implement. Therefore, this thesis puts forward to study glacier mass balance by using data mining technology. At present, the research on geography is mainly focused on time series data about the hydrology similarity search, weather forecast, space remote sensing.This thesis, according to the theory of time series data mining and current research status,puts forward a prediction model based on the Glacier Mass Balance prediction model, which using the climate data collected by meteorological automatic observation station on the Jade Dragon Snow Mountain glacier as the research object. Using method of fitting key points of extreme slope and linear feature to process the glacier data and extract the feature data and analyzing the trend of factors related time series data to compress data; The method of correlation removal proposed in this thesis is used to reduce dimensions; The cluster analysis of data sequences is using the current relatively mature clustering algorithm, that is K-means;According to clustering results, we can find out the rules and catch useful knowledge and combine practical observation to verify the result.The collected climate data is devoted to data mining, and the mountain precipitation rule model and the material balance prediction model are obtained respectively. According to the data of precipitation and temperature collected by weather station under the mountain, the data of precipitation and temperature in the 4800 meters above sea level is recovered. The mountain precipitation rule model is used to explain the relationship between precipitation in the glacier region and related climate elements. The model indicates that precipitation and temperature have vital impacts on glacier retreat changes, and temperature plays a leading role in the change of glacier. Finally, we predict the glacier mass balance through the changeof temperature in recent decades, this thesis provides theoretical basis for sustainable tourism development in the local tourism management department.Using the model to analyze the change of material balance index, the material balance is found to be in a negative state in recent decades, that is, glacier ablation volume is greater than the amount of accumulation, which is consistent with the results of glacier experts in Jade Dragon Snow Mountain glacier environmental observation station of Cold and Arid Regions Environmental and Engineering Research Institute, Chinese Academy of Sciences.Therefore, we can use this method to predict the future trend of glacier mass balance. The application of data mining in the material balance is still in its infancy, and the corresponding methods and technologies are not very mature. In the future, the method and technology of data mining on the Glacier Mass Balance prediction should be further studied.