Study On Meteorological Data Organization Method And Spatial-temporal Differentiation Characteristics Extraction Method Based On The Hadoop

Temperature factor is a hot magnet in meteorological research. A deeply understanding on regularity of air temperature can significantly improve the precision of the meteorological forecast, which will have a profound influence on guiding agricultural production and ecological management. With the continuous improvement on the level of information in meteorological service, related research and management activities acquired and accumulated a lot of meteorological information. However, due to the characteristics of meteorological information data, such as complex categories, totally different data formats, decentralized management and etc., the development of public meteorological information service have been greatly hindered. Meanwhile, current researches on meteorological data have some weaknesses, for instance, a low processing efficiency of processing massive meteorological data and a low level of multiple research technology integration. As a result, the jrequirements of the timely meteorological services with rich information cannot be met well. In addition, it remains a lack of studies on characteristics of temperature spatial-temporal differentiation in the perspective of large scale and multiple dimensions. Nonetheless, Hadoop, an open-source cloud computing framework, still provides a new approach for the study of massive meteorological data.With such a background, this dissertation firstly introduces the cloud computing, geographic data analyzing and processing methods based on empirical meteorological studies. Then, it designs the data organization method and extracts characteristics of the daily air temperature differentiation characteristics with the techniques of Hadoop and GIS, also taking the regional effects on the temperature into consideration, sourcing hourly observed data of global surface air temperature in the period of 2009 to 2013 from the U.S. National Climatic Data Center (NCDC). The main contents and outcomes of this dissertation are as follows:(1) This dissertation for the characteristics of the temperature observed log data, to study the organization method of massive spatio-temporal meteorological data based on HBase. Therefore, this paper researches and designs the table structure, indexing system and data storage method of the meteorological data. An improvement plan for the operational parameters of HBase is also proposed. In addition, the problem that HBase does not support the secondary auxiliary index has been solved by introducing the index system of Solr distributed search engine.(2) This dissertation studies the extraction method of daily temperature spatio-temporal differentiation characteristics with the platform of Hadoop and GIS. Based on the MapReduce programming model, this dissertation researches and designs the method of temperature factor extraction and statistic, and distributed algorithm for the temperature diurnal variation curve also has been designed. Then, on this basis, the dissertation researches and designs the influencing factors extraction method and computing method of the temperature diurnal variation(3) The author has carried out simulating experiment to verify the efficiency and accuracy of the proposed meteorological data organization method and daily air temperature spatio-temporal differentiation characteristics extraction method and the results are positive.