Data Mining for Typhoon Image Collection

This paper introduces the application of data mining methods to the analysis and prediction of the typhoon. The testbed for this research is the typhoon image collection that we established, now archiving approximately 34,000 typhoon images created from satellite images of geostationary meteorological satellite GMS-5. We claim that this data collection is a medium-sized, well-controlled, and richly-variational scientific database that are suitable as a testbed for data mining research. The main challenges of this paper is twofold: the analysis and the prediction of the typhoon. For the analysis, we apply various methods such as principal component analysis and self-organizing map to characterize or visualize the statistical properties of typhoon cloud patterns. We then apply an instance-based learning method for analogy-based prediction using past similar patterns, and also established similarity-based image retrieval system of the typhoon image collection. However, we also point out that we should not overlook fundamental difficulty in typhoon prediction from past similar patterns due to the chaotic nature of the atmosphere.


Asanobu KITAMOTO, "Data Mining for Typhoon Image Collection", Proceedings of the 2nd International Workshop on Multimedia Data Mining, pp. 68-77, 2001年8月 (in English)

BibTeX フォーマット

      @InProceedings{ k:mdm01,
      author = {Asanobu KITAMOTO},
      title = {Data Mining for Typhoon Image Collection},
      booktitle = {Proceedings of the 2nd International Workshop on Multimedia Data Mining},
      pages = {68-77},
      year = 2001,
      month = 8,
      doi = {},
      note = { (in English)},



| リンク 1 | リンク 2 |