Big Data Analysis Using Deep Learning on Extreme Weather Events such as Typhoons
Project/Area Number |
16K12466
|
Research Category |
Grant-in-Aid for Challenging Exploratory Research
|
Allocation Type | Multi-year Fund |
Research Field |
Perceptual information processing
|
Research Institution | National Institute of Informatics |
Principal Investigator |
Asanobu Kitamoto 国立情報学研究所, コンテンツ科学研究系, 准教授 (00300707)
|
Co-Investigator(Renkei-kenkyūsha) |
FUDEYASU Hironori 横浜国立大学, 教育学部, 准教授 (00435843)
|
Project Period (FY) |
2016-04-01 – 2018-03-31
|
Project Status |
Completed (Fiscal Year 2017)
|
Budget Amount *help |
¥3,380,000 (Direct Cost: ¥2,600,000、Indirect Cost: ¥780,000)
Fiscal Year 2017: ¥1,170,000 (Direct Cost: ¥900,000、Indirect Cost: ¥270,000)
Fiscal Year 2016: ¥2,210,000 (Direct Cost: ¥1,700,000、Indirect Cost: ¥510,000)
|
Keywords | 台風 / 深層学習 / ディープラーニング / 気象衛星画像 / パターン認識 / ドボラック法 / 温帯低気圧化 / 時系列モデル / 画像情報処理 / 機械学習 / 気象情報 / 衛星画像 / データセット |
Outline of Final Research Achievements |
Typhoons are important events both in terms of meteorological research and society, but its analysis on intensity and structure has been dependent on manual analysis by human experts. Hence this research proposed a new method for analyzing typhoons from the viewpoint of big data through the creation of the large-scale dataset of meteorological satellite images on typhoons and the application of machine learning, or deep learning. We tackled four topics, namely "classification of typhoon grades," "regression of typhoon central pressure," "transition from typhoons to extra-tropical cyclones" and "extension to a temporal model." In particular, we obtained interesting results on "transition from typhoons to extra-tropical cyclone," in which we proposed a new deep learning-based index called "extra-tropical transition index." Comparison between the index and JMA best track revealed that transition timing announced from JMA is about a half day later than deep learning-based index.
|
Report
(3 results)
Research Products
(8 results)