WEKO3
アイテム
Sequential Data Assimilation and Its Application to Tsunami Analysis in the Japan Sea
https://ir.soken.ac.jp/records/782
https://ir.soken.ac.jp/records/7823b02b13f-1563-464c-a6dc-db492651a53e
| 名前 / ファイル | ライセンス | アクション |
|---|---|---|
要旨・審査要旨 (330.2 kB)
|
||
|
本文 (4.1 MB)
|
| Item type | 学位論文 / Thesis or Dissertation(1) | |||||
|---|---|---|---|---|---|---|
| 公開日 | 2010-02-22 | |||||
| タイトル | ||||||
| タイトル | Sequential Data Assimilation and Its Application to Tsunami Analysis in the Japan Sea | |||||
| タイトル | ||||||
| タイトル | Sequential Data Assimilation and Its Application to Tsunami Analysis in the Japan Sea | |||||
| 言語 | en | |||||
| 言語 | ||||||
| 言語 | eng | |||||
| 資源タイプ | ||||||
| 資源タイプ識別子 | http://purl.org/coar/resource_type/c_46ec | |||||
| 資源タイプ | thesis | |||||
| 著者名 |
中村, 和幸
× 中村, 和幸 |
|||||
| フリガナ |
ナカムラ, カズユキ
× ナカムラ, カズユキ |
|||||
| 著者 |
NAKAMURA, Kazuyuki
× NAKAMURA, Kazuyuki |
|||||
| 学位授与機関 | ||||||
| 学位授与機関名 | 総合研究大学院大学 | |||||
| 学位名 | ||||||
| 学位名 | 博士（学術） | |||||
| 学位記番号 | ||||||
| 内容記述タイプ | Other | |||||
| 内容記述 | 総研大甲第1044号 | |||||
| 研究科 | ||||||
| 値 | 複合科学研究科 | |||||
| 専攻 | ||||||
| 値 | 15 統計科学専攻 | |||||
| 学位授与年月日 | ||||||
| 学位授与年月日 | 2007-03-23 | |||||
| 学位授与年度 | ||||||
| 値 | 2006 | |||||
| 要旨 | ||||||
| 内容記述タイプ | Other | |||||
| 内容記述 | This thesis is organized by two parts. In part I, new data assimilation framework for tsunami simulation models and tide gauge data is introduced. Yamato Rises in the middle of the Japan Sea is analyzed using tide gauge data of Okushiri (Hokkaido-,Nansei-Oki Earthquake) Tsunami occurred in the Japan Sea in 1993. The result of this analysis indicates that the Yamato Rises might be shallower than the existing topography data sets. Many physical variables and nonlinear operations of the simulation model cause difficulty in data assimilation. In part II, methods how to manage high dimensionality and nonlinearity are discussed from the viewpoint of data assimilation. In the earlier sections of part II, the management of nonlinearity is discussed through performance comparison among ensemble based filters used in data assimilation.<br /> In the later sections ofpart II, new smoothing scheme is introduced. Numerical experiments are performed using Nikkei 225 stock price data. Management of high dimensionality is discussed to reduce mainly the memory space required from the large-scale data assimilation.<br /> Past studies about tsunami have been based on making and performing numerical simulation models and comparing observed data with simulation results. Bottom topography data set in these simulations is usually fixed as a priori boundary conditions. However, the data sets have errors which cause inaccurate simulation results because propagation speed of tsunami depends on the sea depth. The accuracy of the topography data is insutficient yet regardless of the recent updates attributed to the ship acoustic measurements and satellite altimeter data analysis. Therefore, bottom topography correction is required to simulate tsunami accurately. In addition, bottom topography correction may improve the knowledges in oceanography and meteorology. <br /> Data assimilation is a technique for a synthesis of information from a dynamic numerical model and observation data especially in geosciences. It is formulated by a state estimation or a parameter identification problem in a state space model. The system and observation models in a state space model correspond to numerical model-based simulations and satellite and/or ground-based measurement systems, respectively. In the new approach, tsunami simulation model and the tide gauge data correspond to the system model and the observation data. Since the observations are sparse in this problem, refinement of parametrization is also introduced. <br /> To validate the method, comparative experiments Called identical twin experiments are performed using two kinds of bathymetries; one is artificial bathymetry and the other is the Japan Sea bottom topography. The results demonstrate that the new approach works well for the purpose of correction of inaccurate bottom topography. Based on this framework, the bottom topography is corrected for the Okushiri Tsunami case in the Japan Sea. The effectiveness of the new approach is confirmed and new findings are obtained about Yamato Rises in the middle of the Japan Sea. The estimated topography for the Yamato Rises is shallower than the average of the available data set over most of the area. Additionally, Nihonkai-Chubu Earthquake Tsunami which occurred in the Japan Sea in 1983 is simulated using the original and corrected bottom topography. Tsunami arrival time prediction is improved under the corrected bottom topography. <br /> In the earlier sections of part ll, performances are compared between the ensemble fiIters and smoothers. Numerical experiments show the superiority of the particle filter to the ensemble Kalman filter with the nonlinear observation system. Nextly, the smooth bootstrap technique is discussed in view of relationship between ensemble filters and genetic algorithms. Numerical expertments are also conducted under two types of simulation models. Based on them, applicability of the smooth bootstrap technique in data assimilation is discussed. <br /> In the latter of part II, a new scheme is demonstrated within the particle smoother. Degeneration problem causes some estimation bias in the particle smoother because resampling from same support is repeated. A straightforward approach to overcome the problem is to increase particles. However, it is difficult because space complexity in the particle smoother is linear order to the time length to be smoothed. The suggested new scheme, called recursive recomputation scheme, can reduce required memory space from linear to logarithmic order maximally. As a result, the number of particles can be increased drastically and length effectively smoothed can be extended. The result of numerical experiment using Nikkei 225 stock index shows desirable aspects of the new scheme in estimation. | |||||
| 所蔵 | ||||||
| 値 | 有 | |||||
