| 摘要: | 摘要 傳統電離層斷層掃描結果之解析度受限於掃描區間之離散網格尺寸，當在電波路徑數目有限之情況下，利用縮小網格尺寸(亦即增加掃描區間之網格數)提升結果解析度之方法會產生人為假象(artifact)。本論文提出局部高密度網格斷層掃描演算法(algorithm)，包含代數演算法(algebraic reconstruction techniques, ART)、倍數代數演算法(multiplicative algebraic reconstruction techniques, MART)、以及同步迭代演算法(simultaneous iterations reconstruction technique, SIRT)修正此問題。局部高密度網格之離散方式是將掃描區間內電子濃度變化較劇烈之區域使用較小之網格，此方式不會大幅增加整個掃描區間之網格數，因此可避免路徑數與網格數差異過大之問題，可有效提升結果之解析度且無人為假象之產生。所建立之演算法經由US-TEC資料庫之GPS衛星資料，進行事件之分析與驗證。驗證上採用觀測資料比對。分析結果發現電離層斷層掃描技術可利用局部高密度網格之方式大幅改善反演結果之解析度，可有效監測電離層中存在之不規則體，例如Sporadic E或是電漿泡狀結構等。 Abstract Resolution of the traditional ionospheric tomography is limited to the pixel size for discretizing the domain of interest. Specifically, with limited number of propagation paths between the ground stations and the satellite orbit, the reduction of the pixel size for improvement on the resolution often leads to artifacts in the results retrieved from the standard algorithms, such as the algebraic reconstruction technique (ART), the multiplicative algebraic reconstruction technique (MART), and the simultaneous iteration reconstruction technique (SIRT). This issue is mainly related to the ill-posed nature of the system equations from the ionospheric tomography. In this thesis, a local pixel-refinement technique is proposed for resolving this issue. The proposed technique exploits an inhomogeneous domain discretization. That is, in the domain of interest, the regions with high density gradient are dicretized using fine pixels, whereas the ones with low density gradient are dicretized using coarse pixels. With this discretiztion technique, the number of pixel is relatively reduced compared to the one form entire-domain fine discretization. With this technique, it is shown that the resolution of the reconstruction results is improved, and there is no observation of the aforementioned artifacts. The proposed technique is validated through the US-TEC data. Several events are used to demonstrate the effectiveness of the proposed technique. Indeed, with the proposed technique, the ionospheric irregularities, such as sporadic E layers and plasma bubbles, are identified in the reconstruction data using the standard ART, MART, and SIRT tomography algorithms. Indeed, the classical tomography algorithms with the proposed discretization technique are shown the effective tools for monitoring the ionospheric fine structures. |
