結合永久散射體雷達差分干涉法與全球衛星定位系統計算地表三維變形;Decomposing Three-Dimensional Land Motion from A Synergy of PSInSAR and GNSS Stations

NCU Institutional Repository > 工學院 > 土木工程研究所 > 博碩士論文 > Item 987654321/83046

請使用永久網址來引用或連結此文件: http://ir.lib.ncu.edu.tw/handle/987654321/83046

題名:	結合永久散射體雷達差分干涉法與全球衛星定位系統計算地表三維變形;Decomposing Three-Dimensional Land Motion from A Synergy of PSInSAR and GNSS Stations
作者:	簡留玄;Jian, Liu-Xuan
貢獻者:	土木工程學系
關鍵詞:	永久散射體雷達差分干涉;GNSS連續站;三維拆解;地表垂直運動;地層下陷;PSInSAR;GNSS;Decomposition;Surface vertical deformation;land subsidence
日期:	2020-07-31
上傳時間:	2020-09-02 14:32:03 (UTC+8)
出版者:	國立中央大學
摘要:	雷達差分干涉技術(Differential Synthetic Aperture Radar Interferometry, DInSAR)之高時間、空間解析度特性有助於地表變形監測，依據不同的地表性質、大氣干擾程度、影像品質、解纏錯誤等因素，其精度最佳可達公分等級。永久散射體雷達差分干涉法(Persistent Scatterer InSAR, PSInSAR)則藉由挑選在多組干涉對中長期穩定之像素進行差分干涉，降低同調性低的像素干擾、減少解纏錯誤，以提供高精度變形速率，其精度則可達公厘等級，進行地表微變分析。然而，雷達衛星觀測之變形為視衛星(Line-of-sight, LOS)方向之一維變形，地表水平及垂直方向運動皆產生貢獻。此外，利用Sentinel-1影像分析全台尺度之大範圍地表形變時，容易受到解纏錯誤累積的影響。因此，此研究利用永久散射體雷達差分干涉作為基礎，協以GNSS連續站的克利金內插法，做為控制點消除大氣誤差、軌道誤差、解纏錯誤等其他誤差，並利用GNSS水平向觀測值，消除視衛星方向中水平向位移之貢獻，推估垂直向變形量。因GNSS提供每日觀測量，PSInSAR之時間序列皆可進行三維拆解，提供高時間及空間解析之連續時間垂直變形。初步成果利用獨立GNSS垂直觀測量進行交叉驗證(leave-one-out)得到相關係數為0.91，差值均方根為12.56公厘。另外，我們也利用常見基於雷達衛星觀測幾何獲得垂直向變形之方法：以升降兩軌雷達影像進行聯合反演，比較其精度，與GNSS垂直觀測量之相關係數為0.52，差值均方根為39.24公厘。相較之下，利用GNSS進行三維拆解的成果較佳，展現GNSS做為控制點可有效減少誤差，尤其當研究區的範圍橫跨山區。最後利用連續時序垂直向的空間變化分析台灣西部地表變形的時空變化，可看出彰化無固定下陷季節，雲林與嘉義沿海除溼季外皆有明顯下陷、屏東則是於乾季時明顯下陷。;Several studies had applied the Differential Synthetic Aperture Radar Interferometry (DInSAR) and Persistent Scatterer InSAR (PSInSAR) techniques to reveal land deformation time series for event-based or county-scaled analysis in Taiwan. However, for a national-wide analysis from a merged swath of Sentinel-1 Interferometric Wide (IW) mode, some residual errors need to be mitigated during the unwrapping procedure. Therefore, this study aims to use a network of dense GNSS stations as control points in Taiwan, to tie the PSInSAR deformation with GNSS data through a Kriging scheme, to calculate the absolute Line-of-Sight (LOS) deformation from PS points, and eventually to obtain the decomposed three-dimensional displacements from a single orbit. Leave-one-out strategy with individual GNSS stations are applied for validating the GNSS-based decomposition model. Also the accuracy of vertical deformation from GNSS-based decomposition is compared to the presented effective method of the combination of dual-orbit imageries. The correlation and root-mean-square error of the difference (RMSE) is 0.91 and 12.56 mm, respectively, in the vertical displacement estimates, which outperforms solutions from dual-orbit combination whose correlation and RMSE is 0.52 and 39.24 mm, respectively. The result demonstrates the error from lengthened unwrapping path is well constrained by GNSS network. From this dataset, we can clearly identify the seasonality and spatial pattern of land subsidence caused by different types of land use. Especially in the western alluvial plain of Taiwan, Changhua, Yunlin, Chiayi and Pingtung county are further analyzed with precipitation and subsidence rate to compare the investigation from MOEA.
顯示於類別:	[土木工程研究所] 博碩士論文

文件中的檔案:

檔案	描述	大小	格式	瀏覽次數
index.html		0Kb	HTML	91	檢視/開啟

在NCUIR中所有的資料項目都受到原著作權保護.

社群 sharing

資料載入中.....