利用大地測量與地震觀測資料探討臺灣造山帶與琉球隱沒帶地殼形變

Abstract

地殼形變是研究地體構造演化與斷層活動災害的重要學問，利用大地測量與地震觀測資料於時間與空間上的取樣以及約制，能幫助地球科學家了解地殼運動速率與斷層面上滑移速率。這兩個議題的了解往往取決於地表與地殼變形的觀測資料在時空上取樣密集與精確程度，因此，觀測資料網的建置與發展是提升區域地殼變形解析能力相當重要的必備工作。本論文以擁有高密度大地測量觀測網與地震觀測網的臺灣地區為例，利用分析多重的更新資料(包含GPS位移、重定位地震、震源機制解)提升臺灣造山帶與其鄰近琉球隱沒帶的地表與地殼變形解析能力，進而對這兩個區域的地殼變形行為做出描述與解釋。 本論文主要由兩個計畫所組成。在第一個計畫中，我結合臺灣地區更新至2015年的GPS連續觀測網位移資料與CWBSN、TSMIP地震觀測網P波初達波資料分析臺灣的近期地表與地震變形。關於臺灣造山帶地殼形變與其機制過去已有許多地球物理觀測資料如地表應變、地殼應力、地殼非均向性構造等累積研究成果，但這些研究的觀測資料對於地殼形變的空間解析能力相異，尤其，普遍在深度上僅呈現整個地殼內部平均結果。近期的非均向性構造研究成果指出造山帶地殼變形在深度上具有層狀變化的現象與機制，因此，我希望以不同於地震波非均向性構造的物理角度，利用最新的地表應變與地殼應力資料提升地殼變形空間解析度，並釐清其內部變形是否存在與深度相關的變化。使用臺灣地區上述大地測量與地震波相更新資料，我求得地表水平速度場與地殼內部震源機制解，進而計算出臺灣地區過去以來在空間上最佳解析度的地表應變場與地殼應力場，以及第一個在深度上具有層狀解析能力的地震斷層區域性型態。計算過程中考慮發生於造山帶內的Mw 7.6集集大地震在震後期間資料所紀錄到的近震源區GPS測站速度偏移以及應力方向偏移等效應做修正，減少其影響應變場與應力場長期平均的計算結果。新的應變與應力場表現出造山帶地殼變形在深度上具有顯著的方向改變與斷層型態變化，造山帶碰撞區以20公里為界呈現上下部地殼之間應力場異質性，此異質性在地殼增厚區與薄化區的整體空間更加顯著；反之，地殼應力場空間上相對均質性則表現於地殼山根區。在第一個計畫中藉此成果，新的地表應變場與地殼應力場大提供了過去相似成果中所無法解析的地殼變形於深度上的變化，同時得出地表應變型態可能反映地殼內部變形的區域性深度範圍。 在第二個計畫中，利用臺灣東北部高密度連續GPS測站網與大量的重定位地震、震源機制解資料聯合分析，我首次研究東北部外海琉球隱沒帶(最南段)之斷層面在間震期潛移滑動行為。過去整個琉球海溝被認為板塊界面屬於弱耦合，並且以潛移方式釋放在間震期隱沒過程中地殼累積應力，因此，歷史紀錄上琉球海溝極少發生規模8以上地震。這個假說已藉由GNSS資料在大部分琉球海溝所觀測的慢滑事件推論得知，但在最南段琉球海溝尚缺乏觀測資料支持此假說。使用上述的連續GPS測站網紀錄到的慢滑事件，以及同步發生區域地震活動觀測資料，我發現時間序列上潛移訊號與區域性地震活動有良好時空對應關係，並在地震活動過後仍有持續潛移的訊號被觀測到。這些潛移訊號藉由計算累積位移與累積地震數量之間所呈現的非線性關係可驗證乃由慢滑事件所引起。為了瞭解這些偶發性長期慢滑事件在隱沒帶可能的發生位置以及時空演育過程，我以格點搜尋法求取斷層錯位模型最佳參數，並且以TDefnode code反演連續GPS三分量時間序列。反演斷層滑移模型能表現連續GPS觀測紀錄最大地表變形量區域，並得出偶發性慢滑事件累積滑移量產生的變形等同於地震矩規模6.4至6.6地震破裂能量釋放。藉由分析慢滑事件與地震在時空上的關係，我們也發現潛移滑動區域與地震密集帶存在著調和現象。這個計畫的研究結果直接驗證了最南段琉球隱沒帶的偶發性潛移機制，並且推論此機制可能重現於多重地震週期中。 最後，我彙整並討論兩個計劃的成果與未來的工作展望。此外，也將另一個在博士期間完成的子計畫：花蓮北部板塊交接帶現今的地殼變形，簡介並附錄於論文之後。

Constraints of geodetic and seismological observations on the crustal deformation in space and time, help geoscientists understand the significant issues about the rates of crustal motion and fault slips. Understanding of the two issues strongly depends on how dense and accurate observations we can have, and it is therefore of importance to develop the networks. In this dissertation, by analyzing multiple data sets from abundant continuous GPS (cGPS) and earthquake observations in Taiwan region, the crustal deformation is better clarified at Taiwan orogen and the nearby Ryukyu Trench. This dissertation mainly consists of two projects. In the first, crustal deformation of the Taiwan orogen is studied by jointly using the updated data sets of surface positions and P-wave first arrivals from the cGPS and CWBSN/TSMIP networks, respectively. The crustal deformation was estimated from geophysical observations from the states of surface strain, crustal stresses, and anisotropic structures. However, most of them shown average results of the overall crust in depths. Recent anisotropic studies indicate that the Taiwan orogeny is layered mechanism, and therefore new strain and stress data are used to verify the mechanism which is independent of anisotropic structures in physical meanings. By jointly analyzing the two updated data sets, updated strain and stress fields with so far the best resolution in the space is obtained. The styles of faulting in Taiwan orogen with resolution in depths is provided for the first time. In the estimation, to minimize the postseismic effects of Mw 7.6 Chi-Chi earthquake on long-term average of states of strain and stress, the effects on biases of cGPS velocities and stress directions are modified. The deformation fields can present heterogeneity of crustal deformation in the vertical space at a depth of 20 km in the main collision zone. The spatial heterogeneity of faulting styles is founded to be strong and weak in the areas of thinned/thickened and overthickened crust, respectively. According to the results of the first project, new results of strain and stress fields greatly improve the spatial resolution of crustal deformation and reveal regional crustal deformation from the surface to a maximum depth involved in geodetic strains. In the second project, Slow Slip Events (SSEs) in the southernmost Ryukyu Trench is studied for the first time by jointly using data of cGPS, relocated seismicity, and earthquake focal mechanisms. The Ryukyu Trench was considered that the crustal accumulated stresses are released by the interplate aseismic slips partly, in other words, the Trench is interseismic weak coupling. The hypothesis is still an open question at the southernmost Trench. By analyzing the cGPS position time series systematically, transient displacements are founded correlate to strong seismicity spatiotemporally and were succeed by a continued slip. The transient displacements are verified by a nonlinear relationship between cumulative number of earthquakes and cumulative displacements. To know the locations and evolution of episodic, long-term SSEs in the Trench, the displacements are inverted by TDefnode code based on dislocation fault modeling via a grid search method. The best fitting models can explain the areas of peak displacement and reveals slip amounts of the episodic SSEs resulting in deformation which is equivalent to Mw 6.4‒6.6 earthquake failure. Analyzing the relationship between SSEs and seismicity, the aseismic and seismic moment release in the shallow Trench is spatiotemporally modulated. The slow-slip mechanism is directly verified and suggested to exist in multi-earthquake cycles probably. I summarized the achievements for the two projects, and their prospections for future works. Additionally, another side project done in my Ph.D. period is also briefly introduced and supplemented: Current Crustal Deformation at the Junction of Collision to Subduction around Hualien area, Taiwan.