歐亞板塊在臺灣南部之特徵分析： 來自SALUTE地震資料的觀察

Abstract

台灣造山帶位於菲律賓海板塊與歐亞板塊之交界，板塊快速聚合與頻繁地震顯示該區域造山作用極為活躍。雖然已有眾多研究探討台灣中部與南部的板塊交互作用，但對於其最佳構造模型仍未有定論。其中一項關鍵問題，是歐亞板塊於台灣中部向東隱沒的範圍與延續性，以及其與向西北隱沒的菲律賓海板塊之間的互動關係。為了釐清此一構造特徵，SALUTE團隊佈設了密集的地震觀測網，包含29座陸地寬頻地震站與8座寬頻海底地震儀，分布於台灣南部及其東方近海地區，涵蓋由隱沒過渡至碰撞的構造帶。本研究利用來自不同方位遠震事件的P波初達波走時以及振幅比值異常進行分析，結果顯示，走時殘差除在西部平原相對較慢，皆能以區域三維速度模型之走時解釋，振幅比值在不同事件中呈一致趨勢，並能觀察到多路徑到時效應，特別是在潮州斷層與縱谷區域為低點，呈現W型振幅比值，對應班尼奧夫帶兩側之散焦效應。綜合波形、走時與振幅比值等觀測資料，推論臺灣中部至南部下方可能存在一隱沒板塊結構。 為了解可能的隱沒板塊形貌，本研究運用有限差分法進行數值模擬，結合多筆地震事件之震源參數與不同隱沒板塊幾何形貌（含傾角、長度、寬度、上邊界深度與速度變化率），建構模擬資料庫。透過與觀測資料之波形與振幅比值比較，選取最佳擬合模型以評估板塊構造對波場特性的影響，本研究所假設的伸直板塊模型能有效對應東西兩側之邊界位置與振幅比值的散焦效應，根據各事件最佳模型，東邊界位於中央山脈與花東縱谷之邊界，西邊界可能位於東經120.3-120.6度之間，惟於中央山脈下方所觀測到之振幅比值提高無法單以隱沒板塊模型解釋。進一步納入淺部速度模型後，模擬已能初步重現觀測中的W型振幅比值趨勢，差異可能與地殼中橫向速度異質性與構造界面等細節有關。結果顯示，單獨使用隱沒板塊模型或淺部速度模型皆無法完整重現觀測特徵，唯有兩者結合，方能合理解釋振幅比值之空間分布與W型趨勢。

The Taiwan orogen is located at the plate boundary between the Philippine Sea Plate (PSP) and the Eurasian Plate, where a high convergence rate and frequent seismic activity indicate that the ongoing transition from subduction to collision is highly active. Despite extensive research on the complex plate interactions in central and southern Taiwan, a consensus has yet to be reached on the most appropriate tectonic models. To investigate this, the SALUTE project deployed a dense network of 29 broadband seismic stations on land and 8 broadband ocean bottom seismometers (BBOBSs) across southern Taiwan and eastern offshore regions. In this study, we analyzed the P-wave first-arrival travel times and amplitude ratio anomalies from teleseismic events originating from various azimuths. The travel-time residuals, except for slower velocities observed in the Western Plains, can be explained by the regional 3D velocity model. The amplitude ratios exhibit consistent patterns across different events, with multipathing effects observed. Notably, low amplitude ratios are observed near the Chaozhou Fault and the Longitudinal Valley, forming a characteristic W-shaped pattern that corresponds to defocusing effects on both sides of the Benioff zone. By comparing synthetic data generated using the 2-D finite-difference method with observations, we identified the best-fitting models to evaluate the impact of slab geometry on wavefield characteristics. The simplified slab model proposed in this study effectively matches the boundary positions on the east and west sides and reproduces the defocusing effects observed in the amplitude ratios. According to the optimal models for individual events, the eastern boundary lies between the Central Range and the Longitudinal Valley, while the western boundary likely falls between 120.3°E and 120.6°E. However, the increased amplitude ratios observed beneath the Central Range cannot be explained solely by the subducting slab model. These results suggest that only through the integration of both the subducting slab model and the shallow velocity model can the spatial distribution and W-shaped pattern of amplitude ratios be reasonably explained.