琉球隱沒帶西南端的剪力波非均向性及對地幔流場特性的意涵

Abstract

琉球隱沒帶西南端接近台灣東北部區域，為菲律賓海板塊斜向碰撞並隱沒至歐亞大陸板塊的聚合邊界。過去研究台灣本島底下震波非均向性性質皆顯示快波極化方向大致與臺灣造山帶走向平行，呈近南-北向，主要與岩石圈的變形有關。關於臺灣-琉球隱沒帶的非均向性特性和隱沒帶周圍地幔流場機制，因受限於測站分布，相關的研究較少。 本論文使用台灣島北部和最接近臺灣的兩個琉球群島寬頻測站所記錄的震波資料，分析剪力波分離參數，即S快波極化方向( )與快慢波分離的時間差( )。觀測結果顯示，琉球群島上的兩個測站IGK和YNG剪力波分離，其快波極化波方向大致與琉球海溝的方向平行；台灣北部彭佳嶼島的PCYB站結果則垂直琉球海溝；在台灣島東北部測站TWBB與TIPB的快S波極化方向由東西向轉為南北向。¬另外遠震S與SKS波分離的延遲時間約為1.5至2秒，而近震S波約為0.2至1秒，且與波線路徑呈正比關係，推測其震波非均向性範圍至少延伸至三百公里深處。 典型的地幔楔流場是由隱沒板塊向下拖曳引起垂直海溝的地幔對流模式，可由簡單的二維角隅流場來描述。因為地幔的主要組成礦物A-型態橄欖石晶體的快軸(a-軸)傾向平行剪切力即流場方向排列，與觀測剪力波分離所得到平行海溝方向抵觸。但近年來礦物變形實驗發現在含水量高、應力大的弧前環境，橄欖石主要以B型態的形式存在，即b-軸為快軸，所以a-軸(慢軸)平行流場方向則造成b-軸(快軸)平行海溝走向，與觀測結果符合。 為了測試弧前區域的B型態橄欖石對所觀測的平行海溝的非均向性有多大的影響，透過簡化的二維地幔楔流場運動模型，計算礦物因剪切應力變形呈晶格優選排列型式和以及對應的非均向震波速度構造，然後藉由數值模擬剪力波在該速度介質中傳遞產生的合成波，比較合成與實際記錄波形所量測的分離參數異同，進而了解隱沒板塊的地幔流場與弧前區域非均向性的關係。 結果顯示，不管是隱沒板塊近震或遠震S波的模擬結果，快方向在模型設置B型態橄欖石存在的弧前區域，均為平行海溝的方向。但是遠震SKS波的模擬結果，其快方向在弧前區域為垂直海溝方向的結果，意味著隱沒板塊下方的地幔A型態橄欖石優選排列的非均向性性質主導著SKS波的分離。另外我們也測試了隱沒板塊與地幔楔蛇紋岩化非均向性構造對剪力波分離的影響，結果與B型態橄欖石模擬結果相似。因此，在臺灣-琉球隱沒板塊邊緣地帶，可能存在於隱沒板塊下方沿海溝走向平行的側向流場以及從邊緣回流至地幔楔的三維流場，才足以解釋觀察到的震波非均向性結果。

The southern Ryukyu trench off northeast Taiwan marks the plate boundary where the Philippine Sea plate obliquely subducts beneath the Eurasian plate. The Ryukyu trench is oriented in a NE-SW direction to the east of 124oE and bends more E-W oriented to the west. To illuminate the nature of the mantle wedge flow around the edge of the Ryukyu subducting slab, we map the strength and geometry of upper mantle seismic anisotropy constrained by shear wave splitting analysis of local S waves and teleseismic SKS and S waves recorded at two stations IGK and YNG closest to Taiwan operated by the Japan F-net seismic network. Our results indicate that the fast directions of both local and teleseismic shear waves observed at the two stations are approximately parallel to the local strike of the Ryukyu trench. Fast directions observed at station PCYB located in the Okinawa trough to the north of Taiwan show more trench-perpendicular. The fast directions observed at stations TWBB and TIPB in the northeast corner of Taiwan exhibit the transition from east-west to north-south directions. The split time delays of the teleseismic SKS and S waves are about 1.5-2 s, while those of the local S waves increase with the propagation path lengths ranging from 0.2 to 1 s, suggesting that seismic anisotropy at least extend down to 300 km depth. A 2D corner flow induced by the drag of a downgoing slab represents the classical view of the mantle flow at SZs. The lattice preferential orientation (LPO) of the fast a-axis of olivine is known to be parallel to the flow direction, that is, trench normal for the corner flow geometry, which is contradictory to the observed trench-parallel fast directions. Recent laboratory experiments on olivine aggregates suggest that under the condition of higher stress and higher water content that may exist in the fore arc region, the B-type fabric becomes dominant with the faster b-axis aligned parallel to the trench. To test whether the B-type texture in the fore-arc region can be sampled by finite-frequency local and teleseismic shear waves resulting in the observed trench-parallel anisotropy, we conduct full wave propagation modeling through the upper mantle structure in the framework of a 2D corner flow field using the numerical pseudospectral method. The adopted seismic velocity structure is based on LPO of olivine and pyroxene aggregates and the corresponding anisotropic elastic structure calculated from the plastic deformation model. The predicted fast directions from synthetic split SKS waves recorded above both the mantle wedge and subslab mantle are parallel to the flow direction, implying that the trench-normal anisotropy induced by the A-type fabric in the back arc and subslab mantle predominantly contribute to the observed SKS splitting. We also test the model with serpentinization in hydrated mantle wedge and subducted slab. The resulting pattern of shear wave splitting is similar to that from B-type fabric. Therefore, the alternative model like complex 3D geometry of the mantle wedge flow around the slab edge may need to explain the observed seismic anisotropy.