| 摘要: | 全球衛星定位系統(GPS)已經成為地殼變動及地體動力學研究的利器，連續密集的GPS觀測可以有效的用於偵測震前、同震和震後變形。2003年12月10日4時38分(UTC時間)，台東成功發生Mw=6.5之地震，GPS觀測的地表最大水平及垂直同震位移分量分別為126 mm及263 mm (近震央位置)。本研究使用GPS所觀測到的成功地震地表同震變形資料，以彈性半無限空間錯位模式，逆推最佳的斷層參數以及斷層面上的同震滑移量分佈，試圖了解震源斷層的特性及幾何型貌。 根據已知的斷層分佈圖，我們將縱谷斷層中段分為南北兩個走向略為不同的子斷層；又由餘震分佈資料獲知斷層在淺層之傾角較大，進入深部後，傾角趨於平緩，故將斷層分為淺層及深部兩個不同傾角的子斷層，共四個子斷層。模式中，約制南北兩斷層的長度皆為45 km，深部斷層寬度為50 km，斷層走向分別為：北段N25.3°E、南段N22°E。經過一連串的搜尋，我們獲得其餘三個斷層參數的最佳解：淺層斷層傾角(Dip1)為59°、深部斷層傾角(Dip2)為30°，淺層與深部斷層連接處深度(D)為21 km，平均滑移量為166 mm。最大同震滑移量為855 mm，約位於震源南方15 km處，而地震矩為3.4 10 dyne-cm。本研究為補強格點搜尋無法估計參數信心區間的缺點，針對三個斷層參數Dip1、Dip2及D，另外利用拔靴法(bootstrap)加以搜尋；而所求三個參數之最佳解分別為，59°、28°及22 km，與格點搜尋結果相近。在95%的信心區間下，分別落於48°~63°、5°~35°及18~27 km，顯示對於深部斷層傾角(Dip2)較無法解析。 而在震後變形部分，則分別對震後初期（0.1年內）及震後120天兩種不同衰減模式做擬合。初步結果發現其震後變形初期，大致上以 之對數函數模式為主；而在震後120天內，則大多以 之指數函數模式來加以描述，其各測站之分量震後鬆弛時間不盡相同。若要對其物理機制作更多解釋，則未來必須利用更長的資料及更嚴密之模式加以計算，並有賴其他相關地球物理研究共同討論。 Global Positioning System (GPS) has become an efficient tool for studying the seismic deformation and geodynamics. Dense continuous GPS data can be used in investigating the preseismic, coseismic and postseismic deformations. The Mw6.5 Chengkung earthquake occurred at 04:38 on 10 December 2003. The GPS observed coseismic displacements reached 126 mm and 263 mm in the horizontal and vertical components, respectively. Using the coseismic surface displacements of Chengkung earthquake, the optimal parameters of fault geometry and coseismic slip rate on the seismogenic fault are inverted by assuming a simple fault model in elastic half-space. According to the geologically recognized fault traces, the middle Longitudinal Valley Fault (LVF) is divided into two sub-faults with two a little bit different strikes. The distribution of the aftershocks shows that the dip of the seismogenic fault is larger in the shallow part, while then become smaller in the deeper part. Therefore, we divide each segment of fault into shallow and deep subfaults with different dip-angles. In our model, we define the length of each sub-fault to be 45 km, the width of the deep ones to be 50 km, and the strikes of the north and the south ones are N25.3∘E and N22∘E, respectively. Through a series of systematic searches, the optimal solutions of the three unknown parameters, the dip of the shallow sub-faults (Dip1 = 59∘), the dip of the deep sub-faults (Dip2 = 30∘) and the depth of the interface between shallow and deep sub-faults (D = 21 km) are derived. The average slip is 166 mm; the maximum coseismic slip is 855 mm and is located 15 km to the south of the hypocenter. The GPS derived seismic moment is 3.4*1025 dyne-cm. The bootstrap method is employed to estimate the confidence intervals of the three parameters, Dip1, Dip2 and D. The optimal solutions derived from bootstrap are 59∘, 28∘ and 22 km, respectively, which are consistent with the results from the grid search. Under 95% confidence intervals, the above three parameters are ranged from 48∘to 63∘, 5∘to 35∘ and 18 to 27 km, respectively. It indicates that the deep fault dip (Dip2) is less resolvable. For the postseismic deformation, we focus on fitting the transient decay forms of the primary (< 0.1 year) and the first four months of the postseismic data. The results show that the data are well fit by the logarithmic function a + b log10(t), during the primary of postseismic deformation. During the first four months, it is well fit by the exponential function a0 +a1 e-t/t, and the relaxation time of each component for each station are not quite the same. If we want to investigate the basic physical mechanism of postseismic deformation, more longer period of GPS data and a more sophisticated model are required in the future. |
