台灣西南部半動態基準之建立

Establishment of Semi-Dynamic Datum in SW Taiwan

臺灣目前採用的”靜態式”大地坐標基準（Taiwan Geodetic Datum 1997, TWD97），因境內菲律賓海板塊與歐亞板塊之擠壓，導致隨著時間的累積而漸失其精度。本研究蒐集中央地質調查所2004至2010年台灣西南部221個GPS移動站的水平速度資料，應用克立金內插法以及塊體模型法產生區域內的水平速度模型，並據此建立半動態基準以維護區域內坐標系統的精度與延長使用年限，同時，本研究亦另使用區域內獨立的44個GPS連續站的水平速度資料進行模型精度之驗證。由分析結果得知，以克立金內插法建立的地表水平速度模型，移動站的模型推估值與觀測值的差異量，在N、E方向的均方根值均為±0.3mm/yr；連續站則在N、E方向分別為±1.9mm/yr與±2.2mm/yr。以塊體模型法建立的速度模型，其移動站的模型推估值與觀測值的差異量，於N、E方向分別為±2.0mm/yr與±2.8mm/yr；連續站的差異量則分別為±2.0mm/yr與±2.9mm/yr。最後亦將坐標轉換法之成果納入比較，結果指出在沒有瞬間大型的板塊位移發生時，克立金內插法建立的速度模型可以維持其精度約25年，塊體模型可維持約15年，而坐標轉換法則維持不到3年。 The precision of Taiwan’s current static geocentric datum (Taiwan Geodetic Datum 1997, TWD97) is losing with time due to the collision between the Eurasian and the Philippine Sea plates. In order to sustain the precision of Taiwan’s datum, observations from 221 campaign-mode GPS stations and 44 continuous GPS stations in SW Taiwan are used to examine the performance of semi-dynamic datum. In this study, we first estimated a horizontal velocity field in SW Taiwan using the data from 221 campaign-mode GPS stations. Then the velocity models were constructed by adopting an exponential function in a Kriging interpolation model and a tectonic block model (DEFNODE), respectively. The RMS misfits of velocity model inferred from the Kriging interpolation in N and E components are ±0.3 mm/yr and ±0.3 mm/yr, respectively. The RMS misfits of velocity model derived from the block model in N and E components are 2.0 mm/yr and 2.8 mm/yr, respectively. Next, observations from 44 continuous GPS stations were adopted to realize the precision of semi-dynamic datum. The performances of the Kriging interpolation are ±1.9 mm/yr in N component and ±2.2 mm/yr in the E component, respectively. For the block model, the RMS of residual errors were averaged ±2.0 mm/yr in N component and ±2.9 mm/yr in E component. The coordinate transformation is also used in this study for comparison with the velocity models. Assuming no large coseismic offsets and the surveyed coordinate difference less than 24 mm, the precision of the Kriging interpolation model is estimated to survive ~25 years, the precision of block model could be kept about 15 years, and the precision of coordinate transformation is evaluated to preserve less than 3 years.

The precision of Taiwan’s current static geocentric datum (Taiwan Geodetic Datum 1997, TWD97) is losing with time due to the collision between the Eurasian and the Philippine Sea plates. In order to sustain the precision of Taiwan’s datum, observations from 221 campaign-mode GPS stations and 44 continuous GPS stations in SW Taiwan are used to examine the performance of semi-dynamic datum. In this study, we first estimated a horizontal velocity field in SW Taiwan using the data from 221 campaign-mode GPS stations. Then the velocity models were constructed by adopting an exponential function in a Kriging interpolation model and a tectonic block model (DEFNODE), respectively. The RMS misfits of velocity model inferred from the Kriging interpolation in N and E components are ±0.3 mm/yr and ±0.3 mm/yr, respectively. The RMS misfits of velocity model derived from the block model in N and E components are 2.0 mm/yr and 2.8 mm/yr, respectively. Next, observations from 44 continuous GPS stations were adopted to realize the precision of semi-dynamic datum. The performances of the Kriging interpolation are ±1.9 mm/yr in N component and ±2.2 mm/yr in the E component, respectively. For the block model, the RMS of residual errors were averaged ±2.0 mm/yr in N component and ±2.9 mm/yr in E component. The coordinate transformation is also used in this study for comparison with the velocity models. Assuming no large coseismic offsets and the surveyed coordinate difference less than 24 mm, the precision of the Kriging interpolation model is estimated to survive ~25 years, the precision of block model could be kept about 15 years, and the precision of coordinate transformation is evaluated to preserve less than 3 years.