臺灣半動態基準之建立與展望

臺灣位於板塊邊界的區域，會因板塊間的相對運動使得靜態國家坐標系統隨時間的推進而漸失去其原有的精度。然而，臺灣現行的TWD97[2010]仍屬於靜態的坐標系統，故如何有效延續此國家坐標系統的精確性乃為國家測繪工作的一項重要任務，而建立臺灣地區的半動態基準則扮演了相當重要的角色。半動態基準包含了靜態大地基準(即TWD97[2010])與地表變形。地表變形模式則包含了「速度網格模型」和「位移網格模型」。在建立速度網格模型方面，本研究完成全臺2000 年01 月至2016 年09 月437 個GPS連續站資料之計算，獲得全臺的水平速度場，並結合785個經濟部中央地質調查所2002 年至2015 年之GPS水平速度場，利用克立金空間內插法建立臺灣地區的水平速度網格模型。在建立位移網格模型方面，本研究藉由比較利用不同精度星曆計算得到的同震位移場，顯示所有計算結果近乎一致，亦即利用超快速星曆求解GPS資料即可在最短時間內建立高精度的近即時地表同震位移場。同時在建置震源模型方面，亦指出單一斷層滑移量模型所獲得的位移網格模型較空間內插模型所得到的結果更加合理。本研究進一步利用半動態基準的地表變形模型檢核內政部國土測繪中心2013 年與2016 年基本控制點檢測成果。採用半動態基準可將2013 年之坐標轉換至2016 年的計算坐標，再將其與2016 年實際觀測的檢測坐標成果進行差異比較發現，兩者有相當高的一致性，有91%的測站坐標差異量在6 公分以內。而剩餘未能被適當估算的基本控制點多分布在山區，可能是由大規模崩塌等區域活動所造成。本研究也建議半動態基準可以配合內政部國土測繪中心現行之e-GNSS系統，利用e-GNSS基準站的坐標時間序列協助檢驗地表變形模式之速度模型精度變化，幫助決定地表速度場更新時機之判斷。最後，就高程半動態基準的發展而言，除了解析長期穩定的震間垂直速度場之外，未來還需要掌握例如地層下陷等非自然因素造成的地表垂直位移，以達到高程基準更新維護之目的。

關鍵字

半動態基準；地表變形模式；克立金空間內插；斷層錯位模型；電子化全球衛星即時動態定位系統

並列摘要

The accuracy of static national coordinate system in any country, located at the present-day plate boundary such as Taiwan, will be decreased due to the relative motions among different plates. However, the current national coordinate system in Taiwan, TWD97[2010], is still a static datum. Therefore, maintaining the accuracy of national coordinate system in high level efficiently is the main issue in Taiwan. One solution of this issue is to establish a semi-dynamic datum. The semi-dynamic datum is consisted of a static geodetic datum, i.e., TWD97[2010] in Taiwan and a surface deformation model, containing the velocity grid model and the displacement grid model. In this study, observations of the 437 continuous GPS sites from Jan. 2001 to Sep. 2016 were adapted to estimate the horizontal velocity field. We also integrated the horizontal velocity field between 2002 and 2015 from 785 campaign-mode GPS sites surveyed by the Central Geological Survey to derive the horizontal grid velocity model in Taiwan by using the Kriging spatial interpolation method. Furthermore, the coseismic displacements of 2016 Meinong earthquake estimated using ultra-rapid ephemerides are consistent with the results using final ephemerides. In other words, the near real-time coseismic displacement field would be solved using ultra-rapid ephemerides in the future. In addition, the displacement grid model inferred from the uniform slip model is more suitable than the model derived from the spatial interpolation. The resurveying results of the basic control points from the National Land Surveying and Mapping Center (NLSC) in 2013 and 2016, respectively, were used for the exterior checking on the accuracy of semi-dynamic datum in this study. The calculated coordinates in 2016, derived from observations in 2013 using the semi-dynamic datum, are nearly corresponding with the observed coordinates in 2016. The coordinate differences in 91% stations are less than 6 cm. The outliers of comparisons are mostly located in the mountain range, which may be caused by the deep-seated landslides. In addition, the e-GNSS established by the NLSC is proposed to be used for checking the accuracy of velocity model and for the timing decision of velocity model renewal. Finally, in order to develop the vertical semi-dynamic datum, we have to estimate the vertical velocity field and collect land subsidence information simultaneously for improving and maintaining the vertical datum in a high accuracy level.