- 期刊
臺灣GNSS連續追蹤站之時間序列資料處理與分析
PROCESSING AND ANALYZING GNSS TIME SERIES AT CORS STATIONS IN TAIWAN
摘要
全球衛星導航系統（Global Navigation Satellite System, GNSS）應用廣泛，具備可全天候連續作業、儀器操作簡便、高頻率與高精度三維坐標定位成果等特點，以GNSS連續站或移動站進行地球科學研究、測量、製圖或變形監測已成趨勢。以臺灣而言，目前平均分布全臺之GNSS連續追蹤站（Continuously Operating Reference Station, CORS）約有五百之數，可提供各地連續之地表運動行為供各界應用。另一方面，由於地表位移、調整GNSS接收儀天線位置等自然因素或人為因素，導致觀測站點位本身位置不連續，使資料判讀有誤，影響分析成果品質。此外，為得較正確之地表變位訊號，需以適當模式濾除資料蒐集過程中因環境因素產生之雜訊。因此，本研究發展自適應不連續點偵測、趨勢線擬合、粗差過濾、趨勢線精化、統計測試以及速度場與精度估計等步驟，建立觀測站點位運動行為之自動化分析模型，並以全臺實際GNSS觀測站之長期觀測資料實作：速度精度優於10^(-4)m/yr之比例於e、n、u三方向分別為96%、96%、90%，後驗單位權標準差於e、n、u三方向之整體平均為（±0.004,±0.004,±0.009）m，其驗證了本方法之適用性與穩定性，根據完整GNSS時間序列資料處理與分析流程合理評估並增進時間序列資料品質，提升速度場成果正確度與可靠度，作為後續地表位移相關分析之所需。
並列摘要
At present, the global navigation satellite system (GNSS), with 24-hour observation, simple operating instruments, and precise positioning coordinates with high frequency, is widely applied to various fields, such as geoscience, with the GNSS being used for surveying, mapping, and monitoring the deformation. Approximately 500 GNSS continuously operating reference stations (CORS) are constructed and evenly distributed in Taiwan, providing local behaviors of the ground motion for the application of various fields. However, both natural (e.g., earthquakes) and human factors (e.g., adjusting the antenna of the GNSS receiver) may result in the discontinuity of the station position and even cause the misinterpretation of the data. To accurately interpret the data, the displacement signals should be collected correctly by an applicable calibrating method to eliminate noises from the environmental factors. This study attempted to build the automation model to describe the motions of the station by employing adaptive detection of discontinuous points, trend line fitting, gross error filtering, line fitting with optimization, statistic testing, and estimation of velocities with accuracy. The long-term observation data of the GNSS observatories would be used to verify the applicability and stability of the method. Finally, the ratios of standard deviation of velocity better than 10^(-4)m/yr were 96%, 96% and 90% in the three directions of e, n and u respectively. Moreover, the overall average of the standard deviation of the posterior unit weight in the three directions of e, n and u was (± 0.004, ± 0.004, ± 0.009) m. Those results meant the complete procedure of GNSS time series data processing and analyzing could be provided to evaluate the data quality rationally and to improve the data reliability properly for relevant research requirements in a surface displacement monitoring task.