中長距離高精度GPS動態定位應用於強地動的研究

Using medium-range Kinematic high accuracy GPS relative positioning to surdy ground motion

由地震觀測資料，可以獲得板塊界面的位置、地震發生的規模、地面產生的瞬時速度與加速度；但對於地面造成的瞬時地形變、無震滑移、以及震後變形，卻不易精確獲得。而高解析度的GPS衛星連續觀測資料，可即時記錄地動位移向量，得以和地震學的研究緊密結合，提供更多的資訊，因而應用於強地動的中長距離高精度GPS動態定位研究，日益顯得重要。差分的中長距離GPS動態定位技術，取決於週波未定值的求解，除了考慮衛星軌道誤差、訊號多路徑的影響、天線相位偏移等，更須注意隨距離而增加的誤差，如對流層及電離層的大氣誤差，造成週波未定值不易正確求得，進而影響定位精度。針對強地動的研究要求，本文除了分析即時求解的動態定位技術及單一歷元求解方法外，更引進後處理的動態定位技術，比較中長距離的定位精度及穩定性。定位成果均經過統計分析及粗差偵測，針對時間序列分析去除伴隨每一恆星日重複出現的系統誤差，如多路徑效應等。本文以集集地震（ML=7.3）之同震位移及2002年331花蓮外海地震（ML=6.8）的即時地表變形運動為例，分析其定位精度及效能。透過以上兩個實例發現，慎選卡曼濾波（kalmanfilter）的即時求解動態定位技術，可求得地震瞬時的位移向量，但對於長時間的定位結果，則易於失真，而單一歷元的求解方法，於高採樣率（1秒）的花蓮外海地震即時地表變形運動，可獲得其運動過程，但無法求得低採樣率（30秒）的集集地震之同震位移結果。 By analyzing the seismic data, the position of the plate boundary, the magnitude of the earthquake and the instantaneous acceleration of the monitor station can be obtained, but the simultaneous ground motion, a-seismic slip and post-seismic deformation is hard to obtain. The continuous observation GPS data in high sampling rate can record the instantaneous ground motion to supply more information for seismology study, hence the medium-range high accuracy GPS kinematic positioning research for strong motion applications is increasingly important. High accuracy medium-range GPS kinematic positioning is determined by obtaining the correct ambiguities. Besides, the effect of the multipath, orbit errors, and phase center variations of the antenna, and the distance dependent biases such as tropospheric and ionospheric biases, present major challenges in the estimation of the ambiguities. In order to study the strong ground motion during the earthquake, this paper is not only analyzing the real-time ambiguity resolution techniques and the single epoch solution of the kinematic positioning of the commercial software, but also applying the post-processing algorithm to compare the accuracy and stability of the positioning. After outlier detection by using the interquartile range edit criterion, the positioning results employ a sliding 3-sideral-day window to eliminate the sidereal period noise (Bock et al., 2000), such as multipath and phase center variation. In this study, the strong ground motion is obtained from a ML=6.8 earthquake occurred on 31 March 2002 off northeastern Hualien (“3.31” earthquake), by applying these GPS data to analyze the position accuracy and stability of the different positioning algorithms. Through analysing the “3.31” earthquake, it has been shown that instantaneous displacement can be estimated by choosing the suitable Kalman filtering of the real-time positioning algorithm can estimate the instantaneous displacement but there is a large drift after a long time positioning. On the other hand, the ground motion can be obtained by selecting the high sampling rate of the single epoch solution algorithm, but there is a jump for a few epochs.

By analyzing the seismic data, the position of the plate boundary, the magnitude of the earthquake and the instantaneous acceleration of the monitor station can be obtained, but the simultaneous ground motion, a-seismic slip and post-seismic deformation is hard to obtain. The continuous observation GPS data in high sampling rate can record the instantaneous ground motion to supply more information for seismology study, hence the medium-range high accuracy GPS kinematic positioning research for strong motion applications is increasingly important. High accuracy medium-range GPS kinematic positioning is determined by obtaining the correct ambiguities. Besides, the effect of the multipath, orbit errors, and phase center variations of the antenna, and the distance dependent biases such as tropospheric and ionospheric biases, present major challenges in the estimation of the ambiguities. In order to study the strong ground motion during the earthquake, this paper is not only analyzing the real-time ambiguity resolution techniques and the single epoch solution of the kinematic positioning of the commercial software, but also applying the post-processing algorithm to compare the accuracy and stability of the positioning. After outlier detection by using the interquartile range edit criterion, the positioning results employ a sliding 3-sideral-day window to eliminate the sidereal period noise (Bock et al., 2000), such as multipath and phase center variation. In this study, the strong ground motion is obtained from a ML=6.8 earthquake occurred on 31 March 2002 off northeastern Hualien (“3.31” earthquake), by applying these GPS data to analyze the position accuracy and stability of the different positioning algorithms. Through analysing the “3.31” earthquake, it has been shown that instantaneous displacement can be estimated by choosing the suitable Kalman filtering of the real-time positioning algorithm can estimate the instantaneous displacement but there is a large drift after a long time positioning. On the other hand, the ground motion can be obtained by selecting the high sampling rate of the single epoch solution algorithm, but there is a jump for a few epochs.