台灣是GNSS觀測網密度最高的國家之一,同時也位於板塊邊界的地震活躍區。除了現有的地震儀資料外,若能加入GNSS地表位移資料,將可提供分析地震震源的另一套參考依據。本研究提出使用GNSS高頻即時觀測及點震源參數反演模型測試近年台灣的一些重要地震事件,其中包含規模為Mw 6.4的2018年花蓮地震。我們期望這個方法能輔助現有的以地震儀資料為基礎的逆推系統,以克服其在中大地震發生後的近場儀器超格問題,達到即時使用近場GNSS真實地表位移資料來快速進行計算的目的。未來台灣若能將高頻GPS建立於即時系統,本研究之GNSS逆推模型將能在中大型地震的即時地震預警或海嘯預警上提供重要的資訊。 ;This research proposes to build an earthquake source inversion model based on GPS observations from the GNSS network in Taiwan, one of the most seismically active countries in the world. The model uses the daily and high-rate coseismic displacements to invert for focal mechanism and the centroid location of moderate to large earthquakes simultaneously. The method uses a point source approach to evaluate source parameters in an efficient way without a priori knowledge of fault geometry, and the results can otherwise provide constraints on subsequent finite source modeling of fault slip. We will apply this method to investigate the 2018 Mw6.4 Hualien earthquakes, which has a relatively moderate magnitude but complex rupture scenario and near-field surface displacements. We also attempt to promote this qusai-real time method to a more urgent requirement such as the early-warning.