臺灣地處歐亞大陸板塊與菲律賓海板塊碰撞擠壓交界位置，地震活動頻繁，但對於仍有許多小規模地震活動在非已知出露地表之斷層附近，這些地震活動之位置，常以盲斷層來做解釋。中央氣象局地震觀測網隨著時間的演進，地震觀測能力不斷提升，自2012年後全面升級為24位元即時地震觀測站，除了可以觀測到過往未能觀測到之更微小的地震事件外，在納入地震目錄之小規模地震觀測數量，也因此以倍數增加。 由於地震規模與釋放能量對數成正相關，小規模地震之破裂，在幾何空間上之影響範圍相較於大規模地震微小許多，本研究透過地震挑波及定位程序改善，以縮小不確定因素影響，經重新地震波相到時挑選，並採用三維地震波速度模型進行地震定位與雙差分重新定位，可獲得更為精確之地震定位結果。 接著修改第二階地震矩方法之處理程序，在前述地震定位結果為基礎上，選擇規模3到4間地震做為目標地震，並設定條件尋找個別的候選經驗格林函數，進行視震源時間函數計算，再以走向0˚~359˚及傾角0˚~90˚共32,760個組合進行第二階地震矩逆推運算，經由誤差評估及候選經驗格林函數篩選，取得最終目標地震的破裂位態。 此結果趨勢與較大規模地震震源機制之斷層面解一致，又本研究方法可獲得地下更多且更細微之破裂位態，有助於未來對未知地下構造進一步了解。 ;Taiwan is located at the collision and compression boundary between the Eurasian Plate and the Philippine Sea Plate, resulting in frequent seismic activity. Many of these small-scale seismic activities occur near faults that are not exposed on the surface, often explained by blind faults. Over time, the earthquake observation capabilities of the Central Weather Bureau′s seismic network have continuously improved. Since 2012, it has been upgraded to a 24-bit real-time earthquake observation system. This upgrade not only detects smaller seismic events that were previously undetectable but also significantly increases the number of observed small-scale earthquakes included in the earthquake catalog. Due to the logarithmic correlation between earthquake magnitude and released energy, the rupture of small-scale earthquakes has a much smaller geometric impact compared to large-scale earthquakes. This study improves earthquake picking and location procedures to reduce uncertainties. By reselecting earthquake arrival times and employing a three-dimensional seismic velocity model for earthquake location and double-difference relocation, more accurate earthquake locations can be achieved. Furthermore, modifications were made to the processing procedures of the second-order seismic moment method. Based on the aforementioned earthquake locations, earthquakes with magnitudes between 3 and 4 were selected as target events. Individual candidate empirical Green′s functions were selected under specified conditions for apparent source-time function calculations. Subsequently, the second-order seismic moment inversion was conducted using 32,760 combinations of strike (0° to 359°) and dip (0° to 90°), followed by error assessment and candidate empirical Green′s function filtering, to obtain the final rupture configuration of the target earthquake. The obtained rupture configuration trends consistently with fault plane solutions of larger-scale earthquakes, while the proposed method allows for more detailed characterization of underground ruptures, facilitating further understanding of unknown subsurface structures in the future.
