根據先前的研究顯示台灣區域之中小規模(Mw < 6.5)地震的點震源參數可以藉由線上即時監測系統(Real-time Moment Tensor, RMT)獲得。此方法是利用預先計算好之格林函數資料庫進行長週期(10 – 50 s)寬頻即時訊號連續逆推，結合格點搜尋方法，可以同時獲得震源位置、時間以及地震矩張量。此方法不需等待氣象局發布快速地震測報資訊，在實際應用上相當有效率。然而，點震源為極簡單之震源描述，在面對破裂過程複雜之大地震的同時可能面對系統無法完整描述震源破裂的問題。在本研究中提出了一個結合了多頻段RMT系統和多點震源地震矩逆推技術的研究流程，同時即時監測大地震，並快速分析其震源破裂之特性。多頻段RMT系統目前於離線測試當中，其中包含了四個子系統，每個子系統的頻段參數設定適用於不同的規模範圍。利用此系統，可以即時得到不同規模之地震的點震源參數。點震源參數得到之後，更詳細之破裂過程可以經由多點震源地震矩張量逆推技術來分析。此技術可偵測震源破裂過程中貢獻震源滑移較大的子事件，藉此快速掌握初步震源破裂之特性。為了檢測系統的可靠性，本研究方法除了經過理論波形測試之外，另於台灣和日本地區發生的顯著中大地震中選出六筆地震事件做為離線測試，並將得到的震源模型和有限斷層模型比較。結果顯示，利用本研究提出之方法得到的震源模型和有限斷層模型在空間和時間上是相吻合的。值得注意的是，利用了即時的資料和快速分析得到之結果，對於大地震後的災後應變(地震，海嘯災害)可以提供更精確的參考資訊。;It is shown that point-source parameters of events of small-to-moderate sizes (Mw<6.5) can be automatically determined by the Real-time Moment Tensor monitoring system (RMT) in Taiwan. The RMT system continuously inverts long-period (10-50s) wave field with a pre-calculated grid-based Green’s function database. This efficient approach tremendously reduces the time lag between event detection and focal mechanism determination. However, large earthquakes may distort the point-source assumption due to the frequency band. To derive stable point-source parameters of large earthquakes, the multiple-frequency bands RMT system was implemented. Each sub-system of the multiple-frequency bands RMT systems is specifically designed to estimate stable moment tensor solutions for earthquakes in a given magnitude range. Besides point-source parameters, dominant attributes of rupture processes can be analyzed using a multiple-source moment tensor inversion technique. The multiple-source analysis determines significant sub-events which account for large seismic moment during source ruptures (i.e. asperity). A procedure combining the multiple-frequency bands RMT system and the multiple-source moment tensor analysis is proposed. Some significant earthquakes in Taiwan and Japan are analyzed using the proposed approach as real-event exercises. The multiple-source model yields simple and robust determination of complex seismic source features. The characterized sub-events are in good agreement with asperities demonstrated in finite-fault models. By taking advantage of real-time determination of overall rupture processes, this approach can provide reliable assessments of associated hazards.
