重建1951年花東縱谷地震序列震源破裂特性與縱谷斷層情境地震模擬之研究

Source rupture and ground motion simulations of 1951 Longitudinal Valley Earthquake Sequences and future earthquake scenario

林珈樺(Chia-Hua Lin)

指導教授 ： 胡植慶

共同指導教授 ： 李憲忠

國立臺灣大學/理學院/地質科學研究所/碩士(2016年)

https://doi.org/10.6342/NTU201601658

摘要

台灣東部花東縱谷被視為歐亞板塊與菲律賓海板塊的縫合帶，而花東縱谷斷層系統由北到南貫穿整個花東縱谷，屬於台灣地區活躍的斷層系統。在1951年10月到12月的期間，四個規模大於6.9的地震陸續發生在花蓮外海及花東縱谷斷層沿線，且記錄到數千個餘震，餘震震源隨時間由北往南傳遞，此地震事件被稱為「花東縱谷地震序列」，當時在縱谷沿線觀察到總共約90公里的同震地表破裂。本研究針對此的震序列做進一步的探討，以期了解縱谷斷層的滑移特性。本文主要分為兩大部份，第一部分藉由同震地表變形量以非負最小平方法(Non-Negative Least Square, NNLS)對三個不同斷層模型進行反演，求得斷層面上之滑移分布(slip pattern)，再透過譜元素法(Spectral Element Method, SEM)計算縱谷斷層分段破裂之地震波傳遞情形，並以模擬之地動分布圖(ShakeMap)和瞬時速度波場(snapshot)之結果，討論其波傳特性；第二部分則以台灣地震模型(Taiwan Earthquake Model)的縱谷斷層幾何作為基礎，考慮地震發生時，斷層滑移分布的不確定與隨機性，透過隨機滑移分布模型(stochastic slip model)建立出十個隨機滑移分布，並模擬縱谷斷層全段破裂在不同震源破裂情境下的地震波傳情形，最後統計不同破裂起始點的平均地動分布與地動極值分布。數值模擬的統計結果顯示，當縱谷斷層發生破裂，花東縱谷地區在各種震源破裂模型中，受到影響最為顯著，地表加速度值大多超過250 cm/s2。若破裂起始點在縱谷斷層中段，全台的地表加速度量值皆會大於80 cm/s2，而儘管距離縱谷斷層較遠的台北盆地、宜蘭平原、台灣南部地區，因受到破裂方向性與場址效應的影響，在特定的破裂分布情形下，會產生超過預期的地表加速度量值。藉由逆推而得的同震滑移分布與隨機滑移分布模型所建立之震源破裂模型的波傳模擬結果，我們可以了解當縱谷斷層發生破裂時，對台灣的影響範圍，並以統計的方式降低斷層面的滑移不確定性對地動分布的影響，期望更接近未來可能的地震情境，作為未來地震防災的參考依據。

關鍵字

1951年花東縱谷地震序列 ； 縱谷斷層 ； 震源逆推 ； 隨機滑移分布模型 ； 三維地震波傳模擬

並列摘要

The Longitudinal Valley (LV) in the eastern Taiwan is considered as the suture zone between the Eurasia Plate and the Philippine Sea Plate. Thousands of earthquakes are occur in this area every year. The Longitudinal Valley Fault (LVF) is a seismically active structure, which is located along the LV. During the time period from October to December in 1951, lots of large earthquakes occurred between Hualien and Taitung area, including four major earthquakes (M > 6.9) and thousands of aftershocks. This earthquake series is known as the Longitudinal Valley Earthquake sequence. Coseismic surface rupture with a total length of approximate 90 km were observed along LV. In order to understand the characteristics of source rupture and resultant strong ground motion, this study is comprised of two different parts. In first part, we reconstructed the source model and strong ground motion time history of this earthquake sequence. Inversion of the coseismic displacement data was first conducted. Based on the inverted slip distribution, we performed 3D forward simulation using the Spectral Element Method. Therefore, the second part of the thesis focuses on ground motion prediction for scenario earthquakes. We performed wave propagation simulation with ten stochastic rupture scenarios and examined the results collectively. The numerical simulation results showed that the PGA larger than 250 cm/s2 distributed along LV in eastern Taiwan in all cases. If the rupture started in the middle of LVF, PGA larger than 80 cm/s2 could be detected in the entire island. In the particular stochastic source rupture models, the PGA might be larger than expected in some places far from LVF due to source radiation and directivity effect, such as Taipei basin, Ilan and southern part of Taiwan. The models we presented in this thesis for both historical and scenario events can serve as reference for future in-depth seismotectonic studies and hazard assessment.

並列關鍵字

1951 Longitudinal Valley earthquake sequence ； Longitudinal Valley Fault ； source rupture model ； stochastic slip model ； 3D wave propagation simulation

參考文獻

內政部(1980)中華民國台灣地區三角點成果表。

姜介中(2009)利用驗潮紀錄估計臺灣沿岸地表垂直運動。國立台灣大學海洋研究所碩博士班碩士論文，共88頁。

Abe, K. (1981). Magnitudes of large shallow earthquakes from 1904 to 1980. Phys. Earth Planet. Inter., 27(1), 72-92.

Aki, K. (1967). Scaling law of seismic spectrum. J. Geophys. Res., 72(4), 1217-1231.

Andrews, D. J. (1980). A stochastic fault model, 1, Static case. J. Geophys. Res., 85, 3867–3877, 1980a.