| 摘要: | 在本論文研究中，我應用了多樣的分析方法，來探討並精進地震災害的評估。研究地區是台灣西南部，也是一個地震活動相對活躍的區域。我利用庫倫應力轉移作為工具，應用在兩個主要地震災害評估的層面：1）同震中「次要」的地表災害，2）地震分佈及大地震的預報。
第一部分，我們探討2016年規模6.4美濃地震的同震地表變形。地震在泥岩區造成了地表上拱。基於庫倫應力理論，我們計算了地震斷層滑移引發的區域三維應變改變的張量。結果顯示同震地表上拱區域與應力/應變改變有密切關聯：1）深處地殼（5-14公里）為擠壓性的應變，2）近地表（0-3公里）為張開型應變。這樣由地震產生的應力/應變改變型態，我們解釋造成了泥岩地區泥火山/泥灌入體深部儲氣儲水層受壓，使液體快速上升造成地表上拱。基於庫倫應力轉換的分析，我們將台灣西南部陸上三條垂直線型排列的泥灌入體作為模型中的受力斷層面（Recieving fault)，結果也是顯示了在地表上拱地區的泥灌入體有深部擠壓（5-6公里）、淺部拉張（0-4公里）。另外，庫倫應力轉換的計算，也顯示了同震應力改變，對於區域的基底滑脫面不適合發生逆衝滑移，反而是適合正斷層型態的低角度滑移，估計這個滑移型態不利於泥灌入體的地表上拱。
在第二部分的研究，我們探索是否可以預報地震分佈及大地震（規模六或以上）的發生。以台灣西南部而言，近20年來，從原本平靜狀態，進入到一個地震活躍階段，2010、2012、2016年接續發生了三個規模六的地震，作為我們探討的研究對象。基於庫倫應力改變的理論，加上地震速度狀態的動摩擦理論，我們假設規模大於四的地震，作為影響地殼應力改變的最主要來源，佔了絕大部分的比例。在這樣的假設下，我們計算過去15年來（從2005到2021），逐年的地震分佈的預報，包括地震頻率及空間分佈。在預報計算上，我們是採用Helmstetter等(2007)推導的算式。逐年預報分析與實際地震資料的比對顯示：1）在我們的假設下，地震的時空分佈的在逐年預報估算上，有相當不錯的吻合度，2）在3個規模六的地震震央區，在地震發生前幾年，預報的估算顯示有地震頻率逐年增加的現象，3）然而，在大地震發生前，會有數年（2010甲仙地震前一年，2016美濃地震前3年），地震分佈的預報有高估的情形，而高估的位置就在事後大地震的震央附近，我們傾向解釋為「應力虧損」（stress deficit）。這種現象，似乎可以作為大地震預報的指標。;In this thesis, I concentrate on a variety of analyses, intending to improve how to better assess the seismicity hazards, in a seismically active area. Here I take southwestern Taiwan as a case study area to study how the Coulomb stress transfer can be used as a useful tool for seismic hazard assessments, with two main aspects: 1) secondary effects of co-seismic surface damages and 2) forecast for seismicity and large earthquake.
Firstly, we investigated the 2016 Meinong earthquake (Mw 6.4) in southwestern Taiwan, which caused a surface pop-up in an area of 10x15 km2 with a maximum uplift of 12 cm, where lies an array of mud volcanoes and possible underlying mud diapir. We calculated the 3D strain tensor in a 3D mesh with 5x5x2 km grids in the epicentral area induced by the Coulomb stress change due to coseismic fault slip. We obtained substantial contraction strain (10-5-10-6) that occurred in a lobe showing “squeezing” at the depth of 5-14 km below the surface pop-up area. Dilatation strain (10-5-10-6) occurred at a shallow level (0-3 km) with a radial pattern around the surface pop-up area. Combining with local geology, which is composed of Mio-Pliocene ~5-km-thick mudstone in a fold-thrust belt, we interpret that the 2016 Meinong coseismic surface pop-up was closely related to mud diapirs/volcanoes, which were likely reactivated by a sudden increase of fluid pore-pressure in the basal reservoir (at 5-6 km depth) and dilatation in the shallow level. We also explored the potential effects of the Coulomb stress transfer on nearby receiver faults – including three arrays of mud diapir, the regional decollement, a suspected backthrust, and one thrust close to the pop-up area. Our results show that the Coulomb stress transfers a) favor NNE-trending mud diapirs in the coseismic pop-up area, with a combination of clamping stress changes at 5-6 km depth and unclamping stress changes at 0-4 km depth, and b) it does not favor triggered thrust slip on the regional thrusts.
Secondly, we focus on the forecast for seismicity and large earthquake in time and space in SW Taiwan during the past decades. In order to do so, we investigated the evolution of seismic rate and pointed out some anomalies with increases or decreases in rates (i. e., precursor indicators). These observations help to forecast the large earthquakes of M >6, by monitoring the seismicity rate during the seismic crisis of 3 large consecutive earthquake sequences over a time span of 6 years from 2010-2016. Furthermore, we adopted the theory of Coulomb stress changes and ate-and-state-dependent friction and applied them to all the earthquakes of M >4, to compute triggering seismicity from 2006 to 2010 and make a series of 1-yr forecast for seismicity rates from 2011 to 2020, which are following the large earthquakes M > 6. Combining all these results with the evolution of seismic rate in time and space, we observed an interesting pattern of changes of seismic rate with the 3 large earthquakes actually occurred in the study area during this time period, i.e., the 2010 Jiashian, 2012 Wutai, and the 2016 Meinong earthquakes.
In more detail, one critical key for determining the seismic rate anomaly is to determine background secular seismic rate in the target area. For the perturbation of aftershocks by the 3 large earthquakes in the study area, we use the “completeness magnitude”, Mc, as a threshold to determine whether and when the aftershocks had returned to the background seismicity level. First of all, we estimated the aftershock durations (ta) for a series of 3 earthquakes (M > 6) since 2010 in SW Taiwan. By carefully defining aftershock affecting area, specifying background levels, considering all reliable triggering seismicity within the area at the early period following the mainshocks, and adopting Mc = 1.6 within 10 days step following each mainshock until we obtained the most accurate aftershock rate model as a consequence, we estimated good values for aftershock periods (ta) of the three earthquakes.
We computed “one-year forecast seismicity rates” from 2006 to 2010 to obtain the effect of previous earthquakes on it. Note that we apply earthquakes of M > 4 year by year before the forecast period for the Coulomb stress change calculation. The results show a common pattern of seismic change: a trend of increase in forecast seismicity effect from earthquakes within a few years followed by a decrease of forecast seismicity effect from earthquakes within one year before a big earthquake (i.e., seismicity anomaly). This pattern can apply to the 2010 Mw6.3 Jiashian and the 2016 Mw 6.4 Meinong earthquakes. It also infers that there exists a period, likely a few years, for the accumulation of stress to induce the big earthquake in SW Taiwan. As for bigger earthquake like the 2016 Meinong earthquake, we obtain a higher increase in forecast seismicity from the cumulative difference in forecast seismicity by 4 years of earthquakes before the mainshock.
Moreover, we calculated another series of “1-yr forecast maps for seismicity rate” from 2011 to 2020 with the effect of the notable earthquakes in SW Taiwan M > 6 since 2010. The serial 1-yr forecast seismicity rate results are in good agreement with the evolution of actual seismicity from the CWB catalog both in time and space, which gives us good confidence in this method. Furthermore, with the help of a series of forecast results before the Meinong earthquake which was affected mainly by the previous large earthquake - the 2012 Mw 6.0 Wutai earthquake, we observe an area with a significantly high rate of > 3 events/month (i.e., seismicity rate anomaly) and with a significant increasing forecast seismicity rate with a value of >1.5 events/month compared with the background seismicity rate. This forecast seismicity “hot spot” area actually corresponds rather well with the epicentral area of the Meinong earthquake, showing a plausible sign of a coming large earthquake. Furthermore, we also obtained the abnormally increased forecast seismicity area (>3 events) close to the location of coming large earthquakes as in the cases of the Meinong and the Wutai earthquakes from the cumulative difference between one-year forecast seismicity of 4 years and 2 years before the large earthquakes, respectively. So, we found seismic precursor indicators for the approximate time occurring of large earthquakes as well as the potential occurrence of large earthquakes in an area, and then the aftershock diffusion zone of the coming large earthquake as well as the proximate location of the large earthquake.
In summary, the coulomb stress transfer is an important factor affecting seismic hazards in a region both in the short-term as co-seismic surface damages and in the long-term as triggering seismicity, especially for large coming earthquakes M > 6. |
