| 摘要: | 921集集大地震之後，經研究發現國內過去提出的機率式地震危害度分析(probabilistic seismic hazard analysis, PSHA)悉數低估了中部地區的危害度，也突顯出唯有考慮活斷層與選擇適當的強地動衰減式，才能獲致合理的PSHA結果。因此本研究重新檢討台灣地區的震源模式、震源參數與強地動衰減式，並以邏輯樹方法完成台灣地區的PSHA，進一步瞭解危害度結果的不確定性來源，同時也透過參數拆解分析瞭解貢獻危害度的主要震源特性。本研究的震源模式除了傳統的區域震源外，另外以三維面震源模型模擬活斷層與隱沒帶板塊震源。在地震規模機率分布的推估方面，採用台灣地區1900年至1999年地震矩規模(MW)的獨立主震目錄，建立各區域震源的截切指數模式；並以各活斷層的滑移速率(S)資料建構特徵地震模式。在強地動衰減式方面，首先將震源分為地殼內部與隱沒帶兩類，並區分不同類地盤的測站，考慮斷層上盤及下盤測站的強地動差異，採用測站距斷層面最短距離，並採用具近場規模飽和效應的Campbell衰減模式進行迴歸分析，獲得八組不同的強地動反應譜衰減式及資料之標準差。 本研究完成之PGA與0.2秒及1.0秒反應譜加速度的危害度等值圖，顯示斷層的活動度與危害度關係密切，尤其在斷層上盤幾何中心處危害度較高。中央山脈是危害度的低區之一，明顯分隔其東西兩側的花東縱谷與西部麓山帶的危害度高區，台灣島內危害度最低的區域在桃園的西北沿海及高雄外海一帶。另外，將本研究的PSHA結果除以僅考慮區域震源的PSHA結果，發現比值的高區分佈於斷層上盤，尤其在背景地震發生率特別低的中部地區以及新竹-苗栗一帶。 由於各場址的鄰近震源參數組合不同，所以影響各場址PSHA不確定性的主要因素不相同。因此特別針對邏輯樹分支有變化的參數，探討參數變化對PSHA不確定性的影響程度，結果發現台灣地區的S影響不確定性最高，其次是mu；斷層震源的特徵地震模式b值不影響不確定性，但是對於遠離斷層震源且使用截切指數模式的區域震源而言，b值影響相當明顯；斷層傾角影響PSHA不確定性，主要是位於斷層傾斜面末端的地表投影處。 台灣都會地區的PSHA結果經震源拆解後，發現僅採用區域震源不能正確反應出長再現週期的危害度，以475年再現週期為例，唯有考慮斷層震源才能使近場的PGA危害度超過0.4g以上。北部區域經震源拆解後發現，除鄰近活斷層的區域外，其餘區域受隱沒帶震源影響不容忽視，其影響範圍可以遍及桃園的西北外海區域，建議未來北部區域在地震防災上，需要加強注意隱沒帶震源的威脅。 本研究PSHA成果已合理反應出斷層震源的危害度，可進一步提供建築法規中耐震設計相關規定的參考。經過不確定分析後，提供未來地球科學界與地震工程界在降低PSHA不確定性的研究上一個明確的方向。各都會地區的參數拆解成果，可作為地震防災上進行地震境況模擬的候選震源參考。 Before the occurrence of the September 21, 1999, Chi-Chi Taiwan earthquake, the previous studies of probabilistic seismic hazard analysis (PSHA) by different agencies and authors totally underestimated the seismic hazard in Central Taiwan. The importance to incorporate fault sources and closest distance to fault in developing the ground-motion attenuation relationships and in PSHA. Using logic tree to model parameters at nodes and weightings at branches in PSHA could obtain the uncertainty of PSHA influenced by parameters. Further, we could understand the characteristic of sources in vicinities by deaggregating hazard contributed from different magnitude and distance bins. We adopt 3-D plate source to model fault planes and subduction zone plates except regional sources. Truncated-Exponential model developed by mainshock catalog in MW from 1900 to 1999 was used to describe the magnitude distribution of regional sources; Characteristic-Earthquake model developed by fault slip rate was used to describe the magnitude distribution of active fault and subduction interface sources. The ground-motion level of subduction plate sources are always higher than crustal sources. Hence, it is in need of adopting suitable attenuation relationship for each source in PSHA, especially the attenuation relationship of curustal source including the ground-motion data of Chi-Chi earthquake sequence, and which reveals the hanging-wall effect and site condition for specific site. According to the iso-seismic hazard map of PGA, 0.2sec and 1.0sec spectral acceleration, there are obviously correlation between the hazard level and the activity of fault, especially on the center of the hanging wall. the highest hazard level can be found in eastern longitudinal valley and western foothills to coast plain, separated by the central range in low hazard level. Furthermore, the hazard level considering faults activity divided by regional sources shows that the prominent ratio always distribute on hanging wall, especially on the low background seismicity region in the past, such as TaiChung and HsinChu-MiaLi region. The iso-seismic hazard map could be referenced by the Taiwan Building Code in the future. For specific site, the major parameters that influence uncertainty of PSHA are different because of various combinations of source parameters in vicinities. After uncertainty analysis by logic tree, we could understand the variation of parameter and how it influences the PSHA result. The most important one is the slip rate of fault, the second one is mu; the b-value of Characteristic- Earthquake model influence the uncertainty unapparent, but Truncated-Exponential model of regional sources which far away from active fault influence the uncertainty conspicuous. The variation of fault dip also influence uncertainty of PSHA obviously at the end of down-dip fault plane projection surface. The result of PSHA uncertainty analysis could help geo-scientists and earthquake engineers to estimate and judge the parameters more efficiently. From the deaggregation of PSHA in several metropolis of Taiwan, results show the hazard contributed from each distance and magnitude bin by different return period. The deaggregation could provide information for hazard mitigation while choosing scenario earthquakes. After considering fault activity in PSHA, the PGA level of near-field always exceed 0.4g in 475yr return period, however, it is impossible to reach the proper hazard level without fault source in PSHA especially in long return period. The northern Taiwan is threatened not only an active fault in vicinity, but also subduction plate sources more than what we have known before. Therefore we should pay more attention on subduction zone sources in hazard mitigation of northern Taiwan in the future. |
