P波在地震預警上之研究

Abstract

地震預警乃利用地震發生後之初達P波所帶來的資訊，在破壞性的地震波尚未到達前推估其地震規模或強度及其發生位置，並對可能發生地震災害的地區提出警告。 本論文分為兩部份研究初達P波在地震預警上之測試，目的在於提供地震預警系統在實際應用上快速準確的方法。 第一部份研究比較及測試 (平均地動週期)與 (最大地動週期)兩方法在地震預警上規模預估的準確性，並著重於濾波程序對其之影響。 研究分析日本K-NET強震觀測網中16個地震矩規模6到8.3之近站地震紀錄。 計算過程根據一系列先前之研究進行一次Butterworth 0.075Hz高通濾波(Wu and Kananmori, 2005a, 2005b, 2008a, 2008b; Wu et al., 2006; 2007)進而改變濾波之pole值。 結果顯示pole值為2時有最佳的規模預估準確性。 另一方面， 的計算過程則依循(Wurman et al., 2007)之研究，結果顯示規模預估誤差明顯較大。 然而改變其濾波程序為如同 方法中所使用的Butterworth濾波後，顯示pole值為5時結果可以得到明顯改善。 進而平均此兩種方法所得最佳結果可以得到更為準確之規模預估。 第二部份研究則選取一組裝設於一棟三層樓建築物的強地動儀陣列系統所得之地震紀錄進行 以及Pd(初達P波三秒間之最大地動位移相之震幅)方法之測試。 此系統1996年至2006年間所記錄之地震中，僅有規模6.1，震央距14.5km之台東地震對此建築物造成破壞，提供本研究比較破壞性與非破壞性地震之差異。研究發現可能由於紀錄訊噪比不佳造成 方法在規模預估上並不理想。 然而此組資料中，僅有造成破壞的台東地震之Pd逾0.5 cm，此結果符合以往之研究 (Wu and Kanamori, 2005b,2008)，顯示此例中以Pd門檻值做為破壞性指標為較佳之地震預警方法；吾人亦發現以此方法進行地震預警，其地震儀裝設之位置對此結果並不會造成明顯差異。

Earthquake early warning (EEW) is to estimate the magnitude or intensity as well as the location of an event with the information of first several seconds of P wave, and to provide a warning for regions where damages are likely to occur. This thesis is separated to 2 parts to test EEW methods using P wave information. The purpose is to provide a quick and accurate method for practical EEW system. The first part of this study compares (average ground motion period) and (dominant ground motion period) methods, and emphasizes on the filtering process. K-net strong motion records in Japan were used, and 16 events with magnitude 6 to 8.3 were chosen to be analyzed. A 0.075 Hz high-pass Butterworth filter was applied for determination of based on our previous studies (Wu and Kananmori, 2005a, 2005b, 2008a, 2008b; Wu et al., 2006; 2007), and the results show that using two poles in the filters has the best magnitude estimates. On the other hand, was determined using Wurman et al. (2007) procedure, and the magnitude estimation was found to have large uncertainty. However, adding a 0.075 HZ high-pass Butterworth filter the same as procedure with five poles could obviously reduce the uncertainty. In addition, the best results of the two methods could be averaged to provide more accurate magnitude estimates. In the second part of this study, records from a strong motion array sensor system which installs at a building of Fire Fighting Bureau of Taitung county were selected to examine the and Pd methods. From 1996 to 2006, only Mw 6.1 Taitung earthquake with epicentral distance 14.5 km caused damage to this building. This fact provides this study to compare the difference between destructive and non-destructive events. According to the results, the method doesn’t provide good magnitude estimation in this case probably because of the low signal to noise ratio (SNR). On the other hand, only Pd of the destructive Taitung earthquake exceeds 0.5 cm which conforms to the results proposed by Wu and Kanamori (2005b, 2008). This result shows that in this case, it is better to warn a destructive event with a Pd threshold. In addition, it is also found that by this way to provide EEW, the location for the installment of seismometers doesn’t make obvious difference.