以今日複雜的結構物環境，很難再以原始定義的震度或最大地表加速度來推估可能的震災。最近的研究結果顯示譜震度（spectral intensity）比最大地表加速度有較好的震災指標性。基於前人在譜震度的研究基礎，本研究的目的是要推求台灣地區及各分區的短、中、長週期控制區段的平均譜震度衰減律，期望其結果能應用於實際的地震危害度評估作業。近年來，相當多學者建議以震矩規模作為迴歸參數。不過以震矩規模作為迴歸參數，至少具有二個缺點，其一是：早期的地震資料沒有震矩規模（芮氏規模相對較為完整），若使用轉換公式推求則引入一個更不能掌握的誤差；另一方面，由震矩規模的定義看，震矩規模（∝ μ DA）對應的是斷層的平均永久位移，而震度表現的確是短週期（10 秒或更短）地震波的特性，因此，以震矩規模作為迴歸參數，在物理意義上不若以表達短週期特性的芮氏規模妥善。本計畫將以芮氏規模為迴歸參數，但在長週期控制區也將討論以震矩規模作為迴歸參數的狀況並比較之。我們認為最佳的距離參數應該是場址到最重要的能量輻射中心，本計畫亦將仔細探討距離參數的選擇方法。本計畫將使用基因演算法迴歸譜震度衰減律，並同時計入規模、距離、及強地動觀測誤差於迴歸分析演算。 Nowadays, the structural environment is becoming so complicated that the originally defined intensity scale and peak ground acceleration (PGA) are not adequate to do a good job for assessing earthquake damage. Recent studies showed that spectral intensity is a better earthquake damage index than PGA. The purpose of this project is to derive the average spectral intensity attenuation relations in short, medium, and long period ranges. Hopefully, the results can be used to asses the future earthquake hazard potential in Taiwan. In recent years, quite a few earthquake engineers suggested to use moment magnitude as a parameter in the regressing analysis of attenuation relations. However, this procedure at least has two drawbacks. First, there are a bunch of past earthquake data without moment magnitude record. It will introduce one more less controlled uncertainty if one uses some formulas to estimate moment magnitude from other scales. Second, from definition, moment is proportional to μ DA . It means that moment magnitude represents the character of average permanent slip of an earthquake fault. However, the intensity is associated with short period characteristics of earthquake waves. Hence, local magnitude may be a better parameter than moment magnitude in intensity attenuation relationships. We shall adopt local magnitude as the regression parameter in this project. In the meantime, we shall also use moment magnitude in the intensity attenuation formulas of long period controlled ranges as a comparison. In this project, we shall try to define the best distance measurement from the most important energy source of an earthquake to a strong motion recording site. This project will use genetic computing procedure and taking the errors of magnitude, distance, and strong motion observation into account in the regressing analysis.
