報告書番号

U99003

タイトル（和文）

震源の不均質過程を考慮した高周波地震動評価法の提案 --1923年関東地震の強震動評価--

タイトル（英文）

Estimation of High Frequency Seismic Motion Considering Heterogeneous Source Process

概要 （図表や脚注は「報告書全文」に掲載しております）

統計的グリーン関数法を拡張することにより、震源の不均質すべり分布を考慮できる高周波地震動評価法を提案し、1923年関東地震の震源モデルに適用した。提案した二種類の手法は、いずれも地震モーメントの保存を制約条件として、高周波地震動を再現できるすべり速度関数を設定した。2つのアスペリティをもつ1923年関東地震の震源モデルに適用した結果、以下の点が明らかになった。１．提案手法は、均質震源モデルを用いた従来法に比べて短周期地震動をより強く励起する。２．提案手法のうちの一方は、理論的にはバリアモデルと呼ばれる震源モデルと等価であり、与えられた震源の不均質すべり分布に対して加速度振幅の上限値を与える。従って、同手法は重要構造物に対する安全側の入力地震動評価法として位置付けられる。これにより、構造物の耐震性能照査レベルに応じて手法を選択することにより、入力レベルを多段階に設定することが可能となる。

概要 （英文）

The ground motion in a short-period range of the 1923 Kanto earthquake was estimated by the stochastic Green's function method, which is based on the empirical Green's function method but uses the artificial seismogram as the element event. A heterogeneous slip distribution on the main shock fault is also considered. The artificial ground motion substituted for the observed is calculated from the theoretical source spectrum and the empirical path and site effect. The site response beneath the evaluation point was obtained from the spectral ratios between the recent observed motions at the evaluation point and the synthesized ones that have the same magnitude and hypocentral distance as the observed, respectively. The observed events used in the analysis are of source depth less than 100km. To estimate the stress parameters of these events, the reference event was chosen from relatively large events and its stress parameter was determined from the records of other bedrock stations. Then, the stress parameters of the other events were obtained by taking a spectral ratio to the reference event. By correction of the stress parameters, the standard deviation of the estimated site response was reduced effectively in the frequency range higher than 1Hz.We assumed the slip distribution of the target event to have two independent large slip regions (asperities), following Wald and Somerville (1995). To describe this heterogeneous source process for the semi-empirical method, two ways of the conversion from spatial slip variations on the fault to stress parameter distributions (or slip velocity function) were proposed. In the first model the stress parameter on each subfault is assumed to be proportional to the slip. This means the rise time has a fixed value over the whole fault plane. In the second model the seismic waves radiated from asperities and whole fault were calculated separately, on assumption that both of which obey the homogeneous source model, and finally they were superposed in time domain. This model takes account of the stopping phase on the periphery of the asperity as well as the entire fault, therefore the high frequency waves are radiated more strongly within the asperities. We call the first method "the proportion method", and second "the superposition method" in this paper. As a result the superposition method generates remarkably larger short-period ground motions than the conventional method, which uses a homogeneous source model, whereas the proportion method shows relatively similar amplitude to the conventional method in the range of the standard deviation estimated from the randomness of phase property.

報告者

キーワード