2002 Fiscal Year Final Research Report Summary
Studies on an updating procedure for efficiently estimating spatial distribution of earthquake ground motion indices fro early damage assessment
| Project/Area Number |
13650525
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
構造工学・地震工学
|
|---|
| Research Institution | Kagawa University |
|---|
Principal Investigator |
NODA shigeru Kagawa University, Faculty of Engineering Professor, 工学部, 教授 (80135532)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
KIYONO jyunji Kyoto University, Graduate School of Engineering, Associate Professor, 大学院, 助教授 (00161597)
|
|---|
| Project Period (FY) |
2001 – 2002
|
| Keywords | early earthquake damage assessment / ground motion indices / stochastic interpolation / real-time estimation / updating algorithm / Kriging / post-earthquake emergency management |
|---|
| Research Abstract |
Recently, several earthquake monitoring networks have been established in order to use earthquake information for early warning or damage assessment of urban systems. However, there is a limit to the number of seismographs that can be installed. In such a case, it is important to estimate earthquake motion by interpolation and extrapolation while taking into account the spatial distribution of seismic wave-forms and ground conditions and its uncertainty, as well as to evaluate the accuracy of the estimation.
in this paper, an updating procedure was efficiently estimating the spatial distribution of earthquake ground motion indices when a one or more observed data are sequentially available. The update equations of simple and universal kriging were derived. The characteristic values of earthquake motion follow a lognormal distribution. A simple kriging is an interpolation problem when stochastic characteristics (mean values and covariance) of earthquake motion are specified in terms of location, and a universal kriging is such a problem when the spatial change in mean values in unknown. Numerical results show that as the number of observed data increases, better estimates are recursively obtained, and at the same time the error variances reduce. It is found that this procedure offers a significant computational saving when compared to a naive kriging implementation.
|
