 |
| Titel |
Analysis of Single-Path Sigma Using SMART-1 Array Data in Taiwan |
| VerfasserIn |
C. H. Sung, C.-T. Lee |
| Konferenz |
EGU General Assembly 2012
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 14 (2012) |
| Datensatznummer |
250064685
|
|
|
|
|
|
| Zusammenfassung |
| The standard deviation of the logarithmic residuals in ground-motion prediction may directly
influence the result of probabilistic seismic hazard analysis, especially in lower probability.
“What method can reduce the sigma more?” is a hot issue in recent years. Now, we
focus on “How to clarify the minimum variance of ground motion residuals from
median attenuation in nature”. This study differs from most previous studies in that it
examines the variability of ground motion for individual earthquakes using dense array
recordings. The basic assumption is that the dense array recordings can be treated as
repeated measurements for a given seismic source, ray path, and site condition. The
variation in peak ground motion (PGA) over short distances is used to directly measure
the statistical properties of PGA and in particular, to measure the variability in
PGA. We used the SMART 1 array recordings from four earthquakes (1986/07/30,
1986/11/16, 1986/05/20 and 1983/09/21) to analyze the single source to single
station variance. The SMART-1 array consists of a center station (C-00) and three
concentric rings (named inner I, middle M, and outer O) of radii 200, 1000, and 2000 m,
respectively, with 12 equally spaced stations on each ring. The similar topography and soil
conditions for the stations in the SMART 1 array reduce the effect of variations in
the local site conditions on the recorded PGA. For this reason, SMART 1 data are
less affected by variations in site conditions than other strong motion data sets.
Finally, we analyze the single path sigma by using variogram method. Measured
semi-variances at different distance bins (lags) and plot a semi-variogram. Nugget effect is
commonly found in each single earthquake. This means even the distance between
two stations is equal to zero; there still is significant difference in ground-motion
residual. This is potentially the most promising approach to find a natural sigma
for use in probabilistic seismic hazard ananlysis and is worth pursuing in future
studies. |
|
|
|
|
|
|