08.04.2013 Views

Chiou and Youngs PEER-NGA Empirical Ground Motion Model for ...

Chiou and Youngs PEER-NGA Empirical Ground Motion Model for ...

Chiou and Youngs PEER-NGA Empirical Ground Motion Model for ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Appendix B<br />

Estimation of Distance <strong>and</strong> Geometry Measures <strong>for</strong> Earthquakes without Finite<br />

Rupture <strong>Model</strong>s<br />

Approach<br />

There are 110 earthquakes in the <strong>NGA</strong> data set that do not have finite fault models, <strong>and</strong><br />

there<strong>for</strong>e have only epicentral <strong>and</strong> hypocentral distances listed in the flat file. Rupture<br />

distances <strong>and</strong> source-site geometry parameters were estimated <strong>for</strong> the recordings from<br />

these earthquakes by simulating 101 possible rupture planes <strong>for</strong> each earthquake,<br />

computing the distance measures <strong>for</strong> each site <strong>for</strong> each simulation, <strong>and</strong> then taking the<br />

median of these values <strong>for</strong> use in completing the metadata in the <strong>PEER</strong>-<strong>NGA</strong> data base.<br />

The estimated parameters include RRUP, RJB, RSEIS, RRMS, the source=site angle θSITE, the<br />

hanging wall <strong>and</strong> foot wall indicators FHW <strong>and</strong> FFW, <strong>and</strong> the fault rupture width W <strong>and</strong><br />

depth to top of rupture, ZTOR.<br />

Simulation Process<br />

The first step was to simulate the fault rupture dimensions. The rupture area, A, was<br />

simulated using the Wells <strong>and</strong> Coppersmith (1994) relationship <strong>for</strong> all fault types:<br />

log( = log( Area)<br />

A) −3.<br />

49 + 0.<br />

91M<br />

σ = 0.<br />

24<br />

(B-1)<br />

Figure B-1 shows the data <strong>for</strong> rupture area versus M <strong>for</strong> the finite rupture models in the<br />

<strong>PEER</strong>-<strong>NGA</strong> database. The black lines show a linear fit to the data (solid = mean, dashed<br />

= 90% confidence on mean). The blue line is Wells <strong>and</strong> Coppersmith (1994), which falls<br />

within the 90% confidence interval <strong>and</strong> provides a closer fit to the data <strong>for</strong> smaller<br />

magnitude earthquakes of primary interest in the simulation, as those earthquakes without<br />

finite fault models are principally of magnitude < M 6.<br />

The aspect ratio of the rupture was then simulated using the relationship:<br />

log( = NM<br />

RV<br />

log( AR)<br />

3.<br />

097<br />

AR) ( 0.<br />

01752 − 0.<br />

00472F<br />

− 0.<br />

01099F<br />

) * ( M − 4)<br />

σ = 0.<br />

16 (B-2)<br />

where FNM <strong>and</strong> FRV are (0,1) dummy variables <strong>for</strong> normal <strong>and</strong> revere earthquakes,<br />

respectively. This relationship was defined by fitting the aspect ratio data <strong>for</strong> the <strong>PEER</strong>-<br />

<strong>NGA</strong> finite fault model data set (Figure B-2). Correlation in the residuals between<br />

rupture area <strong>and</strong> aspect ratio was low (

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!