- Ground-motion model is:
where G is in g, a = -3.07, b = 0.83, c = -1.33, d = 0.15, e = 0.54, p = 0.0023, ν = 0.37 (inter-event),
ϵ = 0.55 (intra-event) and σ = 0.66 (total) (using 1-step regression) and a = -2.59, b = 0.87, c = -1.53,
d = 0.13, e = 0.53, p = 0.0029, ν = 0.39, ϵ = 0.55 and σ = 0.67 (using 2-step regression).
- Use 4 site classes (Lee et al., 2001):
- Very dense or stiff soil
- Stiff soil
- Soft soil
Evaluate site terms based on residuals w.r.t. model derived without site terms.
- Derive model to study components of variability and spatial correlation of ground-motion residuals.
- Use data from Taiwan Strong-Motion Instrumentation Program (>650 16bit or 24bit stations within 7
arrays) from 1993 to 2004.
- Focal depths < 30km.
- Select records with clear P- and S-wave onsets and signal-to-noise ratios (of Fourier amplitude spectra of
S-wave and pre-event noise) > 2.
- Seek simplest reasonable functional form that can describe general features of ground motions.
- Examine residuals w.r.t. rhypo.
- Believe that positive p coefficient is due to: peculiarities of data, region and/or azimuth-specific propagation
path effects or site-specific effects within Taipei basin and Ilan area.
- Check impact of grouping data from arrays (TAP, TCU, CHY and ILA) and within subgroups of site
class. Compute average group-dependent correction factors: D = a + bMw + cR for each array and site
class. Find significant (using F-test) trends. Use residuals from model including these correction factors to
compute variability components. Find corrections reduce σ.
- Find magnitude-dependency in total residuals but inter-event residuals that are independent of magnitude,
based on residuals binned in 0.5 magnitude-unit bins.