- Ground-motion model is:
where PSA is in g, a = 0.3872, b = -0.0009, c1 = 1.0240, c2 = 1.0870, c3 = 1.1390 and σ = 0.2847.
- Use 3 site classes following approach of Zhao et al. (2006) (see Section 2.245) that uses horizontal-to-vertical
response spectral ratios:
- SC-I. 844 in larger dataset.
- SC-II. 390 in larger dataset.
- SC-III and SC-IV. 414 in larger dataset.
Originally had very soft soil category but insufficient data. Use this approach because V s,30 lacking and
because it accounts for predominant period of the site. Find clear differences in average H/V ratios among
the different classes.
- Focal depths 0–29km with most around 10km, which agrees with seismotectonics of Algeria.
- Combine data from Algeria, Europe (Ambraseys et al., 2000) and western USA (USGS and CDMG)
because insufficient data from larger events to develop model solely using Algerian data. Based on a residual
analysis, find using only Algerian data leads to over-prediction of ground motions from larger earthquakes.
- Algerian data from SMA-1, SSA-1 and Etna instrumentsv from 1980 to 2014. Instrument and baseline
correct records. Then bandpass Ormsby filter records using cut-off of 25Hz (roll-off width of 3Hz), and
0.12–0.2Hz (with roll-off width of 0.06Hz) for digital and 0.2–0.3Hz (with roll-off width of 0.1Hz) for
- Remove data from distances likely beyond distance at which instrument triggering is likely using: 1)
distances suggested Boore et al. (2014) and 2) distances estimated by predicted PGAs< 10cm∕s2 using
model of Fukushima et al. (2003).
- Find no trends in residuals, including w.r.t. region (Algeria, Europe or W. USA).