- Ground-motion model is [based on Abrahamson et al. (2013, 2014)] (for median):
_{Lin}= 660, c_{4}= 8.6, a_{1}= 1.350, a_{2}= -1.087, a_{3}= 0.275, a_{4}= 0.121, a_{5}= -0.592, a_{6}= 1.780, a_{8}= 0.0, a_{10}= -0.397, a_{11}= -0.200, a_{12}= -0.120, a_{13}= 0.670, a_{14}= -0.168, a_{15}= 1.100, a_{17}= -0.0062, a_{25}= 0.0015, a_{26}= -0.0007, a_{27}= 0.0031, a_{28}= 0.0035, a_{29}= -0.0010, a_{31}= 0.252 and a_{35}= 0.380. - Ground-motion model is (for aleatory variability):
_{1}= 0.734 and s_{2,all}= 0.520, s_{3}= 0.440, s_{4}= 0.350, s_{2,NoJP }= 0.450 and s_{4,NoJP }= 0.322. - Use 3 faulting mechanisms:
- Strike-slip
- Other rake angles. 221 events. F
_{RV }= F_{N}= 0. - Reverse
- Rake angles between 30 and 150
^{∘}. 79 events. F_{RV }= 1, F_{N}= 0. - Normal
- Rake angles between -30 and -150
^{∘}. 26 events, mostly 4.6 ≤ M_{w}≤ 6. F_{N}= 1, F_{RV }= 0.

Use two earthquake types:

- Class 1
- Mainshocks. F
_{AS}= 0. - Class 2
- Aftershocks. Events with centroid r
_{jb}< 15km (CR_{jb}). F_{AS}= 1.

Use two locations w.r.t. vertical projection of the top of rupture:

- Hanging wall
- F
_{HW }= 1. - Foot wall
- F
_{HW }= 0.

Use five regional terms to adjust model w.r.t. base model (all other regions, dominated by California):

- Taiwan
- F
_{TW }= 1 - China
- F
_{CN}= 1 - Japan
- F
_{JP }= 1 - Middle East
- F
_{ME}= 1 - Italy
- F
_{IT }= 1

- Vertical-component NGA-West 2 model corresponding to horizontal model of Abrahamson et al. (2013, 2014) (see Section 2.362 for details of data and approach used to develop model). Use similar database and functional form but aspects are different.
- Functional form does not include terms for nonlinear site amplification or soil depth used by Abrahamson
et al. (2013, 2014). Simulations of nonlinear site amplification not consistent with data. Notes that
reduced confidence in using the model away from an average site with V
_{s,30}= 450m∕s because of lack of nonlinear terms. - Use same database as Abrahamson et al. (2013, 2014) but remove records without vertical component (55 records) and those for which vertical component is questionable (98 records).
- Regress in a number of steps following approach of Abrahamson et al. (2013, 2014).
- Smooth coefficients to assure smooth spectra and to make model extrapolate in reasonable manner.
- Find ϕ from Japanese data much higher than from Californian or Taiwanese data, which relate to shallower
soils in Japan. Recommend s
_{2,NoJP }and s_{4,NoJP }for use outside Japan. - Note that V
_{s,30}scaling for Japanese sites is distance dependent but V_{s,30}slope in model is distance independent. Use V_{s,30}slope from r_{rup}≤ 50km for model because most relevant for applications. Note that at larger distance may be a misfit between predictions of regional model and observations from Japan. - Examine inter-event residuals w.r.t. M
_{w}and separated by region. Find no trends. - Examine intra-event residuals w.r.t. V
_{s,30}, Sa_{1180}(predicted spectral acceleration for V_{s,30}= 1180m∕s) and Z_{1}(depth to V_{s,30}= 1km∕s horizon). Find no trends.