- Ground-motion model is:
where PSA is in g, α

_{1}= 7.0887, α_{2}= 0.2058, α_{3}= 0.0589, β_{1}= 2.9935, β_{2}= -0.2287, ξ = -0.854, γ = -0.0027 and ϕ = 0.08 (for M_{w}≤ 6.75) and α_{1}= 9.0138, α_{2}= -0.0794, α_{3}= 0.0589, β_{1}= 2.9935, β_{2}= -0.2287, ξ = -0.854, γ = -0.0027 and ϕ = 0.08 (for M_{w}> 6.75). σ = 1.18 + 0.035ln(T) - 0.06M, where for M_{w}< 5 use σ for M_{w}5, for M_{w}> 7.5 use σ for M_{w}7.5, for T < 0.05s use σ for T = 0.05s and for T > 3s use σ for T = 3s. - Uses V
_{s,30}to characterise sites. Notes that V_{s,30}is not being used to account for nonlinearity but only to better fit the observations. Presents coefficients for 450 ≤ V_{s,30}≤ 2000m∕s. Only 34 records from V_{s,30}> 1200m∕s hence for V_{s,30}> 1200m∕s recommends using V_{s,30}= 1200m∕s. - Uses 2 mechanisms:
- Strike-slip
- Mechanisms 0 (strike-slip) and 1 (normal, only 39 records of all data from M
_{w}≥ 4.5) of Flatfile. F = 0. - Reverse
- Mechanisms 2 (reverse), 3 and 4 (reverse-oblique) of Flatfile. F = 1.

- Model update of Idriss (2008) (see Section 2.295).
- Model derived within NGA West 2 project, using the project database (Ancheta et al., 2014). Data principally from: California (74 events) with some data from Idaho (1) and Nevada (1), Taiwan (5), China (55), Japan (5), New Zealand (2) and other countries (Canada, Mexico, Italy, Turkey and Iran) (15).
- Uses PSA(0.01s) to approximate PGA rather than true PGA.
- Selects records with M
_{w}≥ 4.5, all the required independent and dependent parameters and at free-field location. Because finds a significant change in decay of PGA (and spectral values) beyond r_{rup}= 150–175km, excludes more distant records. Next excludes events with < 3 records and multiple records from same earthquake at arrays (e.g. Taiwan SMART). - Examines data binned into: soft soil sites (100–211m∕s), nonlinear soil sites (211–450m∕s) and quasi-linear
sites (≥ 450m∕s). Assesses rough shear strain using PGV/V
_{s,30}for all records and plots against PGA. Finds that for quasi-linear sites the behaviour of sites in 1999 Chi-Chi mainshock is significantly different than remaining records and, hence, excludes these data. Estimates thresholds for nonlinear ranges based on modulus reduction curves for each bin. Finds that ≤ 4% of records in quasi-linear bin could be in mildly nonlinear range. Henceforth uses only data from this bin. - Examine total residuals w.r.t. M
_{w}, r_{rup}and V_{s,30}. Finds that model fits well in range 5.2–7.9 for PGA, 5.3–7.3 for T = 0.2s and 5.2–7.5 for T = 1s. - Does not include depth to top of rupture (Z
_{TOR}) in model because finds, based on residuals, that this parameter does not bias results for Z_{TOR}< 13km. - Does not include dip angle (δ) in model because finds, based on residuals, that this parameter does not
bias results for 24 < δ ≤ 90
^{∘}. - Does not include distance to surface projection of top of rupture R
_{x}or r_{jb}because not independent distance measures. - Examines residual plots w.r.t. depth to 1.0km∕s and 2.5km∕s shear-wave horizons. Finds little dependence and hence does not include these parameters.
- Insufficient data (95 records) on hanging wall to examine the effect of this parameter.