- Ground-motion model is:
where Y is in m∕s2, a = -4.976 ± 0.013, b = 1.281 ± 0.006, c = -1.660 ± 0.008, h = 1.485 ± 0.036, e = 0.161 ± 0.001, τ = 0.237 (inter-event), ϕ = 0.283 (intra-event) and σ = 0.370.

- Focal roughly uniformly distributed between 0 and 5km.
- Update of Sharma et al. (2013) (see Section 2.375) with much larger set of records, retrieved from the period 24/07/2007 to 18/11/2010.
- Data from 29 different stations.
- Data roughly uniformly distributed for all M
_{w}and R_{hypo}up to about 20km. Only a small proportion of data for larger distances. - Only use data with signal-to-noise ratio > 10 in frequency range 0.5–35Hz. Instrument correct and remove mean and trend from records. Bandpass filer with zero-phase shift and 4-pole Butterworth filter with cut-offs of 0.7 and 35Hz. Retain waveforms within the interval: origin time and time corresponding to 98% of the total energy (to remove coda). Taper signal with a 0.1% cosine-taper function. Then differentiate the records and filter again to remove noise introduced due to differentiation.
- Do not include M
^{2}term because of theoretical and observational reasons. - Site terms (s
_{j}) derived from residuals using same approach as Sharma et al. (2013). Find that addition of these terms improve fit to data. Do not report s_{j}here due to lack of space. - Find no significant trends in residuals w.r.t. M
_{w}and r_{hypo}. - Find positive trend in residuals w.r.t. focal depth.