- Ground-motion model is:
where Y is in cm∕s2, C

_{1}= 1.8207, C_{2}= 0.3506, C_{3}= -1.2775, C_{4}= -0.1370, R_{0}= 10 and σ = 0.3445 (for unweighted regression); C_{1}= 2.4911, C_{2}= 0.3647, C_{3}= -1.7654, C_{4}= -0.0575, R_{0}= 8 and σ = 0.3902 (for weighted regression); and C_{1}= 2.7831, C_{2}= 0.4956, C_{3}= -2.6029, C_{4}= 0.4220, R_{0}= 15 and σ = 0.3546 (for unweighted regression excluding Wenchuan aftershocks). - Use 2 site classes:
- Rock
- S = 0
- Soil
- S = 1

- Use data from National Strong-Motion Observation Network System of China.
- Select data using these criteria: r
_{epi}< 110km, M ≥ 4.5, both horizontal components available and PGA ≥ 0.01g on at least one horizontal component. Data from 6 mainshocks (5 ≤ M ≤ 6.4), their aftershocks and aftershocks of the 2008 Wenchuan (M_{s}8) mainshock. Most data (837 records) from aftershocks of Wenchuan earthquake. Most data from M < 4.8 and r_{epi}< 110km. - Use 3 regression approaches: 1) without weighting, 2) with weighting except for Wenchuan aftershocks and
3) without weighting and excluding Wenchuan aftershocks, to study influence of these data. Results vary.
Prefer weighted regression since predictions match observations from Ninger M
_{s}6.4 earthquake the best. - Compare observations (grouped by magnitude ranges) and predictions.
- Plot absolute (not using logarithms) residuals w.r.t. r
_{epi}.