- Ground-motion model is:
Y is in unknown units, a1 = 0.5258, a2 = -1.7383 (sic), a3 = -1.0153, a4 = 0.4079, a5 = -0.0006,
a6 = -0.179, τ = 0.350 (inter-event), ϕ = 0.0.567 (intra-event), σ = 0.666 (total), σSS = 0.586
(single-station) and bracketed σSS = 0.508 (see below).
- Use V s,30 to characterise sites. V s,30 estimates often from extrapolated profiles shallower than 30m or
proxies. 227 ≤ V s,30 ≤ 649m∕s with vast majority in range 240 ≤ V s,30 ≤ 480m∕s.
- Focal depths ≤ 30km.
- Derive model to examine single-station σ. Note that model is not expected to be used for engineering
- Data from National Strong-Motion Observation Network System from end of 2007 to end of 2015. 125
events are Wenchuan aftershocks, 39 are Lushan aftershocks and 22 are other events in Sichuan region.
- Exclude data with: rjb > 200km, Ms > 7.0, Ms < 4.0, zero focal depths and stations with unknown
V s,30. Only use a single record from the same site and same event. Only use 3-component records. Exclude
poor-quality records (e.g. multiple events, spikes, noise or very high/low amplitudes) after visual inspection.
Exclude events with < 4 records.
- Data from 103 different stations.
- Believe higher ϕSS found may be due to heterogeneous propagation medium in region (boundary of Sichuan
basin/eastern margin of Tibet plateau) and/or use of aftershock records.
- Reasonably uniform distribution of data w.r.t. Ms and rjb except all records with rjb < 10km are from
Ms < 5.5.
- rjb estimated from hypocentral locations and focal mechanisms or locations of seismogenic faults.
- Bandpass filter using acausal Butterworth filter with low cut-offs of 0.06 to 0.35Hz (depending on
magnitude) and uniform high cut-off of 30Hz. Visually inspect Fourier amplitude spectra and integrated
displacements to check selected cut-offs.
- Examine unit covariance between pairs of coefficients to check if necessary and resolvable.
- Compute inter- and intra-event residuals. Find no clear trends. Examine average site residuals w.r.t. V s,30
and find no trends.
- Consider the 47 stations (contributing 1463 records) that have recorded ≥ 10 records. 91 records from one
station. Compute δS2S and ϕss,s. Find considerable variation, which relate to various sources and paths.
Compute ϕSS for four distance ranges: < 50 (403 records), 50–100 (504), 100 - 150 (396) and > 150km
(160). Find higher values for shortest bin. Compute bracketed ϕss values using records obtained at stations
within a 10∘ bracket of event-to-station azimuths, thereby excluding sparse ray paths. Use data from 327
records at 21 of 47 stations to compute these bracketed ϕss values. Find considerable reduction in values
w.r.t. unbracketed estimates.
- Compute event-corrected single-path single-station standard deviations ϕss - sp using clusters of Lushan
aftershocks recorded by several stations. Find considerable reduction in ϕ over total and single-station