- Ground-motion model is (similar to Bindi et al. (2011a)): where Y is in cm∕s2, Mr = 5.0, a = 0.071, b1 = 0.603, b2 = -0.019, c11 = -1.895, c21 = 0.286,
c12 = -0.926, c22 = 0.035, c13 = -1.838, c23 = 0.511, c14 = -2.256, c24 = 0.455, h = 6.701, fNF = 0.035,
fTF = 0.181, sB = 0.050, sC = 0.203, δbas = -0.060, τ0.106 (inter-event), ϕ = 0.318 (intra-event) and
σ = 0.336 (total) when using rjb; and a = 0.053, b1 = 0.619, b2 = 0.023, c11 = -2.036, c21 = 0.150,
c12 = -0.949, c22 = -0.006, c13 = -2.129, c23 = 0.383, c14 = -2.257, c24 = 0.455, fNF = 0.011,
fTF = 0.192, sB = 0.059, sC = 0.212, δbas = -0.059, τ0.109 (inter-event), ϕ = 0.327 (intra-event) and
σ = 0.344 (total) when using rhypo (h is then the focal depth).
- Use 3 site Eurocode 8 (EC8) classes:
- V s,30 > 800m∕s. CA = 1 and other Cis are zero. About 300 records.
- 360 < V s,30 ≤ 800m∕s. CB = 1 and other Cis are zero. About 800 records.
- 180 < V s,30 ≤ 360m∕s. CC = 1 and other Cis are zero. Most near-source records. About 1600 records.
Data from 300 different stations. Most stations classified based on geological information.
- Use 2 classes based on location w.r.t. basins (either Po Plain or in smaller basins in Apennines):
- Middle of a basin (C1). Δbas = 1. About 950 records (all site class C stations). Most near-source
- Otherwise. Δbas = 0. About 1500 records.
- Use 3 faulting mechanism classes using classification of Zoback (1992):
- ENF = 1 and ETF = 0. 11.8% of records.
- ETF = 1 and ENF = 0. 47.8% of records.
- ENF = ETF = 0. All have Mw < 5.0. 35.9% of records.
Discard the 4.5% of strike-slip records as not representative of area.
- Use 4 locations w.r.t. Po Plain and eastern Alps (PEA) and distance (corresponds to index j of c1 and
- Sites within PEA and R ≤ Rh.
- Sites within PEA and R < Rh.
- Sites within northern Apennines (NA) and R ≤ Rh.
- Sites within NA and R < Rh.
Rh = 70km. Use relation LATref = -0.33LONs + 48.3, where LONs is station longitude and LATref
is reference latitude. Positive differences between station latitude and LATref imply PEA and negative
differences imply NA.
- Data from 44.0–46.3∘N and 8.0–13.5∘E. Most ( 60% of records) data from 20/5/2012 (Mw6.1) and
29/5/2012 (Mw6.0) Emilia earthquakes and aftershocks plus the Friuli 1976–1977 sequence. Data from
permanent and temporary stations of national Italy network (Rete Accelerometrica Nazionale), INGV
network plus 3% from other Italian, French and Swiss networks. Data extracted from ITACA 2.0 database.
- Select data with: Mw or ML > 4.0, rjb or repi < 200km and focal depth < 30km.
- Most data from MW < 5.5, rjb ≤ 120km and focal depths 5–10km.
- Station distribution is uniform in northern Apennines, eastern Alps and central Po Plain whereas few
stations in western Alps and western Po Plain.
- About 10% of records with rjb < 10km.
- Zero-pad acceleration time-series. Apply 2nd-order acausal time-domain Butterworth filter. Remove
zero-pads to make acceleration and displacement consistent. Typical band-pass frequency range is 0.1 to
40Hz for digital records and 0.3 to 25Hz for analogue records.
- Plot PGAs v rjb for 29/5/2012 Emilia earthquake and find that PGAs at stations to north of Po Plain
are 3–5 times larger than that to the south for rjb > 50km, which relate to reflections off the Moho (SmS
- Do not include magnitude saturation in functional form because no evidence from data.
- Compare predictions and observations for data around Mw6 and Mw4.5 and find good match.
- Plot residuals as histograms and w.r.t. rjb and find no trends.