where SaCU is in cm∕s2, α1= 5.6897, α2= 1.1178, α3= -0.50 and α4= -0.0060 (σ is not reported).
Only use data from Ciudad Universitaria (CU) station, which is the reference site in the hill zone (rock)
of Mexico City. Also derive models for Secretaria de Comunicaciones y Transportes (SCT) and Central de
Abastos (CD), which are in lakebed zone, using same approach.
Weight data so large and small earthquakes equally represented in regression.
Use Bayesian regression where prior probability distributions of coefficients are assigned based on ω2 model
for source, frequency-dependent attenuation parameters and duration from previous studies and random
vibration theory. Coefficients are updated using the records.
Compare observed and predicted spectra. Find match acceptable except for event 13 (Ometepec,
25/04/1989, Mw6.9), which was an anomalously intense earthquake.
Derive model for comparison with less direct ways of estimating motions.