Data from rock and soil sites but do not include site terms in model because of lack of data.
Data from Indian Himalaya taken from PESMOS, Virtual Data Centre (VDC) and Indian and GNSS
network (ISGN).
Data from VDC baseline corrected and bandpass filtered between 0.75–0.9 and 25–27Hz. Data from
PESMOS and ISGN bandpass filtered with cut-offs of 0.1 and 25Hz using 4th order Butterworth filter.
Reject data from Mw< 5 and > 200km because of low signal-to-noise ratios.
Use rhypo because insufficient information on rupture planes.
Use Monte Carlo method to estimate standard errors in coefficients.
Choose functional form through residual analysis of previously-published models.
55In the article there is an extra opening bracket in the equations for F but not a closing bracket so it could be that the terma4+ a5(M - Mh) could be all in brackets.