Hi Ravi

I would recommend to start checking the frequency band over which you perform the SMSIM simulation. This kind of short period trough could be due to high-pass filtering - or absence of HF computation - at frequencies around 5-20 Hz. Have a look at the Fourier spectra of your SMSIM acceleration time series and compare it with the expected typical shape (log-log scale).

Do you have similar troughs whatever the site conditions ? What is the "kappa" value you consider ?

Hoping it will help, all the best

Thanks Prof. Pierre-Yves Bard for the explanation.

I haven't performed the SMSIM simulation. My observation was based on some GMPE (generated using synthetic ground motions using SMSIM simulation). Hence, I do not know the bandwidth of filtering frequency and the 'kappa' used. I understood the the filtering of high frequency may have led to this trough.

Additionally, I want to highlight that the site amplification factor (SAF) from bedrock to different site category (for the GMPE being referred) is of the pattern attached below. This is also based on SMSIM simulation. I believe that the trough in SAF (for site class D) is also due to filtering. I would also like to know, "whether this trough is realistic ?"

The legend in the figure refer to following site class;

A: Vs30 >1500m/s

B: 760-1500m/s

C: 360-760m/s

D:180-360m/s

Hi Ravi;

As dear Prof. Pierre-Yves Bard told that filtering band and the kappa values have crucial effect on results.

Also i think, in GMPE based SAFs such 'deamplifications' below short periods generally originated from earthquake data/dataset(far-field/near-field), critical damping factor and frequnecy resolution in Sa calculatons. So, actually this trough is most likely not 'realistic' in the manner of deamplification of seismic waves .

Thanks Hamdullah Livaoğlu for giving this supporting explanations for the above question. This is indeed helpful.