Source Time Function - output amplitudes

[AashaPancha]

Hello, I have been testing the C6SmoothBump, Ricker, and Gaussian STF, keeping all other source parameters constant:
**source type=C6SmoothBump depth=3000 lat=-41.21 lon=175.01 strike=336 dip=63 rake=31 m0=9.55e14 **
The resultant output displacement values differ by approximately a factor of ten. I would have thought that the source time function would be normalised by the moment so that for a given earthquake magnitude, the resultant ground motion at a specified station location would be somewhat similar for each source type.

I am hoping to compare results to observed earthquake motions, but based on this testing, would like advice on the most appropriate source time function to use.

[andersp]

If you look in the user's guide, section 4, you will see that the Gaussian and C6SmoothBump time functions are similar, i.e., they both start and end at zero and have a non-zero integral. When you use either of these time functions, the results from SW4 model the particle velocities at the stations. However, the time functions are scaled differently and the freq parameter needs to be set differently, see section 4.4.
The Ricker function will give something completely different as it is proportional to the 2nd time-derivative of the Gaussian and the motions will correspond to the 2nd derivative of the velocity. We rarely use this function.
Hope this helps,
Anders