Drift in position

[moizhyd]

Hello,

I am trying to scale an accelerogram with a custom spectrum.
After the scaling (which goes ok) I see that velocity and acceleration look reasonable
but the position shows a large drift with respect to time.

I am wondering why this happens and how I could fix it.
Please let me know if you have any insights regarding this.

Best Regards,
Moiz

[huffte]

Hi moizhyd. First of all, does the displacement drift occur in the unmatched record? If so, then you should likely baseline correct and filter the original record, using SeismoSignal for example, to eliminate the drift. If not, then you might wish to baseline correct and filter the matched record to eliminate the drift. Baseline correction and filtering, when done properly, typically do not alter the response spectrum significantly over a wide range of periods, but there is always the chance that they will.

Are you using SeismoMatch 2018? If so, I believe it already includes baseline correction and filtering options. I'm not certain that SeismoMatch 2016 does. So one approach might be:

1. Perform matching on the record in SeismoMatch
2. Save the matched accelerograms
3. Load the matched accelerograms in SeismoSignal
4. Perform baseline correction and filtering in SeismoSignal
5. Save the adjusted accelerograms
6. Load the adjusted accelerograms into SeismoMatch and perform matching on it.

All this being said, your best option might be to try other records for matching to your target spectrum. If indeed the original record does not have the drift but the matched record does, this would seem to indicate that the alteration through wavelet addition to achieve matching to the target was so dramatic that the original accelerogram was not the best choice for the matching process.

Best of luck moizhyd.

[chengzhen]

I am using SeismoMatch 2018 to match a record with two horizontal components(X,Y) to a target spectra. The AI Atik & Abrahamson (2010) algorithm is chosen. The results show that one component(X) match perfect to the target spectra with no shift appear, but the other component(Y) dose not. After matched, the Y component's acceleration dose not end with zero and the velocity and displacement have a large shift(the original record dose not have). I am wondering what's the problem?
Thank you!

[seismosoft]

Indeed,as huffte correctly pointed out, you will need filtering and/or baseline correction of the record. Note that these facilities are included in the commercial release of SeismoSignal.
Seismosoft Support