Range check error

[moizhyd]

Hello,

I am trying to load a spectrum from a .txt file and get a range error.
Could you please explain what I miss ? The minimum period in my
data is 0.05 and maximum is 4. This is the same as what is defined in
the settings.

Best Regards,
Moiz

[seismosoft]

Are you using SeismoMatch 2018?
We checked the module and it worked fine for us.
Can you describe the exact steps with which you get this problem, since we could not duplicate it?

Regards,
Seismosoft Support

[UCM]

The problem i have is about the spectral matching in periods after 1 seconds. When i used the programa the matched spectrum does not fit with the target spectrum and it supposed that the match sould be done in the range of 0.06 to 4 seconds. What could be the reason of that error ?

I attached the link where you can see the image of the results of the analysis:
https://drive.google.com/file/d/1UeAITz ... sp=sharing

thanks for your time

[seismosoft]

Did you get convergence when running the matching? It seems that the analysis did not converge.
Seismosoft Support

[UCM]

Yes, it did converge, and the methodology that i used is Al Atik and Abrahamson. When i changed the method to Hancock the result converge and the original spectrum fit perfectly. Could it be that the acelerograms were previously corrected by base line and the Al Atik and Abrahamson method is causing the issues ?


Image of spectral matching using Al Atik and Abrahamson methodology
https://drive.google.com/file/d/1ORRAw6 ... sp=sharing


Image of spectral matching using Hancock methodology
https://drive.google.com/file/d/1Y6_X46 ... sp=sharing

Greetings, thanks for your help

[seismosoft]

This is strange...
Can you please send the project file to support@seismosoft.com, so that we have a look?
Seismosoft Support

[seismosoft]

We ran the project and, as we correctly predicted, you did not get convegence with the Al Atik and Abrahamson method. This is why you get such large differences.
Note however that the PGA of the target spectrum is almost 0.20g, whereas the original records are only have very small pga, around 0.007'0.008g, that is 25 time smaller. Furthermore, the spectra of the original records are very different from the target spectrum. In such cases, it is difficult for the algorithm to converge and the new matched records are not always realistic.
Seismosoft Support

[huffte]

I would like to add that this is not unusual. One of the developers of spectral matching techniques has stated that they discard approximately 80% of the records they generate through spectral matching. As noted by the Seismosoft team, it is important to have a reasonably close match to shape and amplitude. Match to shape may be measured through computation of DRMS or MSE. Scaling records prior to matching is also beneficial. So there is a good deal of effort which should go into selecting which records should be candidates for spectral matching - not just any seed record will do.

In fact, even when convergence is achieved, the character of the ground motion may be altered dramatically. For example, pulse-type regions of the signal may be lost. Or, unrealistic integrated displacement histories may result.

For DRMS definition, see: Katsanos, etal., (2009), "Selection of earthquake ground motion records: A state-of-the-art review from a structural engineering perspective"

For MSE definition, see the PEER Ground Motion Database User's Manual:
https://ngawest2.berkeley.edu/peer_grou ... manual.pdf